Article

Automatic liver segmentation for volume measurement in CT Images

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Computed tomography (CT) images have been widely used for diagnosis of liver disease and volume measurement for liver surgery or transplantation. Automatic liver segmentation and volume measurement based on the segmentation are the most essential parts in computer-aided diagnosis for liver CT as well as computer-aided surgery. However, liver segmentation, in general, has been performed by outlining the medical image manually or segmenting CT images semi-automatically because surface features of the liver and partial-volume effects make automatic discrimination from other adjacent organs or tissues very difficult. Accordingly, in this paper, we propose a new approach to automatic segmentation of the liver for volume measurement in sequential CT images. Our method analyzes the intensity distribution of several abdominal CT samples and exploits a priori knowledge, such as CT numbers and location of the liver to identify coherent regions that correspond to the liver. The proposed scheme utilizes recursively morphological filter with region-labeling and clustering to detect the search range and to generate the initial liver contour. In this search range, we deform liver contour using the labeling-based search algorithm following pattern features of the liver contour. Lastly, volume measurement is automatically performed on the segmented liver regions. The experimental measurement of area and volume is compared with those using manual tracing method as a gold standard by the radiological doctors, and demonstrates that this algorithm is effective for automatic segmentation and volume measurement method of the liver.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Depth convolutional neural networks (DCNNs) revolutionized detection and segmentation until recently [11]. An adversarial network was introduced using CNN [12,13] and the graphic cutting approach, with the supervision network integrated to provide the model a faster convergence speed and stronger recognition capability [14][15][16][17]. ...
... This is accomplished by comparing the predicted segmentation map to the actual segmentation map. The U-Net architecture can take advantage of both Fig. 4 Architecture of U-net [13] low-level and high-level features for precise segmentation by integrating the encoder and decoder paths with skip connections [34,35]. Due to its capacity to manage sparse training data and produce accurate segmentation maps, this architecture has been widely used for a variety of medical image segmentation applications, including liver segmentation [36,37]. ...
... Three challenging cases, including (a) Small liver area, (b) Discontinuous liver area, (c) Blurred liver boundary[13] ...
Article
Full-text available
Medical image analysis requires liver segmentation for liver disease detection and treatment. Deep learning approaches, particularly liver segmentation, have demonstrated astounding effectiveness in a variety of medical imaging applications. Using the U-Net architecture, a well-liked and successful deep learning model for semantic segmentation, a liver segmentation approach is suggested in this study. This approach uses 3D abdominal CT images with liver regions identified. The U-Net model collects local and global contextual data via skip links and an encoder-decoder network. Supervised learning and data augmentation are used to develop the network’s generalization ability. Intensity normalization, voxel resampling, and image cropping were used to enhance liver segmentation by improving input data quality and consistency. Post-processing approaches like linked component analysis and morphology improved segmentation results and eliminated false positives. A separate test dataset and conventional assessment criteria as DSC, sensitivity, and specificity were employed to evaluate our liver segmentation approach. A Dice score of 0.9287 indicates a 92.87% overlap between the sets. This is a good result since the segmentation or comparison approach identified and aligned the matching regions in the sets. Train dice loss, train metric dice, test dice loss, test metric dice and mean dice are found to be 0.0223, 0.9733, 0.289, 0.782, and 0.9287 respectively. Lab results reveal that the current liver segmentation approach is accurate and resilient. Comparing present strategy to other cutting-edge liver segmentation methods shows its competitiveness. In conclusion, this study proposes a liver segmentation method based on the U-Net architecture that successfully tackles the difficulties in precisely distinguishing the liver from abdominal CT scans. The suggested method has produced encouraging results, demonstrating its potential for clinical uses in the diagnosis of liver disease, surgical planning, and therapy monitoring.
... This is applicable in the research environment where time constraints are not as important and in the clinical setting. Manual segmentation is not only the most time consuming but it also requires the most operator input and can be sensitive to inter-operator variability [129,[131][132][133]. For this reason, manual segmentation of the whole liver is not suitable for routine clinical use [134]. ...
... Clinical applicable methods must use a fast, accurate, reproducible and robust segmentation technique. Hence, if segmentation of the whole liver proved superior for routine clinical liver fat determination, semiautomatic or fully automatic segmentation algorithms would need to be developed [131,132,[135][136][137][138]. ...
... Poor contrast of the liver tissues compared to surrounding structures makes automated segmentation difficult and less accurate. CT imaging has poor soft tissue contrast, hence the liver has similar intensity with surrounding tissues or organs [131,139]. The strength of MRI is the ability to modify tissue contrast with changes in sequence parameters. ...
Thesis
Liver fat has a greater negative effect on health than fat stored under the skin. Increased liver fat is an indicator of metabolic disorders associated with being overweight or diabetes. This thesis seeks to compare whole liver estimation of liver fat by magnetic resonance imaging (MRI) techniques with methods that sample small areas of the liver, namely the current clinical gold standard, liver biopsy, single voxel MRS in 2 positions and two small regions-of-interest for imaging techniques. Currently, the clinically accepted method to measure liver fat is through biopsy. A liver biopsy requires inserting a needle into the liver and removing a small sample. This procedure has significant risks to the patient. Use of liver biopsy to measure liver fat is not suitable for research because it cannot be repeated on the same patient within a short period of time. Validation of non-invasive techniques has enormous benefits for patient management and research. Magnetic resonance imaging and spectroscopy allows several different methods for determination of liver fat non-invasively. This study used information from a database of patients that had liver biopsy for clinical assessment along with assessment by magnetic resonance imaging and magnetic resonance spectroscopy. The first step was optimizing a liver segmentation method on a sub-set of 15 patients. The optimized whole liver manual segmentation protocol was developed to select liver tissue from the whole organ and exclude non-liver tissue including vessels, image artefacts and image calculation errors when generating the liver fat images. The optimised method was then applied to the whole data set of 63 patients. Direct comparison of region-of-interest versus whole liver methods for calculating liver fat from two different imaging techniques was included. Whole liver segmentation was performed using manual segmentation methods. These are very time consuming and operator dependent methods, particularly compared to ROI methods. However, manual segmentation remains the “gold standard”. ROI methods could be completed in less than one minute and easily incorporated into the imaging time with the patient in the MRI scanner. Investigations to evaluate correlations were done between the following: ROI and whole liver analysis of in-phase/out-of-phase and with/without fat saturation, magnetic resonance spectroscopy, biopsy results using the percentage of hepatocytes with visible fat. All methods tested for the determination of liver fat were highly correlated. There was no consistent improvement in the correlation of either imaging method, ROI or whole liver, compared to MRS or biopsy. However, the calculated value for steatosis was different for almost all methods. This resulted in different numerical ranges being required to match the steatosis grading determined from liver biopsy. This work demonstrates high correlation, yet numerically different values for liver steatosis, determined from the ROI or whole liver from IP/OP and ±FS MRI techniques, compared to MRS or liver biopsy. No benefit was evident for ROI versus whole liver calculation of liver, however quick visual inspection of images with fat dependent image intensity should be included. This work supports the growing body of evidence that MRI can provide safe, non-invasive and reliable methods for determining liver steatosis. However, standardisation of acquisition and analysis methods is required to enable definition of numerical ranges for clinical steatosis grading.
... Surrounding extrahepatic vessels, extrahepatic bile ducts, and the gallbladder were excluded. The liver volume obtained as a result of segmentation was adopted in the study as the gold standard against which other parameters and calculated volumes were compared [22][23][24]. For a better understanding, the workflow has been presented in Figure 1 and the data of the test group has been presented in Table 1. ...
... Surrounding hepatic vessels, extrahepatic bile ducts, and the gallbladder were excluded. The liv ume obtained as a result of segmentation was adopted in the study as the gold st against which other parameters and calculated volumes were compared [22][23][24] better understanding, the workflow has been presented in Figure 1 and the data of group has been presented in Table 1. The results were gathered and analyzed using Statistica 13.3 (TIBCO Software Inc.: Palo Alto, CA, USA (2017)). ...
Article
Full-text available
Background: A reliable assessment of liver volume, necessary before transplantation, remains a challenge. Our work aimed to assess the differences in the evaluation and measurements of the liver between independent observers and compare different formulas calculating its volume in relation to volumetric segmentation. Methods: Eight researchers measured standard liver dimensions based on 105 abdominal computed tomography (CT) scans. Based on the results obtained, the volume of the liver was calculated using twelve different methods. An independent observer performed a volumetric segmentation of the livers based on the same CT examinations. Results: Significant differences were found between the formulas and in relation to volumetric segmentation, with the closest results obtained for the Heinemann et al. method. The measurements of individual observers differed significantly from one another. The observers also rated different numbers of livers as enlarged. Conclusions: Due to significant differences, despite its time-consuming nature, the use of volumetric liver segmentation in the daily assessment of liver volume seems to be the most accurate method.
... A three-stage approach is used by Lim et al. (2006) [28]. The first stage involves image simplification as preprocessing, where an ROI is identified and thresholds are determined using multilevel thresholding. ...
... A three-stage approach is used by Lim et al. (2006) [28]. The first stage involves image simplification as preprocessing, where an ROI is identified and thresholds are determined using multilevel thresholding. ...
