-
[show abstract]
[hide abstract]
ABSTRACT: Objectives The purpose of this study was to compare automated and semiautomated algorithms for analysis of carotid artery wall thickness and intima-media thickness on multidetector row computed tomographic (CT) angiography and sonography, respectively, and to study the correlation between them. Methods Twenty consecutive patients underwent multidetector row CT angiographic and sonographic analysis of carotid arteries (mean age, 66 years; age range, 59-79 years). The intima-media thickness of the 40 carotid arteries was measured with novel and dedicated automated software analysis and by 4 observers who manually calculated the intima-media thickness. The carotid artery wall thickness was automatically estimated by using a specific algorithm and was also semiautomatically quantified. The correlation between groups was calculated by using the Pearson ρ statistic, and scatterplots were calculated. We evaluated intermethod agreement using Bland-Altman analysis. Results By comparing automated carotid artery wall thickness, automated intima-media thickness, semiautomated carotid artery wall thickness, and semiautomated intima-media thickness analyses, a statistically significant association was found, with the highest values obtained for the association between semiautomated and automated intima-media thickness analyses(Pearson ρ = 0.9; 95% confidence interval, 0.82-0.95; P = 0.0001). The lowest values were obtained for the association between semiautomated intima-media thickness and automated carotid artery wall thickness analyses (Pearson ρ = 0.44; 95% confidence interval, 0.15-0.66; P = 0.0047). In the Bland-Altman analysis, the better results were obtained by comparing the semiautomated and automated algorithms for the study of intima-media thickness, with an interval of -16.1% to +43.6%. Conclusions The results of this preliminary study showed that carotid artery wall thickness and intima-media thickness can be studied with automated software, although the CT analysis needs to be further improved.
Journal of ultrasound in medicine: official journal of the American Institute of Ultrasound in Medicine 04/2013; 32(4):665-74. · 1.25 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Ultrasonography has great potential in differentiating malignant thyroid nodules from the benign ones. However, visual interpretation is limited by interobserver variability, and further, the speckle distribution poses a challenge during the classification process. This article thus presents an automated system for tumor classification in three-dimensional contrast-enhanced ultrasonography data sets. The system first processes the contrast-enhanced ultrasonography images using complex wavelet transform-based filter to mitigate the effect of speckle noise. The higher order spectra features are then extracted and used as input for training and testing a fuzzy classifier. In the off-line training system, higher order spectra features are extracted from a set of images known as the training images. These higher order spectra features along with the clinically assigned ground truth are used to train the classifier and obtain an estimate of the classifier or training parameters. The ground truth tells the class label of the image (i.e. whether the image belongs to a benign or malignant nodule). During the online testing phase, the estimated classifier parameters are applied on the higher order spectra features that are extracted from the testing images to predict their class labels. The predicted class labels are compared with their corresponding original ground truth to evaluate the performance of the classifier. Without utilizing the complex wavelet transform filter, the fuzzy classifier demonstrated an accuracy of 91.6%, while utilizing the complex wavelet transform filter, the accuracy significantly boosted to 99.1%.
Proceedings of the Institution of Mechanical Engineers Part H Journal of Engineering in Medicine 03/2013; 227(3):284-92. · 1.21 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Among gynecological malignancies, ovarian cancer is the most frequent cause of death. Image mining algorithms have been predominantly used to give the physicians a more objective, fast, and accurate second opinion on the initial diagnosis made from medical images. The objective of this work is to develop an adjunct computer-aided diagnostic technique that uses 3D ultrasound images of the ovary to accurately characterize and classify benign and malignant ovarian tumors. In this algorithm, we first extract features based on the textural changes and higher-order spectra information. The significant features are then selected and used to train and evaluate the decision tree (DT) classifier. The proposed technique was validated using 1,000 benign and 1,000 malignant images, obtained from ten patients with benign and ten with malignant disease, respectively. On evaluating the classifier with tenfold stratified cross validation, the DT classifier presented a high accuracy of 97 %, sensitivity of 94.3 %, and specificity of 99.7 %. This high accuracy was achieved because of the use of the novel combination of the four features which adequately quantify the subtle changes and the nonlinearities in the pixel intensity variations. The rules output by the DT classifier are comprehensible to the end-user and, hence, allow the physicians to more confidently accept the results. The preliminary results show that the features are discriminative enough to yield good accuracy. Moreover, the proposed technique is completely automated, accurate, and can be easily written as a software application for use in any computer.