Article
Full-text available
Oncology has emerged as a crucial field of study in the domain of medicine. Computed tomography has gained widespread adoption as a radiological modality for the identification and characterisation of pathologies, particularly in oncology, enabling precise identification of affected organs and tissues. However, achieving accurate liver segmentation in computed tomography scans remains a challenge due to the presence of artefacts and the varying densities of soft tissues and adjacent organs. This paper compares artificial intelligence algorithms and traditional medical image processing techniques to assist radiologists in liver segmentation in computed tomography scans and evaluates their accuracy and efficiency. Despite notable progress in the field, the limited availability of public datasets remains a significant barrier to broad participation in research studies and replication of methodologies. Future directions should focus on increasing the accessibility of public datasets, establishing standardised evaluation metrics, and advancing the development of three-dimensional segmentation techniques. In addition, maintaining a collaborative relationship between technological advances and medical expertise is essential to ensure that these innovations not only achieve technical accuracy, but also remain aligned with clinical needs and realities. This synergy ensures their applicability and effectiveness in real-world healthcare environments.
... A three-stage approach is used by Lim et al. (2006) [28]. The first stage involves image simplification as preprocessing, where a ROI is identified and thresholds are determined using multilevel thresholding. ...
... A three-stage approach is used by Lim et al. (2006) [28]. The first stage involves image simplification as preprocessing, where a ROI is identified and thresholds are determined using multilevel thresholding. ...
Preprint
Full-text available
Oncology has emerged as a crucial field of study and treatment in the domain of medicine. Computed tomography has gained widespread adoption as a radiological modality for the identification and characterisation of pathologies, particularly in oncology, enabling precise identification of affected organs and tissues. However, achieving accurate liver segmentation in computed tomography scans remains a challenge due to the presence of artefacts and the varying densities of soft tissues and adjacent organs. This paper compares artificial intelligence algorithms and traditional medical image processing techniques to assist radiologists in liver segmentation on computed tomography scans, and evaluates their accuracy and efficiency. It is noteworthy that although studies have been conducted on liver segmentation in computed tomography scans, they often lack an intuitive and visual component that allows healthcare professionals to manipulate and observe the results obtained, thereby limiting interaction with the outcomes. From the literature review, challenges such as under-segmentation, over-segmentation, and poor boundary detection, as well as the selection of methods to improve the accuracy and efficiency of liver segmentation in computed tomography scanners, are highlighted as needs to be addressed. The importance of future research in understanding the essential features for the study, generating more datasets, improving segmentation efficiency, and developing lightweight artificial intelligence frameworks for liver segmentation is outlined.
... Lim et al. [205,206,207], combined prior-knowledge such as the location of the liver and its intensity distribution with active contours to segment the liver. The rough estimation is determined via thresholding and morphological operations in a multiscale fashion. ...
... The final segmentation is obtained semi-automatically by asking the user to select the slices where the segmentation gave accurate results, and propagate this segmentation iteratively to adjacent slices. An example is depicted in the figure 3.Intensity based liver segmentation workflow used by ©Lim et al.[207] ...
Thesis
To evaluate the status of a liver tumor, we usually perform a biopsy followed by an anatomo-pathological evaluation of the extracted sample. However, the biopsy, due to the small sampling size, does not testify the intra and inter-tumor heterogeneity, thus struggling in assessing precisely the phenotypical characteristics of the patients. Recent progress in medical imaging and data science fields enabled the emergence of a new technique called radiomics, that is partially answering these challenges. In this thesis, we have been focusing on hepatocellular carcinoma, and we built new imaging methods to characterize this widespread pathology. By incorporating temporal information through multiphase images, specialized UNet-like networks have been stacked in a cascaded architecture to provide a semantic segmentation of both the liver and its internal tissue (parenchyma, active & necrotic part of the tumor). To characterize the strong heterogeneity that resides in the tumor, we predict the histological grade on a fine scale (slice-wise), by re-using the features learned from the semantic segmentation network. Our preliminary results enable the production of a fine-detailed map of the tumor that separates well differentiated areas from poorly ones. Even though these results need to be confirmed with a larger cohort, we believe that medical images combined with deep modeling techniques may soon be introduced in a clinical workflow to help diagnose and evaluate the phenotypical characteristics of pathologies such as liver cancer.
... Goodman et al. (2010) suggested a statistical form model (SSM) to automatically segment the liver from abdominal CT images. Lim et al. (2006) developed an intense learning machine (ELM) to decide whether the segmented organ is a tissue in the liver or non-liver. Song et al. (2013) implemented an adaptive fast marching (FMM) method to automatically segment the liver from CT images. ...
... 1 trilateral filter 2 histogram compensation (HE) (Lim et al. 2006) 3 adaptive histogram compensation (CLAHE) (Yussof and Burkhardt, 2009b). ...
Article
Liver cancer remains the most common cause of cancer death worldwide; in recent decades, the epidemiology has improved. Commonly, endoscopic stomach biopsy is performed for early detection of liver cancer to minimise mortality. Picture segmentation is a key technique for comprehension and intensification of the medical image. The purpose of this study was to create a sustainable computer-aided estimating system to determine the risk of liver cancer development, achieved through image processing on a CT image. Initially, the image is enhanced by using anisotropic diffusion filtering with unsharp masking (ADF-USM) technique, and the computer-aided estimating method was developed based on fuzzy C-means clustering, Otsu's, region-dependent active contour and superpixel segmentation dependent iterative clustering (SSBIC). This sustainable approach will allow for the effective selection of high-risk liver cancer populations. The performed sustainable CAD device acts as an assistant to the radiologists, helping to identify the area of cancer in the CT scaffold images, take biopsies from those areas and make a better diagnosis.
... The time consuming of region growing is the most serious problem. The method in [4,[15][16][17][18][19][20][21][22][23] that depends on automatic liver segmentation, it records disadvantages in gray level. In addition to, the approaches that depending on gray level; the estimation of liver gray level is considered the main and primary step thus, the intra and large inter gray level variability will neglect. ...
... Datasets of 30 subjects (Men: 16 When scanning a patient for contrast and noncontrast enhanced CT images the several protocols have been employed. The contrast agents (oral and/or IV) have been usually used to produce contrast enhanced CT images as well as, the scanner itself usually contains on unit of an attenuation correction. ...
Article
Full-text available
Evaluating the diffused fat in the liver requires an accurate segmentation for the liver tissues. Miss segmenting the non-liver organs or tissues may lead to a negative impact on the credibility of the obtained results. The segmenting of liver has been proposed as adaptive method by using non-contrast enhanced CT images (NCT). In this method, minimizing the error of segmenting non-liver tissues is the main objective. The proposed method is improved the robustness without utilized training data in building our model or in calculating all parameters. A fully automatic liver segmentation method is suggested in this paper. In this method, the liver of a subject is segmented using NCT data slice-by-slice. The method of segmentation is based on using thresholding operation, gray-level information, Gaussian gradient transformation, region growing algorithm, distance transformation, edge detection and anatomy information. Data sets of 30 subjects are employed to evaluate the proposed method subjectively. Results show a great capability to separate attached organs from the liver. The ability of the method to segment the liver tissues did not reach to a great level. However, the results of segmented liver can be considered as accepted results for the main objective of this study. The method shows a feasible capability to separate non-liver organs and tissues. The results indicate that chances for mistakenly segmentation for non-liver tissues as liver are very low.
... Also, automated segmentation of small liver tumors showed lower accuracy compared to both manual segmentation and automated segmentation of larger tumors (12,24). As a result, there have been multiple attempts to develop methods of automated liver segmentation using CT (31,32). ...
... Results of our vRECIST measurements are likely to provide an early marker for TACE monitoring (23,28,32). We found that A-vRECIST measurements made using our neural network model could be a good substitute for M-vRECIST measurements and mRECIST (Figure 5). ...
Article
Full-text available
Background: Hepatocellular carcinoma (HCC) is the most common liver malignancy and the leading cause of death in patients with cirrhosis. Various treatments for HCC are available, including transarterial chemoembolization (TACE), which is the commonest intervention performed in HCC. Radiologic tumor response following TACE is an important prognostic factor for patients with HCC. We hypothesized that, for large HCC tumors, assessment of treatment response made with automated volumetric response evaluation criteria in solid tumors (RECIST) might correlate with the assessment made with the more time- and labor-intensive unidimensional modified RECIST (mRECIST) and manual volumetric RECIST (M-vRECIST) criteria. Accordingly, we undertook this retrospective study to compare automated volumetric RECIST (A-vRECIST) with M-vRECIST and mRESIST for the assessment of large HCC tumors' responses to TACE. Methods:We selected 42 pairs of contrast-enhanced computed tomography (CT) images of large HCCs. Images were taken before and after TACE, and in each of the images, the HCC was segmented using both a manual contouring tool and a convolutional neural network. Three experienced radiologists assessed tumor response to TACE using mRECIST criteria. The intra-class correlation coefficient was used to assess inter-reader reliability in the mRECIST measurements, while the Pearson correlation coefficient was used to assess correlation between the volumetric and mRECIST measurements. Results:Volumetric tumor assessment using automated and manual segmentation tools showed good correlation with mRECIST measurements. For A-vRECIST and M-vRECIST, respectively, r = 0.597 vs. 0.622 in the baseline studies; 0.648 vs. 0.748 in the follow-up studies; and 0.774 vs. 0.766 in the response assessment (P < 0.001 for all). The A-vRECIST evaluation showed high correlation with the M-vRECIST evaluation (r = 0.967, 0.937, and 0.826 in baseline studies, follow-up studies, and response assessment, respectively, P < 0.001 for all). Conclusion:Volumetric RECIST measurements are likely to provide an early marker for TACE monitoring, and automated measurements made with a convolutional neural network may be good substitutes for manual volumetric measurements.
... Also, automated segmentation of small liver tumors showed lower accuracy compared to both manual segmentation and automated segmentation of larger tumors (12,24). As a result, there have been multiple attempts to develop methods of automated liver segmentation using CT (31,32). ...
... Results of our vRECIST measurements are likely to provide an early marker for TACE monitoring (23,28,32). We found that A-vRECIST measurements made using our neural network model could be a good substitute for M-vRECIST measurements and mRECIST (Figure 5). ...
Article
Purpose: The aims of the study were to assess the typical and atypical radiologic features of pathologically proven adrenal adenomas and to determine the relationship between the radiologic and histopathologic classification. Methods: We retrospectively studied 156 pathologically proven adrenal adenomas in 154 patients from our institutional databases who have computed tomography (CT) and/or magnetic resonance imaging (MRI) examinations before intervention. We determined the histopathologic diagnosis (typical or atypical) using Weiss scoring and classified the adenomas radiologically into typical, atypical, or indeterminate based on lesion size, precontrast CT attenuation, absolute percentage washout, calcification, and necrosis. The κ statistic was used to assess the agreement between radiologists. The Fisher exact test was used to compare the radiologic and pathological classifications. Results: In consensus, there were 83 typical, 42 atypical, and 31 indeterminate adrenal lesions. Logistic regression model showed that radiologically atypical adenoma was significantly associated with larger size, lobulated shape, higher unenhanced CT attenuation, heterogeneous appearance, nonfunctioning status, absolute percentage washout of less than 60%, and a signal intensity index of less than 16.5%.Pathologically, 147 adenomas were pathologically typical (Weiss 0), and 9 adenomas were pathologically atypical (Weiss 1-2). Radiologically, there was substantial agreement between both readers, with Cohen κ at 0.71. Approximately 98% of radiologically typical adenomas were pathologically typical. Only 17% of radiologically atypical adenomas were pathologically atypical. All radiologically indeterminate adenomas were pathologically typical. However, some of the radiologically indeterminate and typical adenomas still had an atypical component on pathologic analysis, such as necrosis, nuclear atypia, or oncocytic features. Conclusions: Radiologically atypical lesion was significantly associated with larger size and higher unenhanced CT attenuation. Approximately 27% of the cases demonstrated atypical features on imaging. Most radiologically atypical adrenal adenomas are pathologically typical.
... In 2006, Lim et al. [13] discussed the difficulty of constructing a 3D model of the liver and the extraction of features from CT images, they suggested the direct separation of areas suspected of the disease. ...
... The first is mean squre error (MSE) and the second is peak signal to noise ratio (PSNR). (13) where I is the main image of the input CT and K is the image restored after the noise is removed and the details are reinforced. This is the difference between the predicted value of the model or the statistical estimator and the actual value. ...
Article
Full-text available
Liver cancer emerges as a mass in the right upper of the abdomen with general symptoms such as jaundice and weakness. In recent years, liver cancer has been responsible for increasing the rate of deaths. Due to some discrepancies in the analytical results of CT images and the disagreement among specialists about different parts of the liver, accurate diagnosis of possible conditions requires skill, experience, and precision. In this paper, a new integrative model based on image processing techniques and machine learning is provided, which is used for segmentation of damages caused by the liver disease on CT images. The implementation process consists of three steps: (1) using discrete wavelet transform to remove noise and separate the region of interest (ROI) in the image; (2) creating the recognition pattern based on feature extraction by Gray-Level Co-occurrence matrix and hierarchical visual HMAX model; reducing the feature dimensions is also optimized by principal component analysis and support vector machine (SVM) classification, and finally (3) evaluating the algorithm performance by using K-fold method. The results of implementation were satisfactory both in performance evaluation and use of features selection. The mean recognition accuracy on test images was 91.7%. The implementation was in the presence of both descriptors irrespective of feature dimension reduction; with unique HMAX model and feature dimension reduction and application of both descriptors and reduction of feature dimensions and their effect on recognition were measured.
... Researches on human organ segmentation mainly focus on image processing methods (such as thresholds, edges, regions, etc.) [1][2][3][4]. Lim et al. [3] proposed multiple thresholds and morphological filtering methods to get the liver boundary, which is suitable for the CT images with distinct difference between the target and its surrounding tissues. Lin et al. [4] proposed an adaptive region growing algorithm, and used prior knowledge to locate kidney tissue. ...
... Researches on human organ segmentation mainly focus on image processing methods (such as thresholds, edges, regions, etc.) [1][2][3][4]. Lim et al. [3] proposed multiple thresholds and morphological filtering methods to get the liver boundary, which is suitable for the CT images with distinct difference between the target and its surrounding tissues. Lin et al. [4] proposed an adaptive region growing algorithm, and used prior knowledge to locate kidney tissue. ...
Article
Full-text available
The segmentation challenge of adrenal and surrounding tissues lie in the similar CT values and adhesion in a medical image. An adrenal segmentation model (SALS) based on shape associating level set is proposed to segment the adrenal accurately. The objective function of adrenal boundary is expressed by a level set model. The prior shape curve of the adrenal to be segmented is calculated by the gradual change relationship between the adrenal glands in adjacent CT images, and is converted into shape constraint term of the level set models. A 3D Laplace method is used to improve adrenal grey scale information. The improved result of CT image is converted into grey scale constraint term of the level set models. Under the two constraints, the objective function of adrenal boundary converges to the best boundary. The level set methods in the literature obtain a prior shape by training a large number of sample images, and use the shape modes to segment CT images. The SALS model does not depend on the sample images. The adrenal boundaries in sequence of CT images can be directly segmented by SALS model, and the segmented boundaries are more accurate than the traditional level set methods. The SALS model has stronger adaptability to adherent adrenal boundary.
... There are many proposed methods to segment liver in decades [1]- [10]. The most simple and intuitive approaches to perform liver segmentation are thresholding and region growing [1], [2]. ...
... There are many proposed methods to segment liver in decades [1]- [10]. The most simple and intuitive approaches to perform liver segmentation are thresholding and region growing [1], [2]. Active contour model (ACM) approaches [3], [4] were also presented mainly using intensity distributions. ...
Preprint
Accurate segmentation of liver is still challenging problem due to its large shape variability and unclear boundaries. The purpose of this paper is to propose a neural network based liver segmentation algorithm and evaluate its performance on abdominal CT images. First, we develop fully convolutional network (FCN) for volumetric image segmentation problem. To guide a neural network to accurately delineate target liver object, we apply self-supervising scheme with respect to edge and contour responses. Deeply supervising method is also applied to our low-level features for further combining discriminative features in the higher feature dimensions. We used 160 abdominal CT images for training and validation. Quantitative evaluation of our proposed network is presented with 8-fold cross validation. The result showed that our method successfully segmented liver more accurately than any other state-of-the-art methods without expanding or deepening the neural network. The proposed approach can be easily extended to other imaging protocols (e.g., MRI) or other target organ segmentation problems without any modifications of the framework.
... In order to solve the problem of liver CT image segmentation, many methods have been proposed by experts and researchers. Traditional liver segmentation methods are categorized into: intensity threshold (Lim et al. 2006;Soler et al. 2001), region growing (Ruskó et al. 2007;Pohle and Tönnies 2001), and deformable model (Kainmüller et al. 2007;Park et al. 2003). ...
Article
Full-text available
Introduction The automatic segmentation of the liver is a crucial step in obtaining quantitative biomarkers for accurate clinical diagnosis and computer-aided decision support systems. This task is challenging due to the frequent presence of noise and sampling artifacts in computerized tomography (CT) images, as well as the complex background, variable shapes, and blurry boundaries of the liver. Standard segmentation of medical images based on full-supervised convolutional networks demands accurate dense annotations. Such a learning framework is built on laborious manual annotation with strict requirements for expertise, leading to insufficient high-quality labels. Methods To overcome such limitation and exploit massive weakly labeled data, we relaxed the rigid labeling requirement and developed a semi-supervised double-cooperative network (SD- Net). SD-Net is trained to segment the complete liver volume from preoperative abdominal CT images by using limited labeled datasets and large-scale unlabeled datasets. Specifically, to enrich the diversity of unsupervised information, we construct SD-Net consisting of two collaborative network models. Within the supervised training module, we introduce an adaptive mask refinement approach. First, each of the two network models predicts the labeled dataset, after which adaptive mask refinement of the difference predictions is implemented to obtain more accurate liver segmentation results. In the unsupervised training module, a dynamic pseudo-label generation strategy is proposed. First each of the two models predicts unlabeled data and the better prediction is considered as pseudo-labeling before training. Results and discussion Based on the experimental findings, the proposed method achieves a dice score exceeding 94%, indicating its high level of accuracy and its suitability for everyday clinical use.
... 21,22 As the intensities of adjacent organs and tissues are very similar to liver tissue itself, threshold 3 and regiongrowing 4 methods are sensitive to noises and easy to cause leakage of the liver contour into nearby tissues, but they play an essential role in obtaining the initial liver mask. [23][24][25][26] Graph-cuts based methods [7][8][9] have good global optimization properties and topology variability, but they must first define the initial contour of the liver, and it is easy to misclassify lesions as non-liver tissues. Shape model-based methods [12][13][14] can effectively accomplish the initialization or refinement of liver segmentation, but they heavily depend on training shapes, 10 and their matching step is timeconsuming. ...