Journal of Digital Imaging 11/2012; · 1.25 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we present a Computer Aided Diagnosis (CAD) based technique (Atheromatic system) for classification of carotid plaques in B-mode ultrasound images into symptomatic or asymptomatic classes. This system, called Atheromatic, has two steps: (i) feature extraction using a combination of Discrete Wavelet Transform (DWT) and averaging algorithms and (ii) classification using Support Vector Machine (SVM) classifier for automated decision making. The CAD system was built and tested using a database consisting of 150 asymptomatic and 196 symptomatic plaque regions of interests which were manually segmented. The ground truth of each plaque was determined based on the presence or absence of symptoms. Three-fold cross-validation protocol was adapted for developing and testing the classifiers. The SVM classifier with a polynomial kernel of order 2 recorded the highest classification accuracy of 83.7%. In the clinical scenario, such a technique, after much more validation, can be used as an adjunct tool to aid physicians by giving a second opinion on the nature of the plaque (symptomatic/asymptomatic) which would help in the more confident determination of the subsequent treatment regime for the patient.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:3199-202.
-
[show abstract]
[hide abstract]
ABSTRACT: In this work, we present a Computer Aided Diagnosis (CAD) based technique for automatic classification of benign and malignant thyroid lesions in 3D contrast-enhanced ultrasound images. The images were obtained from 20 patients. Fine needle aspiration biopsy and histology confirmed malignancy. Discrete Wavelet Transform (DWT) and texture based features were extracted from the thyroid images. The resulting feature vectors were used to train and test three different classifiers: K-Nearest Neighbor (K-NN), Probabilistic Neural Network (PNN), and Decision Tree (DeTr) using ten-fold cross validation technique. Our results show that combination of DWT and texture features in the K-NN classifier resulted in a classification accuracy of 98.9%, a sensitivity of 98%, and a specificity of 99.8%. Thus, the preliminary results of the proposed technique show that it could be adapted as an adjunct tool that can give valuable second opinions to the doctors regarding the nature of the thyroid nodule. The technique is cost-effective, non-invasive, fast, completely automated and gives more objective and reproducible results compared to manual analysis of the ultrasound images. We however intend to establish the clinical applicability of this technique by evaluating it with more data in the future.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:452-5.
-
[show abstract]
[hide abstract]
ABSTRACT: We present here a novel and patented completely automated IMT measurement system that we developed for common carotid arterial ultrasound longitudinal images, called Carotid Measurement Using Dual Snakes (CMUDS) - a class of AtheroEdge™ system. CMUDS is a dual deformable parametric model (snake) system where the dual snakes evolve simultaneously and are forced to maintain a regularized distance to prevent collapsing or diverging. We benchmarked CMUDS against a conventional single snake (CMUSS). CMUDS is totally automatic while CMUSS is semi-automatic. For performance evaluation, two readers manually traced the lumen-intima (LI) and media-adventitia (MA) borders of our multi-institutional, multi-ethnic, and multi-scanner database of 655 longitudinal B-Mode ultrasound images. CMUDS and CMUSS correctly processed all 665 images. The average IMT biases were equal to 0.030±0.284 mm and -0.004±0.273 mm for CMUDS, and -0.011±0.329 mm and -0.045±0.317 mm for CMUSS. The Figure of Merit of the system was 96.0% and 99.6% for CMUDS and 98.5% and 94.4% for CMUSS. CMUDS improved accuracy (Wilcoxon, p<0.02) and reproducibility (Fisher, p<3 10(-2)), proving that the novel CMUDS system is adaptable to large multi-centric studies, where a standard IMT measurement technique is required.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:484-7.
-
[show abstract]
[hide abstract]
ABSTRACT: In this work, we present a Computer Aided Diagnostic (CAD) technique (a class of Atheromatic systems) that classifies the automatically segmented carotid far wall Intima-Media Thickness (IMT) regions along the common carotid artery into symptomatic and asymptomatic classes. We extracted texture features based on Local Binary Patterns (LBP) and Law's Texture Energy (LTE) and used the significant features to train and test the Support Vector Machine classifier. We developed the classifiers using three-fold stratified cross validation data resampling technique on 342 IMT wall regions. An accuracy of 89.5% was registered. Thus, the proposed technique is accurate, robust, non-invasive, fast, objective, and cost-effective, and hence, will add more value to the existing carotid plaque diagnostics protocol.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:448-51.