Article
Full-text available
Purpose Automatic liver segmentation from computed tomography (CT) images is an essential preprocessing step for computer‐aided diagnosis of liver diseases. However, due to the large differences in liver shapes, low‐contrast to adjacent tissues, and existence of tumors or other abnormalities, liver segmentation has been very challenging. This study presents an accurate and fast liver segmentation method based on a novel probabilistic active contour (PAC) model and its fast global minimization scheme (3D‐FGMPAC), which is explainable as compared with deep learning methods. Methods The proposed method first constructs a slice‐indexed‐histogram to localize the volume of interest (VOI) and estimate the probability that a voxel belongs to the liver according its intensity. The probabilistic image would be used to initialize the 3D PAC model. Secondly, a new contour indicator function, which is a component of the model, is produced by combining the gradient‐based edge detection and Hessian‐matrix‐based surface detection. Then, a fast numerical scheme derived for the 3D PAC model is performed to evolve the initial probabilistic image into the global minimizer of the model, which is a smoothed probabilistic image showing a distinctly highlighted liver. Next, a simple region‐growing strategy is applied to extract the whole liver mask from the smoothed probabilistic image. Finally, a B‐spline surface is constructed to fit the patch of the rib cage to prevent possible leakage into adjacent intercostal tissues. Results The proposed method is evaluated on two public datasets. The average Dice score, volume overlap error, volume difference, symmetric surface distance and volume processing time are 0.96, 7.35%, 0.02%, 1.17 mm and 19.8 s for the Sliver07 dataset, and 0.95, 8.89%, −0.02%0.02%-0.02\%, 1.45 mm and 23.08 s for the 3Dircadb dataset, respectively. Conclusions The proposed fully‐automatic approach can effectively segment the liver from low‐contrast and complex backgrounds. The quantitative and qualitative results demonstrate that the proposed segmentation method outperforms state‐of‐the‐art traditional automatic liver segmentation algorithms and achieves very competitive performance compared with recent deep leaning‐based methods.
... is surface helps to create a geodesic active contour which is the same as the true liver boundary. e gray level methods are presented [62][63][64][65] in which the histogram of the whole volume through the preset gray level range for the identification of liver peak having two thresholds [62]. e purpose of these thresholds is to determine a liver binary volume which is processed heavily to delete the organ through morphological operators. ...
Article
Full-text available
Segmentation of a liver in computed tomography (CT) images is an important step toward quantitative biomarkers for a computer-aided decision support system and precise medical diagnosis. To overcome the difficulties that come across the liver segmentation that are affected by fuzzy boundaries, stacked autoencoder (SAE) is applied to learn the most discriminative features of the liver among other tissues in abdominal images. In this paper, we propose a patch-based deep learning method for the segmentation of a liver from CT images using SAE. Unlike the traditional machine learning methods, instead of anticipating pixel by pixel learning, our algorithm utilizes the patches to learn the representations and identify the liver area. We preprocessed the whole dataset to get the enhanced images and converted each image into many overlapping patches. These patches are given as input to SAE for unsupervised feature learning. Finally, the learned features with labels of the images are fine tuned, and the classification is performed to develop the probability map in a supervised way. Experimental results demonstrate that our proposed algorithm shows satisfactory results on test images. Our method achieved a 96.47% dice similarity coefficient (DSC), which is better than other methods in the same domain.
... Lim vd. [28] ise çalışmalarında, morfolojik prosedürler, kümeleme, etiketleme ve temel görüntü işleme teknikleri ile BT görüntüleri üzerinde karaciğerin otomatik bölütlemesini uygulamışlardır. Bir başka çalışmada, Huang vd. ...
Article
Full-text available
Bilgisayarlı Tomografi (BT) görüntülerinde her bir kesitte ortaya çıkan şekil, sınır ve yoğunluk gibi değişikliklerden dolayı karaciğerin bölütlenmesi zor bir süreç olarak durmaktadır. Diğer bölütleme yöntemleri ile karşılaştırıldığında, derin öğrenme modelleri ile daha başarılı bölütleme sonuçları genel fenomendir. Bu çalışmada, abdomen bölgesinden alınmış BT taramalarındaki kesitler üzerinde karaciğerin bilgisayar destekli otomatik bölütlenmesi için, Maskeli Bölgesel-Evrişimsel Sinir Ağları (Maskeli B-ESA) kullanılarak çoklu-GPU ile hızlandırılmış bir yöntem önerilmiştir. Bu çalışmaya özgü hazırlanan karaciğer BT görüntü veriseti üzerinde, hem tek hem de çift GPU donanımsal yapısı ile deneysel çalışmalar yürütülmüştür. Önerilen yöntem kullanılarak elde edilen sonuçlar ile uzman hekim tarafından bulunan bölütleme sonuçları Dice benzerlik katsayısı (DSC), Jaccard benzerlik katsayısı (JSC), volumetrik örtüşme hatası (VOE), ortalama simetrik yüzey mesafesi (ASD) ve oransal hacim farkı (RVD) ölçüm parametreleri ile karşılaştırılmıştır. Önerilen yaklaşım ile test görüntüleri üzerinde yürütülen deneysel çalışmalarda DSC, JSC, VOE, ASD ve RVD bölütleme başarım metrikleri, sırasıyla 97.32, 94.79, 5.21, 0.390, -1.008 olarak hesaplanmıştır. Bu sonuçlar ile bu çalışma kapsamında önerilen yöntemin, karaciğerin bölütlenmesi için hekimlerin karar verme süreçlerinde yardımcı bir araç olarak kullanılabileceği görülmüştür.
... Although CNN and LightGBM show good performance in wrinkle and cell segmentation, their results are not always satisfactory. For instance, the segmentation includes areas wherein it is unclear whether they contain wrinkles or cells, which degrades the reliability of the segmentation results [36][37][38][39]. Another problem is false segmentation pixels, which appear as noise. ...
Article
Full-text available
Objective and quantitative assessment of skin conditions is essential for cosmeceutical studies and research on skin aging and skin regeneration. Various handcraft-based image processing methods have been proposed to evaluate skin conditions objectively, but they have unavoidable disadvantages when used to analyze skin features accurately. This study proposes a hybrid segmentation scheme consisting of Deeplab v3+ with an Inception-ResNet-v2 backbone, LightGBM, and morphological processing (MP) to overcome the shortcomings of handcraft-based approaches. First, we apply Deeplab v3+ with an Inception-ResNet-v2 backbone for pixel segmentation of skin wrinkles and cells. Then, LightGBM and MP are used to enhance the pixel segmentation quality. Finally, we determine several skin features based on the results of wrinkle and cell segmentation. Our proposed segmentation scheme achieved a mean accuracy of 0.854, mean of intersection over union of 0.749, and mean boundary F1 score of 0.852, which achieved 1.1%, 6.7%, and 14.8% improvement over the panoptic-based semantic segmentation method, respectively.
... We can divide them into two categories, namely traditional methods with no deep learning involved and deep learning methods. Traditional methods can be further divided into seven sub-categories, namely, grayscale based methods (based on histogram [1], based on threshold [2], and based on clustering [3]), graph cut based methods [4], region growing based methods [5], probability map based methods [6], statistical shape model based methods [7], level set based methods [8], active contour model based methods [9]. Deep learning methods can be further divided into 2D methods and 3D methods. ...
Article
Full-text available
Liver cancer has become a major disease that seriously endangers people’s lives and health. In clinical practice, it is necessary to segment the liver, lesions, and other normal organs and tissues in abdominal CT images accurately. However, segmentation of abdominal organs and tissues is a critical and time-consuming process in radiotherapy. Therefore, it is very important to use fully automated and high-precision methods to segment and reconstruct abdominal multi-tissues. This article integrates the automatic recognition process and main segmentation algorithms of skin, bones, intrahepatic blood vessels, liver and lesions. In addition, it supports human-computer interaction to modify the segmentation results, as well as the reconstruction of 3D surface, the measurement of size and volume, and other functions. The threshold segmentation and morphological operations are used in segmentation and reconstruction of skin and bones. And the fuzzy C-means algorithm is utilized in the segmentation and reconstruction of intrahepatic blood vessels. As for the segmentation and reconstruction of liver and lesions, we use convolutional neural network named V-net to accomplish. The experimental results show that the Dice coefficients of the skin and bones segmentation were 98.4% and 92.3%, respectively. The intrahepatic blood vessel segmentation was 78.9%. And the liver and lesions segmentation were 95.4% and 80.4%, respectively. The ensemble algorithm proposed in this paper demonstrated its potential clinical utility in terms of accuracy and time-efficiency.
... The morphological filter utilises region-labelling, and clustering, to examine the liver contour with fixed search range. A labelling-based search algorithm is utilised to deform liver contour [34]. ...
Article
Full-text available
Liver cancer is the major reason for death in this entire world. Manual detection of cancer tissue is found time consuming and difficult. Therefore, the development of an automatic detection approach with high accuracy for liver cancer is considered as the main aim of this work. The image processing approach can use the CAD for the classification of liver cancer in order to assist the physician in the decision‐making process. An automated approach to effective classification of the liver tumour using effective features is conveyed in terms of the CAD system. Traditionally, radiologists delineate the liver and liver lesion on a slice‐by‐slice basis, which is time consuming and susceptible to inter‐ and intra‐rater variations. Automatic methods for the segmentation of liver and liver tumours are, thus, highly demanded in clinical practice. A systematic review is being carried out to detect reported liver cancer from 2013 to 2020. Finally, a more precise technical direction is provided for all researchers in this review. Research gaps for earlier detection and its potential future aspects are also discussed.
... Mittal and Kumari used gradient based segmentation method to find true edges in their work [5]. Lim et al. developed automatic liver segmentation method using adaptive thresholding and iterative morphological filtering [6]. They obtained an accuracy of 96%. ...
Article
Full-text available
Segmentation of liver is the initial and fundamental step for the diagnosis of liver disease, 3-D volume construction and volume measurement. However, segmentation of liver is a challenging task due to it's inter and intra intensity and texture similarities among other organs present in CT abdominal images. A semiautomatic method has been proposed to segment the liver portion from CT abdominal images and their three dimensional volume construction by (i)Noise removal using median filter,(ii)Segmentation of liver portion based on active contour method using sparse field method and(iii)liver volume construction using marching cube method. Evaluation of proposed method is carried out on clinically acquired CT images and effectiveness of algorithm is evaluated by comparing manually segmented liver portion marked by radiologist with proposed method.
... In order to evaluate results of 3d fast marching segmentation method on our data, we have computed the volume of the segmented tumor using an estimation which focus on some characteristics of the CT data such as thickness and interval information of the slice and the size of the voxel [2] and we have used two metrics : volume overlap (m1) and relative absolute volume difference (m2) defined by the following expressions : ...
Conference Paper
Liver tumor treatments, such as percutaneous alcohol injection operation, require the knowledge of tumor position and volume measurement in order to estimate the appropriate dose of injection product and to insure accurate localization of needle tip for an efficient treatment. We propose, in this paper, an approach that estimates the volume and the exact location of the tumor from segmented computed tomography images. The first stage consists on tumor segmentation by means of a 3d fast marching level-set method. In the second stage, we can estimate the tumor volume by computing the number of voxels constituting the segmented region and find out its exact location as the mass center. Later, this 3d location is turned to the radiologist repository. We have tested our approach on two clinical datasets and we have got interesting results and promising performance.
... However, it is less accurate for most tumors, as they usually have irregular borders. So, to reach a good estimation of this kind of volume, some studies have focus on some characteristics of the CT data such as thickness and interval information of the slice and size of the voxel (Lim, Jeong, and Ho 2006). This measurement can be defined by the following equation: ...
Conference Paper
Liver cancer is one of the most popular cancer diseases and causes a large amount of death every year (Parkin, Bray, Ferlay, and Pisani 2005). In order to make decisions and interventions such as radiofrequency ablation (RFA) and percutaneous alcohol injection (PAI), doctors need to know the volume measurement and the center position of the tumor in order to estimate the appropriate dose of alcohol and to guarantee an equitable diffusion. Thus, an important task in radiology is the determination of tumor volume and tumor center position. In this paper, we propose an approach that estimate the volume and the center position from respectively segmented computed tomography (CT) scans and reconstructed tridimensional (3D) mesh of the tumor. At first, the seg-mentation of the liver tumor is carried out. We propose to use the Fast Marching (FM) algorithm (Sethian 1996) which is a kind of the level-set segmentation method. This method requires the selection of one seed voxel in the center of the liver tumor from CT data. Then, we apply the Marching Cubes (MC) algorithm (Lorensen and Cline 1987) to all segmented slices in order to extract an isosurface of the liver tumor and consequently a triangular 3D mesh. Given the resolution of the CT scans, we can estimate the tumor volume by computing the number of voxels which constitute it. The tumor center position can be found by computing the center of gravity of the tumor 3D mesh. Later, this 3D position is turned to the doctor repository: we give the CT scan number in which you detect the center and the coordinates of this center with regard to a reference marker.
... Entre os métodos de segmentação automática se destacam alguns algoritmos baseados em crescimento de região [36][37][38] por meio de sementes posicionadas sobre o tecido, técnicas de segmentação por limiarização do histograma 31,32 seguida de transformações morfológicas 39 de ajuste de atlas morfológicos por meio de transformação afim 40 . ...
Article
Full-text available
Este artigo tem por objetivo apresentar uma abordagem conceitual sobre os principais aspectos envolvidos no processamento e na análise digital de imagens médicas, trazendo exemplos da aplicação na prática clínica e da pesquisa em imagens médicas. Para explorar a temática, o artigo está dividido em seções. A primeira seção apresenta os aspectos relacionados às diferenças entre a imagem adquirida no equipamento e a visualizada nos monitores, levantando alguns elementos relacionados à qualidade da aquisição. A seguir são descritas algumas técnicas de pré-processamento que permitem melhorar e destacar aspectos relevantes das imagens. A próxima seção apresenta os principais métodos de segmentação de objetos de interesse nas imagens. A seguir, duas seções descrevem como representar e descrever de forma quantitativa as características relevantes das imagens, para que elas possam ser analisadas computacionalmente, e os aspectos relativos à análise e ao reconhecimento de padrões em imagens. Finalmente, são apresentados alguns exemplos de esquemas de auxílio computadorizado ao diagnóstico.
... Classification methods consist of grouping individual components of the image such as pixels or sub-images and exploiting their similarities as opposed to the contour approach seeking dissimilarities, while model based methods can be statistical or supported by an atlas. Region growing methods [1], [2], histograms with thresholds [3], voxel classification algorithms [4], and graph cuts [5] are widely used as classification methods to segment the liver, but they often lead to over-segmentation problems [6]. In the contour approaches, some researchers improve the existing methods, for example Shi et al. [7] and Liu et al. [8] present improved deformable shape and contour models respectively. ...
Article
Developing methods to segment the liver in medical images, study and analyze it remains a significant challenge. The shape of the liver can vary considerably from one patient to another, and adjacent organs are visualized in medical images with similar intensities, making the boundaries of the liver ambiguous. Consequently, automatic or semi-automatic segmentation of liver is a difficult task. Moreover, scanning systems and magnetic resonance imaging have different settings and parameters. Thus the images obtained differ from one machine to another. In this article, we propose an automatic model-based segmentation that allows building a faithful 3-D representation of the liver, with a mean Dice value equal to 90.3% on CT and MRI datasets. We compare our algorithm with a semi-automatic method and with other approaches according to the state of the art. Our method works with different data sources, we use a large quantity of CT and MRI images from machines in various hospitals and multiple DICOM images available from public challenges. Finally, for evaluation of liver segmentation approaches in state of the art, robustness is not adequacy addressed with a precise definition. Another originality of this article is the introduction of a novel measure of robustness, which takes into account the liver variability at different scales.
... Three main types of method are used for liver segmentation: classification, active contour detection, and model-based methods. Classification methods consist of grouping individual components of the image such as pixels or sub-images and exploiting their similarities as region-growing methods [3,4], histograms with thresholds [5], voxel classification algorithms [6], and graph cuts [7]. They often lead to over-segmentation. ...
Article
Background: Proper segmentation of the liver from medical images is critical for computer-assisted diagnosis, therapy and surgical planning. Knowledge of its vascular structure allows division of the liver into eight functionally independent segments, each with its own vascular inflow, known as the Couinaud scheme. Couinaud's description is the most widely used classification, since it is well-suited for surgery and accurate for the localization of lesions. However, automatic segmentation of the liver and its vascular structure to construct the Couinaud scheme remains a challenging task. Methods: We present a complete framework to obtain Couinaud's classification in three main steps; first, we propose a model-based liver segmentation, then a vascular segmentation based on a skeleton process, and finally, the construction of the eight independent liver segments. Our algorithms are automatic and allow 3D visualizations. Results: We validate these algorithms on various databases with different imaging modalities (Magnetic Resonance Imaging (MRI) and Computed Tomography (CT)). Experimental results are presented on diseased livers, which pose complex challenges because both the overall organ shape and the vessels can be severely deformed. A mean DICE score of 0.915 is obtained for the liver segmentation, and an average accuracy of 0.98 for the vascular network. Finally, we present an evaluation of our method for performing the Couinaud segmentation thanks to medical reports with promising results. Conclusions: We were able to automatically reconstruct 3-D volumes of the liver and its vessels on MRI and CT scans. Our goal is to develop an improved method to help radiologists with tumor localization.
... A more recent works by Lim are also basing on the tree steps mention above (Lim et al., 2004;Lim et al., 2006;Lim et al., 2005). They estimates the liver grey value by analyzing the intensities distribution on a subset of manually segmented lever images together with prior knowledge of liver location. ...
Article
Full-text available
Segmentation of liver images containing disconnected regions has always been an overlooked problem. Previous works on liver segmentation either ignore this problem or use manual initialization when facing these disconnected regions. Therefore, in this paper we propose a liver level set (LLS) algorithm which is able to segment disconnected regions automatically. The LLS algorithm is based on level set framework together with hybrid energy minimization as the stopping function. By using the LLS algorithm in a looping manner, we allow the current liver boundary to inherit the topological changes from previous images in a 2.5D environment. We also conduct an experiment to obtain an average factor for dynamic localization region sizes based on liver anatomy to improve the segmentation accuracy. These dynamic localization region sizes ensure a more accurate segmentation when compared with manual segmentation. Our experiment gives a respective segmentation result with dice similarity coefficient (DSC) percentage of 87.5%. Plus, our LLS algorithm is able to segment all connected and disconnected liver region automatically and accurately.