-
[show abstract]
[hide abstract]
ABSTRACT: In this work, we have developed an adjunct Computer Aided Diagnostic (CAD) technique that uses 3D acquired ultrasound images of the ovary and data mining algorithms to accurately characterize and classify benign and malignant ovarian tumors. In this technique, we extracted image-texture based and Higher Order Spectra (HOS) based features from the images. The significant features were then selected and used to train and test the Decision Tree (DT) classifier. The proposed technique was validated using 1000 benign and 1000 malignant images, obtained from 10 patients with benign and 10 with malignant disease, respectively. On evaluating the classifier with 10-fold stratified cross validation, we observed that the DT classifier presented a high accuracy of 95.1%, sensitivity of 92.5% and specificity of 97.7%. Thus, the four significant features could adequately quantify the subtle changes and nonlinearities in the pixel intensities. The preliminary results presented in this paper indicate that the proposed technique can be reliably used as an adjunct tool for ovarian tumor classification since the system is accurate, completely automated, cost-effective, and can be easily written as a software application for use in any computer.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:4446-9.
-
[show abstract]
[hide abstract]
ABSTRACT: The carotid intima-media thickness (IMT) is a validated marker of cerebrovascular disease risk. This work presents a new parameter, the IMT variability (IMTV), and compares the IMT and IMTV in symptomatic and asymptomatic Italian patients. 142 patients were analyzed (age 59±11.2 years, 59% males), 42 of which suffered from TIA (transient ischemic attack) or minor stroke. The lumen-intima (LI) and media-adventitia (MA) interfaces were manually traced by a Reader, and automatically traced by an automated system (AutoEdge). These interfaces were then used to measure the IMT and IMTV along the carotid wall. Wilcoxon and Pearson correlation analyses were performed. There was about a 65% correlation between the manual and automated measurements of IMT. There was no statistical difference between the manual and automated IMTV measurements (Wilcoxon signed rank, p>0.7). The observed mean IMT for symptomatic patients (0.83±0.44 mm for Reader vs. 0.82±0.35 mm for AutoEdge) was higher compared to asymptomatic patients (0.78±0.45 mm for Reader vs. 0.74±0.30 mm for AutoEdge). The symptomatic IMTV was about 11% higher than the asymptomatic IMTV when using Reader tracings and 8% higher when using AutoEdge. AutoEdge was very accurate in measuring the IMT and IMTV both for symptomatic and asymptomatic patients. Results showed that the symptomatic subjects had comparable IMT with respect to asymptomatic subjects, but a higher IMTV value.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:2668-71.
-
[show abstract]
[hide abstract]
ABSTRACT: The carotid intima-media thickness (IMT) is a validated marker of cerebrovascular disease risk. This paper presents a new parameter, the IMT variability (IMTV), and compares it between symptomatic and asymptomatic patients taken from a cohort of Italian population. One hundred forty-two patients were analyzed (age 59 ± 112 years, 59% males), 42 of these patients suffered from TIA or minor stroke. The lumen-intima (LI) and media-adventitia (MA) interfaces of the far wall were manually traced by a Reader. We also used a computer-based automated system (called AutoEdge) to obtain the LI/MA interfaces. The LI/MA interfaces were used to measure the IMT and the IMTV along the distal wall of the common carotid artery. Wilcoxon and Pearson correlation analyses were performed. The agreement between the Reader's IMT and the AutoEdge IMT values was 98.7% for the symptomatic (0.83 ± 0.44 mm for Reader, 0.82 ± 0.35 mm for AutoEdge) and 94.9% for the asymptomatic patients (0.78 ± 0.45 mm for Reader, 0.74 ± 0.30 mm for AutoEdge). Correlation was 65% for symptomatic and 68% for asymptomatic patients, respectively. The IMT measured using AutoEdge was 1.2% lower compared to manual measurements in symptomatic population, while 5.12% lower in asymptomatic. The IMTV was 11% higher in symptomatic patients compared to asymptomatic when using manual delineations, 8% higher when using AutoEdge. There was no statistical difference between the manual and automated IMTV measurements (Wilcoxon signed rank, P > 0.7). We conclude that the IMT and IMTV values were very similar between Reader and AutoEdge software when studying symptomatic and asymptomatic patients in Italian population.