... Third, regarding segmentation of liver CT images, traditional methods, such as clustering [27]- [29], and morphological operators [30] used to help clinicians classify texture in feature space. These intensity-based methods are usually efficient and can give excellent results when the liver's intensity is sufficient. ...
Article
Full-text available
Image segmentation is typically used to locate objects and boundaries. It is essential in many clinical applications, such as pathological diagnosis of hepatic diseases, surgical planning, and postoperative assessment. The segmentation task is hampered by fuzzy boundaries, complex backgrounds, and appearances of objects of interest, which vary considerably. The success of the procedure is still highly dependent on the operator’s skills and level of hand-eye coordination. Thus, this study was strongly motivated by the necessity to obtain an early and accurate diagnosis of a detected object in medical images. In this paper, we propose a new polyp segmentation method based on the architecture of a multiple deep encoder-decoder networks combination called CDED-net. The architecture can not only hold multi-level contextual information by extracting discriminative features at different effective fields-of-view and multiple image scales, but also learn rich information features from missing pixels in the training phase. Moreover, the network is also able to capture object boundaries by using multiscale effective decoders. We also propose a novel strategy for improving the method’s segmentation performance based on a combination of a boundary-emphasization data augmentation method and a new effective dice loss function. The goal of this strategy is to make our deep learning network available with poorly defined object boundaries, which are caused by the non-specular transition zone between the background and foreground regions. To provide a general view of the proposed method, our network was trained and evaluated on three well-known polyp datasets, CVC-ColonDB, CVC-ClinicDB, and ETIS-Larib PolypDB. Furthermore, we also used the Pedro Hispano Hospital (PH2), ISBI 2016 skin lesion segmentation dataset and CT Healthy Abdominal Organ Segmentation dataset to depict our network’s ability. Our results reveal that the CDED-net significantly surpasses state-of-the-art methods.
... Image registration accuracy has been investigated for CT to CT liver registration for contrast-enhanced diagnostic CTs [38]. Over the past decade, numerous semi-automatic and automatic approaches for liver segmentation [39,40] on CT that rely on histogrambased methods [41,42], graph cut [43][44][45], region growing [45][46][47], geometric deformable model and level set [48][49][50], probabilistic atlas [51,52], statistical shape models [53][54][55], and recently neural network [56][57][58][59] have been proposed. Despite these efforts, image registration and segmentation remains a challenging task for SIRT application for several reasons: (1) liver is a soft tissue and liver shape is heavily dependent on patient positioning (e.g., the position of the arms); (2) the liver shape in SIRT patients differs from the normal shape, because of preceding treatments (liver resection, liver ablation, chemotherapy) and tumor growth which makes it challenging to use liver segmentation techniques which are dependent on the liver shape for these patients; (3) liver is a soft tissue and its Hounsfield units are similar to those of adjacent organs like the heart, spleen, stomach, and kidney, which makes liver segmentation on non-contrast-enhanced CTs (e.g., CT from MAA study) hard, even for experts; (4) CT from MAA study is not a dedicated diagnostic CT, this low-dose CT usually suffers from streak artifacts; and (5) the interval between the MAA study and the diagnostic high-dose, contrast-enhanced CT from from fluorine-18 fluorodeoxyglucose ( 18 F-FDG) PET/CT study can be up to weeks to even 1 or 2 months and the liver can deform dramatically over time for several reasons, e.g., tumor change. ...
Article
Full-text available
Purpose: We have developed a multi-modal imaging approach for SIRT, combining 99mTc-MAA SPECT/CT and/or 90Y PET, 18F-FDG PET/CT, and contrast-enhanced CBCT for voxel-based dosimetry, as a tool for treatment planning and verification. For radiation dose prediction calculations, a segmentation of the total liver volume and of the liver perfusion territories is required. Method: In this paper, we proposed a procedure for multi-modal image analysis to assist SIRT treatment planning. The pre-treatment 18F-FDG PET/CT, 99mTc-MAA SPECT/CT, and contrast-enhanced CBCT images were registered to a common space using an initial rigid, followed by a deformable registration. The registration was scored by an expert using Likert scores. The total liver was segmented semi-automatically based on the PET/CT and SPECT/CT images, and the liver perfusion territories were determined based on the CBCT images. The segmentations of the liver and liver lobes were compared to the manual segmentations by an expert on a CT image. Result: Our methodology showed that multi-modal image analysis can be used for determination of the liver and perfusion territories using CBCT in SIRT using all pre-treatment studies. The results for image registration showed acceptable alignment with limited impact on dosimetry. The image registration performs well according to the expert reviewer (scored as perfect or with little misalignment in 94% of the cases). The semi-automatic liver segmentation agreed well with manual liver segmentation (dice coefficient of 0.92 and an average Hausdorff distance of 3.04 mm). The results showed disagreement between lobe segmentation using CBCT images compared to lobe segmentation based on CT images (average Hausdorff distance of 14.18 mm), with a high impact on the dosimetry (differences up to 9 Gy for right and 21 Gy for the left liver lobe). Conclusion: This methodology can be used for pre-treatment dosimetry and for SIRT planning including the determination of the activity that should be administered to achieve the therapeutical goal. The inclusion of perfusion CBCT enables perfusion-based definition of the liver lobes, which was shown to be markedly different from the anatomical definition in some of the patients.
... Gray level methods utilize statistics to estimate the intensity distribution of the liver. Lim et al. proposed several methods to segment liver using morphological filters and intensity distributions [38,39], and pattern features [40]. Graph Cut (GC) methods segments the image into background and liver. ...
Preprint
Full-text available
In this work, we report the set-up and results of the Liver Tumor Segmentation Benchmark (LITS) organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI) 2016 and International Conference On Medical Image Computing Computer Assisted Intervention (MICCAI) 2017. Twenty four valid state-of-the-art liver and liver tumor segmentation algorithms were applied to a set of 131 computed tomography (CT) volumes with different types of tumor contrast levels (hyper-/hypo-intense), abnormalities in tissues (metastasectomie) size and varying amount of lesions. The submitted algorithms have been tested on 70 undisclosed volumes. The dataset is created in collaboration with seven hospitals and research institutions and manually reviewed by independent three radiologists. We found that not a single algorithm performed best for liver and tumors. The best liver segmentation algorithm achieved a Dice score of 0.96(MICCAI) whereas for tumor segmentation the best algorithm evaluated at 0.67(ISBI) and 0.70(MICCAI). The LITS image data and manual annotations continue to be publicly available through an online evaluation system as an ongoing benchmarking resource.
... To solve this problem, many researches use a specific process (Militello et al., 2015;Lim et al., 2006;Khayati et al., 2008). This process starts with enhancing the image in a pretreatment phase. ...
... Through the trained SVM model, all super-pixel blocks in the CT images could be automatically classified. The SVM reduces the computational efforts and identifies the initial liver boundary [32]. ...
Article
Full-text available
Accurate segmentation of specific organ from computed tomography (CT) scans is a basic and crucial task for accurate diagnosis and treatment. To avoid time-consuming manual optimization and to help physicians distinguish diseases, an automatic organ segmentation framework is presented. The framework utilized convolution neural networks (CNN) to classify pixels. To reduce the redundant inputs, the simple linear iterative clustering (SLIC) of super-pixels and the support vector machine (SVM) classifier are introduced. To establish the perfect boundary of organs in one-pixel-level, the pixels need to be classified step-by-step. First, the SLIC is used to cut an image into grids and extract respective digital signatures. Next, the signature is classified by the SVM, and the rough edges are acquired. Finally, a precise boundary is obtained by the CNN, which is based on patches around each pixel-point. The framework is applied to abdominal CT scans of livers and high-resolution computed tomography (HRCT) scans of lungs. The experimental CT scans are derived from two public datasets (Sliver 07 and a Chinese local dataset). Experimental results show that the proposed method can precisely and efficiently detect the organs. This method consumes 38 s/slice for liver segmentation. The Dice coefficient of the liver segmentation results reaches to 97.43%. For lung segmentation, the Dice coefficient is 97.93%. This finding demonstrates that the proposed framework is a favorable method for lung segmentation of HRCT scans.
... Boykov et al. [6] proposed Graph Cut algorithm as a boundary based segmentation method. Except these methods, gray level based methods are used for automatic liver segmentation [7][8][9]. Jayanthi [10] compared the various methods for segmentation of the liver. Different algorithms such as seeded region growing, Neutroshophic set with thresholding, label connected were used to segment the liver in the study. ...
Conference Paper
Full-text available
This study is an implementation of liver segmentation on abdomen CT images. The liver organ was segmented by using SLIC super-pixel and AdaBoost algorithms. Firstly, the images were clustered by SLIC super-pixel algorithm. Then, the liver was segmented by AdaBoost classifier. The segmentation process was done automatically. The automatic segmentation is based on the classification of overlapping patches of the image. The results of automatic segmentation and manual segmentation were compared and the efficiency of the method was observed. The best Dice rate was obtained as 92.13% and the best Jaccard rate was obtained as 85.8% on 16 abdomen CT images.
... It is used commonly alongside other metrics. 5,12,19,20,27,28 ...
Article
All medical image segmentation algorithms need to be validated and compared, yet no evaluation framework is widely accepted within the imaging community. None of the evaluation metrics that are popular in the literature are consistent in the way they rank segmentation results: they tend to be sensitive to one or another type of segmentation error (size, location, and shape) but no single metric covers all error types. We introduce a family of metrics, with hybrid characteristics. These metrics quantify the similarity or difference of segmented regions by considering their average overlap in fixed-size neighborhoods of points on the boundaries of those regions. Our metrics are more sensitive to combinations of segmentation error types than other metrics in the existing literature. We compare the metric performance on collections of segmentation results sourced from carefully compiled two-dimensional synthetic data and three-dimensional medical images. We show that our metrics: (1) penalize errors successfully, especially those around region boundaries; (2) give a low similarity score when existing metrics disagree, thus avoiding overly inflated scores; and (3) score segmentation results over a wider range of values. We analyze a representative metric from this family and the effect of its free parameter on error sensitivity and running time.
Chapter
Automatic and accurate segmentation of liver tumors is crucial for the diagnosis and treatment of hepatocellular carcinoma or metastases. However, the task remains challenging due to imprecise boundaries and significant variations in the shape, size, and location of tumors. The present study focuses on tumor segmentation as a more critical aspect from a medical perspective, compared to liver parenchyma segmentation, which is the focus of most authors in publications. In this paper, four state-of-the-art models were trained and used to compare with UNet in terms of accuracy. Two of them (namely, based on polar coordinates and Visual Image Transformer (ViT)) were adopted for the specified task. Dice similarity measure is used for the comparison. A unified baseline environment and preprocessing parameters were used. Experiments on the public LiTS dataset demonstrate that the proposed ViT based network can accurately segment liver tumors from CT images in an end-to-end manner, and it outperforms many existing methods (tumour segmentation accuracy 56%, liver parenchyma 94% Dice). The average Dice similarity measure for the considered images was found to be 75%. The obtained results seem to be clinically relevant.KeywordsLiver tumour segmentationtransformerconvolutional network
Article
Full-text available
Computer-aided liver diagnosis helps doctors accurately identify liver abnormalities and reduce the risk of liver surgery. Early diagnosis and detection of liver lesions depend mainly on medical imaging techniques such as magnetic resonance imaging (MRI) and computed tomography (CT). Segmentation and identification of hepatic lesions in these images are very challenging because these images often come with low resolution and severe noise. Many new machine learning and image analysis techniques have been gradually used on this topic, and their performance is still limited. An automatic and accurate model that incorporates tracking, detection, and diagnosis of hepatic lesions in the 3D volumes of CT and MRI is still lacking. This paper aims to review different models for the automatic detection and diagnosis of the hepatic lesion with CT and MRI and discusses the medical background of liver tumors and the standard elements of the CAD liver diagnosis system. In addition, the concept of federated learning has been introduced, and the fused information from multi-modality (CT and MRI) and the robust and complex features that represent liver lesions accurately have been discussed. More specifically, this paper presents a comprehensive study of the latest work on liver tumor detection and diagnosis, which identifies the contributions of these different approaches and the recommendation model suggested for practical use. Furthermore, this paper was intended to encourage researchers from the medical community, image processing, and machine learning community to pay much attention to the use of deep and federated learning, spiking neuron model, bio-inspired optimization algorithms, fuzzy logic, and neutrosophic logic to address the problems of segmentation and prediction/classification for real-time diagnosis.
Article
In a medical side liver segmentation is a major problem. It is a much more challenging task among other segmentation than a liver segment. So we have to introduce automatic segmentation of liver in CT images and MRI has become significance in medical area, so we are focusing the managing on domain from MRI to CT volumes on the example of 3D and 2D liver segmentation. Here, we are implemented to fuzzy clustering method and cuckoo optimized method to used in this method. In additional that the purpose of this method is to used for liver boundary calculation and segment the values using fuzzy C-means and fuzzy clustering using Support Vector Machine (SVM) for classification). The spatial liver boundary by constraints are named with different method such as the user can choose on particular region of interest and contour method until more accurate results to be obtained.
Article
Full-text available
In clinical picture handling, the division of the liver in figured tomography pictures is of huge importance. To acquire liver segmentation, there is an examination of strategies for dividing the liver and methods utilizing processed tomography pictures. Isolating plans into two classes are self-loader and fully automated programs. The two classes have a few techniques, estimation, related questions; a few downsides will be depicted and explained. Following the similar review for liver division plans, different assessments and scoring are given; we will cautiously highlight the benefits and inconveniences of procedures. A few deficiencies and hardships of the proposed strategies are still focused around.
Book
This book constitutes the refereed post-conference proceedings of the Fourth IFIP TC 12 International Conference on Computational Intelligence in Data Science, ICCIDS 2021, held in Chennai, India, in March 2021. The 20 revised full papers presented were carefully reviewed and selected from 75 submissions. The papers cover topics such as computational intelligence for text analysis; computational intelligence for image and video analysis; blockchain and data science.
Chapter
Full-text available
According to India Brand Equity Foundation (IBEF), 32% of the global food market is dependent on Indian agricultural sector. Due to urbanisation, the fertile land have been utilised for non-agricultural purposes. The loss of agricultural lands impacts the productivity and results with diminishing yield. Soil is the most important factor for the thriving agriculture, since it contains the essential nutrients. The food production could be improved through the viable usage of soil nutrients. To identify the soil nutrients, the physical, chemical and biological parameters were examined using many machine learning algorithms. However, the environmental factors such as sunlight, temperature, humidity, and rainfall plays a major role in improving the soil nutrients since it is responsible for the process of photosynthesis, germination, and saturation. The objective is to determine the soil nutrient level by accessing the associative properties including the environmental variables. The proposed system termed as Agrarian application which recommends crops for the particular land using classification algorithms and predicts the yield rate by employing regression techniques. The application will help the farmers in selecting the crops based on the soil nutrient content, environmental factors and predicts the yield rate for the same.
Chapter
An automatic method for segmenting the liver from the portal venous phase of abdominal CT images using the K-Means clustering method is described in this paper. We have incorporated an interactive technique for correcting the errors in the liver segmentation results using power law transformation. The proposed method was validated on abdominal CT volumes of fifteen patients obtained from Kasturba Medical College, Manipal. The average values of the various standard evaluation metrics obtained are as follows: Dice coefficient = 0.9361, Jaccard index = 0.8805, volumetric overlap error = 0.1195, absolute volume difference = 4.048%, average symmetric surface distance = 1.7282 mm and maximum symmetric surface distance = 38.039 mm. The quantitative and qualitative results obtained in our preliminary work show that the K-Means clustering technique along with power law transformation is effective in producing good liver segmentation outputs. As future work, we will attempt to automate the power law transformation technique.
Research
Medical Image Segmentation is an activity with huge handiness. Biomedical and anatomical data are made simple to acquire because of progress accomplished in computerizing picture division. More research and work on it has improved more viability to the extent the subject is concerned. A few techniques are utilized for therapeutic picture division, for example, Clustering strategies, Thresholding technique, Classifier, Region Growing, Deformable Model, Markov Random Model and so forth. This work has for the most part centered consideration around Clustering techniques, particularly k-implies what's more, fluffy c-implies grouping calculations. These calculations were joined together to concoct another technique called fluffy k-c-implies bunching calculation, which has a superior outcome as far as time usage. The calculations have been actualized and tried with Magnetic Resonance Image (MRI) pictures of Human cerebrum. The proposed strategy has expanded effectiveness and lessened emphasis when contrasted with different techniques. The nature of picture is assessed by figuring the proficiency as far as number of rounds and the time which the picture takes to make one emphasis. Results have been dissected and recorded. Some different strategies were surveyed and favorable circumstances and hindrances have been expressed as special to each. Terms which need to do with picture division have been characterized nearby with other grouping strategies.
Conference Paper
Accurate medical image segmentation can assist doctors in disease diagnosis. It is very important to segment the liver accurately from medical images in the field of the liver. However, the low contrast of tissues and organs and uneven distribution of CT values in abdominal CT images makes liver segmentation difficult. In this paper, we propose a method of combining the improved U-Net network model and the region growing algorithm, the feature information of the pooling layer is directly extracted after two convolution and ReLU functions. The up-sample layer copies the feature information of that is the corresponding down-sampling layer. Softmax Layer calculates the amount of information loss to reduce the loss of feature information. Finally, the region growing algorithm is used to optimize the initial results. Five parameters of medical image segmentation are used to evaluate, DICE can reach more than 95.0%, and other parameters have been increased accordingly. Experimental results show this method can accurately segment the liver area, solve the problems of blurred edges and unclear focus areas, and provide an effective basis for the diagnosis of liver disease
Article
Objective : Herein, a neural network-based liver segmentation algorithm is proposed, and its performance was evaluated using abdominal computed tomography (CT) images. Methods : A fully convolutional network was developed to overcome the volumetric image segmentation problem. To guide a neural network to accurately delineate a target liver object, the network was deeply supervised by applying the adaptive self-supervision scheme to derive the essential contour, which acted as a complement with the global shape. The discriminative contour, shape, and deep features were internally merged for the segmentation results. Results and Conclusion : 160 abdominal CT images were used for training and validation. The quantitative evaluation of the proposed network was performed through an eight-fold cross-validation. The result showed that the method, which uses the contour feature, segmented the liver more accurately than the state-of-the-art with a 2.13% improvement in the dice score. Significance : In this study, a new framework was introduced to guide a neural network and learn complementary contour features. The proposed neural network demonstrates that the guided contour features can significantly improve the performance of the segmentation task.