Echocardiography 07/2012; 29(9):1111-1119. · 1.24 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Automated computer-aided detection systems for measurement of the carotid intima-media thickness (IMT) are becoming popular. These systems yield lumen-intima (LI) and media-adventitia (MA) borders. In this work, we developed and validated a novel and patented completely automated IMT measurement system called carotid measurement using dual snakes (CMUDS): a class of AtheroEdge system (Global Biomedical Technologies, Inc, Roseville, CA). CMUDS is modeled as a dual parametric system corresponding to LI and MA borders with initialization from the far adventitia layer. The novelty of CMUDS is the first-order absolute moment-based external energy, which provides stable deformation. The dual snakes evolve simultaneously and are forced to maintain a regularized distance to prevent collapsing or bleeding. Two independent readers manually traced the LI/MA boundaries of a multi-institutional, multi-ethnic, and multi-scanner database of 665 longitudinal images for performance evaluation. CMUDS was also benchmarked against a previously developed automated technique. CMUDS correctly processed 660 images (99.2% success). The differences between the CMUDS and two manual IMT measurements (mean ± SD) were 0.013 ± 0.216 and -0.021 ± 0.197 mm, respectively. The corresponding figures of merit for CMUDS compared to reader tracings were 98.4% and 97.5%. Compared to the previous technique (IMT differences, 0.022 ± 0.276 and -0.012 ± 0.266 mm), CMUDS improved accuracy (Wilcoxon P < 0.009) and variability (Fisher P < 10(-8)). Among different resolution images from original equipment manufacturer ultrasound scanners, CMUDS performed best with high-resolution images corresponding to 0.0789 mm/pixel. Accuracy in IMT measurement with the proposed automated CMUDS technique makes this system adaptable to large multi-center studies, in which such an IMT measurement system would be very useful tool.
Journal of ultrasound in medicine: official journal of the American Institute of Ultrasound in Medicine 07/2012; 31(7):1123-36. · 1.25 Impact Factor
-
Filippo Molinari,
Kristen M Meiburger,
Luca Saba,
U Rajendra Acharya,
Giuseppe Ledda,
Guang Zeng,
Sin Yee Stella Ho,
Anil T Ahuja,
Suzanne C Ho,
Andrew Nicolaides,
Jasjit S Suri
[show abstract]
[hide abstract]
ABSTRACT: Automated and high performance carotid intima-media thickness (IMT) measurement is gaining increasing importance in clinical practice to assess the cardiovascular risk of patients. In this paper, we compare four fully automated IMT measurement techniques (CALEX, CAMES, CARES and CAUDLES) and one semi-automated technique (FOAM). We present our experience using these algorithms, whose lumen-intima and media-adventitia border estimation use different methods that can be: (a) edge-based; (b) training-based; (c) feature-based; or (d) directional Edge-Flow based. Our database (DB) consisted of 665 images that represented a multi-ethnic group and was acquired using four OEM scanners. The performance evaluation protocol adopted error measures, reproducibility measures, and Figure of Merit (FoM). FOAM showed the best performance, with an IMT bias equal to 0.025±0.225mm, and a FoM equal to 96.6%. Among the four automated methods, CARES showed the best results with a bias of 0.032±0.279mm, and a FoM to 95.6%, which was statistically comparable to that of FOAM performance in terms of accuracy and reproducibility. This is the first time that completely automated and user-driven techniques have been compared on a multi-ethnic dataset, acquired using multiple original equipment manufacturer (OEM) machines with different gain settings, representing normal and pathologic cases.
Computer methods and programs in biomedicine 05/2012; · 1.14 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Subjects suffering from migraine with aura (MwA) present an altered cerebral autoregulation during migraine attacks. It is
still unclear whether MwA sufferers present a normal autoregulation during attack-free periods. In this study, we characterized
cerebral autoregulation in the frequency domain by analyzing the spontaneous oscillations superimposed on the cerebral hemodynamic
signals, as detected by near-infrared spectroscopy (NIRS). Ten healthy women (age: 38.4±9.5years) and ten women suffering
from MwA (age: 35.2±10.5years) underwent NIRS recording in resting conditions and during breath-holding (BH). Being the
NIRS signals during BH nonstationary, we used the Choi–Williams time–frequency distribution to perform spectral analysis.
We considered 256s of signals and quantified the variation in the power of the very-low frequencies (VLF: 20–40mHz) and
of the low frequencies (LF: 40–140mHz) as response to BH. Results showed that BH increases the power in the LF band both
in healthy and MwA subjects. Considering the signal of the deoxygenated hemoglobin, the average power increase in the LF band
was equal to 20%±15.4% for the healthy group and significantly lower, 4.8%±8.3%, in the MwA group (Student’s t test, P<0.02). No significant difference was observed in the VLF band or in the oxygenated hemoglobin signal power variations of
the LF and VLF bands. The resulting data reveal a possible impairment in the carbon dioxide-regulatory mechanism in MwA subjects.