Article
Full-text available
Medical Image Segmentation is an activity with huge handiness. Biomedical and anatomical data are made simple to acquire because of progress accomplished in computerizing picture division. More research and work on it has improved more viability to the extent the subject is concerned. A few techniques are utilized for therapeutic picture division, for example, Clustering strategies, Thresholding technique, Classifier, Region Growing, Deformable Model, Markov Random Model and so forth. This work has for the most part centered consideration around Clustering techniques, particularly k-implies what's more, fluffy c-implies grouping calculations. These calculations were joined together to concoct another technique called fluffy k-c-implies bunching calculation, which has a superior outcome as far as time usage. The calculations have been actualized and tried with Magnetic Resonance Image (MRI) pictures of Human cerebrum. The proposed strategy has expanded effectiveness and lessened emphasis when contrasted with different techniques. The nature of picture is assessed by figuring the proficiency as far as number of rounds and the time which the picture takes to make one emphasis. Results have been dissected and recorded. Some different strategies were surveyed and favorable circumstances and hindrances have been expressed as special to each. Terms which need to do with picture division have been characterized nearby with other grouping strategies.
Article
Full-text available
Liver segmentation is one of the most basic and important parts in computer-aided diagnosis for liver CT. Although various segmentation methods have been proposed for medical imaging, most of them generally do not perform well in segmenting the liver from CT images because of surface features of the liver and difficulty of discrimination from other adjacent organs. In this paper, we propose a new scheme for automatic segmentation of the liver in CT images. The pro-posed scheme is carried out on region-of-interest (ROI) blocks that include regions of the liver with high probabilities. The ROI approach saves unnecessary computational loss in finding the accurate boundary of the liver. The proposed method utilizes the composition of morphological filters with a priori knowledge, such as the general location or the approximate intensity of the liver to detect the initial boundary of the liver. Then, we make the gradient image with the weight of an initial liver boundary and segment the liver region by using an immersion-based watershed algorithm in the gradient image. Finally, a refining process is carried out to acquire a more accurate liver region.
Article
Full-text available
We present a new, interactive tool calledIntelligent Scissorswhich we use for image segmentation. Fully automated segmentation is an unsolved problem, while manual tracing is inaccurate and laboriously unacceptable. However, Intelligent Scissors allow objects within digital images to be extracted quickly and accurately using simple gesture motions with a mouse. When the gestured mouse position comes in proximity to an object edge, alive-wire boundary“snaps” to, and wraps around the object of interest. Live-wire boundary detection formulates boundary detection as an optimal path search in a weighted graph. Optimal graph searching provides mathematically piece-wise optimal boundaries while greatly reducing sensitivity to local noise or other intervening structures. Robustness is further enhanced withon-the-fly trainingwhich causes the boundary to adhere to the specific type of edge currently being followed, rather than simply the strongest edge in the neighborhood.Boundary coolingautomatically freezes unchanging segments and automates input of additional seed points. Cooling also allows the user to be much more free with the gesture path, thereby increasing the efficiency and finesse with which boundaries can be extracted.
Article
Full-text available
Computed tomography (CT) images have been widely used for liver disease diagnosis. Designing and developing computer-assisted image processing techniques to help doctors improve their diagnosis has received considerable interests over the past years. In this paper, a CT liver image diagnostic classification system is presented which will automatically find, extract the CT liver boundary and further classify liver diseases. The system comprises a detect-before-extract (DBE) system which automatically finds the liver boundary and a neural network liver classifier which uses specially designed feature descriptors to distinguish normal liver, two types of liver tumors, hepatoma and hemageoma. The DBE system applies the concept of the normalized fractional Brownian motion model to find an initial liver boundary and then uses a deformable contour model to precisely delineate the liver boundary. The neural network is included to classify liver tumors into hepatoma and hemageoma. It is implemented by a modified probabilistic neural network (PNN) [MPNN] in conjunction with feature descriptors which are generated by fractal feature information and the gray-level co-occurrence matrix. The proposed system was evaluated by 30 liver cases and shown to be efficient and very effective.
Article
Full-text available
A common challenge for automated segmentation techniques is differentiation between images of close objects that have similar intensities, whose boundaries are often blurred due to partial-volume effects. We propose a novel approach to segmentation of two-dimensional images, which addresses this challenge. Our method, which we call intrinsic shape for segmentation (ISeg), analyzes isolabel-contour maps to identify coherent regions that correspond to major objects. ISeg generates an isolabel-contour map for an image by multilevel thresholding with a fine partition of the intensity range. ISeg detects object boundaries by comparing the shape of neighboring isolabel contours from the map. ISeg requires only little effort from users; it does not require construction of shape models of target objects. In a formal validation with computed-tomography angiography data, we showed that ISeg was more robust than conventional thresholding, and that ISeg's results were comparable to results of manual tracing.
Article
Full-text available
The new MPEG-4 video coding standard enables content-based functionalities. In order to support the philosophy of the MPEG-4 visual standard, each frame of video sequences should be represented in terms of video object planes (VOPs). In other words, video objects to be encoded in still pictures or video sequences should be prepared before the encoding process starts. Therefore, it requires a prior decomposition of sequences into VOPs so that each VOP represents a moving object. This paper addresses an image segmentation method for separating moving objects from the background in image sequences. The proposed method utilizes the following spatio-temporal information. (1) For localization of moving objects in the image sequence, two consecutive image frames in the temporal direction are examined and a hypothesis testing is performed by comparing two variance estimates from two consecutive difference images, which results in an F-test. (2) Spatial segmentation is performed to divide each image into semantic regions and to find precise object boundaries of the moving objects. The temporal segmentation yields a change detection mask that indicates moving areas (foreground) and nonmoving areas (background), and spatial segmentation produces spatial segmentation masks. A combination of the spatial and temporal segmentation masks produces VOPs faithfully. This paper presents various experimental results
Article
Full-text available
The results of a study on multiscale shape description, smoothing and representation are reported. Multiscale nonlinear smoothing filters are first developed, using morphological opening and closings. G. Matheron (1975) used openings and closings to obtain probabilistic size distributions of Euclidean-space sets (continuous binary images). These distributions are used to develop a concept of pattern spectrum (a shape-size descriptor). A pattern spectrum is introduced for continuous graytone images and arbitrary multilevel signals, as well as for discrete images, by developing a discrete-size family of patterns. Large jumps in the pattern spectrum at a certain scale indicate the existence of major (protruding or intruding) substructures of the signal at the scale. An entropy-like shape-size complexity measure is also developed based on the pattern spectrum. For shape representation, a reduced morphological skeleton transform is introduced for discrete binary and graytone images. This transform is a sequence of skeleton components (sparse images) which represent the original shape at various scales. It is shown that the partially reconstructed images from the inverse transform on subsequences of skeleton components are the openings of the image at a scale determined by the number of eliminated components; in addition, two-way correspondences are established among the degree of shape smoothing via multiscale openings or closings, the pattern spectrum zero values, and the elimination or nonexistence of skeleton components at certain scales
Article
This paper presents an efficient volume measurement method for quantitative assessments within a virtual liver surgery planning system. The measurement is carried out by hardware-accelerated voxelization of surface-based models in order to be applicable for interactive applications. The focus of this paper is the presentation of our voxelization technique suitable for off-the-shelf PC graphics cards including two different data retrieval approaches. Moreover, the measurement can be performed off-screen while not disturbing the rest of the application. Another important issue discussed in this paper is the applicability study within the planning system. Our measurement toolkit has been applied on several medical objects and delivers interactive response times.
Article
In this paper, the authors have proposed a method of segmenting gray level images using multiscale morphology. The approach resembles the watershed algorithm in the sense that the dark (respectively bright) features which are basically canyons (respectively mountains) on the surface topography of the gray level image are gradually filled (respectively clipped) using multiscale morphological closing (respectively opening) by reconstruction with isotropic structuring element. The algorithm detects valid segments at each scale using three criteria namely growing, merging and saturation. Segments extracted at various scales are integrated in the final result. The algorithm is composed of two passes preceded by a preprocessing step for simplifying small scale details of the image that might cause over-segmentation. In the first pass feature images at various scales are extracted and kept in respective level of morphological towers. In the second pass, potential features contributing to the formation of segments at various scales are detected. Finally the algorithm traces the contours of all such contributing features at various scales. The scheme after its implementation is executed on a set of test images (synthetic as well as real) and the results are compared with those of few other standard methods. A quantitative measure of performance is also formulated for comparing the methods.
Computer Based Automatic Segmentation and Volume Measurement of the Liver and Spleen in Contrast-Enhanced Helical CT Images, A Thesis for the degree of
  • H Y Kim
H.Y. Kim, Computer Based Automatic Segmentation and Volume Measurement of the Liver and Spleen in Contrast-Enhanced Helical CT Images, A Thesis for the degree of Doctor of Philosophy of Medicine in Chungnam National University, 2002.
Efficient volume measurement using voxelization
  • B Reitinger
  • A Bornik
  • R Beichel
B. Reitinger, A. Bornik, R. Beichel, Efficient volume measurement using voxelization, in: Proceedings of Spring Conference on Computer Graphics 2003, 2003.
  • S.-J Lim
S.-J. Lim et al. / J. Vis. Commun. Image R. 17 (2006) 860–875
  • E Gose
  • R Johnsonbaugh
  • S Jost
E. Gose, R. Johnsonbaugh, S. Jost, Pattern Recognition and Image Analysis, Prentice-Hall, Englewood Cliffs, NJ, 1996.