Neurological Sciences 04/2012; 30:105-107. · 1.32 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Plaques in the carotid artery result in stenosis, which is one of the main causes for stroke. Patients have to be carefully selected for stenosis treatments as they carry some risk. Since patients with symptomatic plaques have greater risk for strokes, an objective classification technique that classifies the plaques into symptomatic and asymptomatic classes is needed. We present a computer aided diagnostic (CAD) based ultrasound characterization methodology (a class of Atheromatic systems) that classifies the patient into symptomatic and asymptomatic classes using two kinds of datasets: (1) plaque regions in ultrasound carotids segmented semi-automatically and (2) far wall gray-scale intima-media thickness (IMT) regions along the common carotid artery segmented automatically. For both kinds of datasets, the protocol consists of estimating texture-based features in frameworks of local binary patterns (LBP) and Law's texture energy (LTE) and applying these features for obtaining the training parameters, which are then used for classification. Our database consists of 150 asymptomatic and 196 symptomatic plaque regions and 342 IMT wall regions. When using the Atheromatic-based system on semiautomatically determined plaque regions, support vector machine (SVM) classifier was adapted with highest accuracy of 83%. The accuracy registered was 89.5% on the far wall gray-scale IMT regions when using SVM, K-nearest neighbor (KNN) or radial basis probabilistic neural network (RBPNN) classifiers. LBP/LTE-based techniques on both kinds of carotid datasets are noninvasive, fast, objective and cost-effective for plaque characterization and, hence, will add more value to the existing carotid plaque diagnostics protocol. We have also proposed an index for each type of datasets: AtheromaticPi, for carotid plaque region, and AtheromaticWi, for IMT carotid wall region, based on the combination of the respective significant features. These indices show a separation between symptomatic and asymptomatic by 4.53 units and 4.42 units, respectively, thereby supporting the texture hypothesis classification.
Ultrasound in medicine & biology 04/2012; 38(6):899-915. · 2.02 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Accurate intima-media thickness (IMT) measurement of the carotid artery from minimal plaque ultrasound images is a relevant clinical need, since IMT increase is related to the progression of atherosclerosis. In this paper, we describe a novel dual snake-based model for the high-performance carotid IMT measurement, called Carotid Measurement Using Dual Snakes (CMUDS). Snakes (which are deformable contours) adapt to the lumen-intima (LI) and media-adventitia (MA) interfaces, thus enabling the IMT computation as distance between the LI and MA snakes. However, traditional snakes might be unable to maintain a correct distance and in some spatial location along the artery, it might even collapse between them or diverge. The technical improvement of this work is the definition of a dual snake-based constrained system, which prevents the LI and MA snakes from collapsing or bleeding, thus optimizing the IMT estimation. The CMUDS system consists of two parametric models automatically initialized using the far adventitia border which we automatically traced by using a previously developed multi-resolution approach. The dual snakes evolve simultaneously and are constrained by the distances between them, ensuring the regularization of LI/MA topology. We benchmarked our automated CMUDS with the previous conventional semi-automated snake system called Carotid Measurement Using Single Snake (CMUSS). Two independent readers manually traced the LIMA boundaries of a multi-institutional, multi-ethnic, and multi-scanner database of 665 CCA longitudinal 2D images. We evaluated our system performance by comparing it with the gold standard as traced by clinical readers. CMUDS and CMUSS correctly processed 100% of the 665 images. Comparing the performance with respect to the two readers, our automatically measured IMT was on average very close to that of the two readers (IMT measurement biases for CMUSS was equal to -0.011±0.329mm and -0.045±0.317mm, respectively, while for CMUDS, it was 0.030±0.284mm and -0.004±0.273mm, respectively). The Figure-of-Merit of the system was 98.5% and 94.4% for CMUSS, while 96.0% and 99.6% for CMUDS, respectively. Results showed that the dual-snake system CMUDS reduced the IMT measurement error accuracy (Wilcoxon, p<0.02) and the IMT error variability (Fisher, p<3×10(-2)). We propose the CMUDS technique for use in large multi-centric studies, where the need for a standard, accurate, and automated IMT measurement technique is required.
Ultrasonics 03/2012; 52(7):949-61. · 1.84 Impact Factor
-
IEEE Transactions on Image Processing. 01/2012; 21:1211-1222.
-
[show abstract]
[hide abstract]
ABSTRACT: The development of completely automated techniques for arterial wall segmentation and intima-media thickness measurement requires the recognition of the artery in the image frame. Conceptually, automated techniques can be thought of as the combination of two cascaded stages: artery recognition and wall segmentation. In this paper, the authors show three carotid artery recognition systems (CARS) that are fully automated.
The first technique is based on a first-order derivative Gaussian edge analysis (CARSgd). The second method is based on an integrated approach (CARSia) that combines image feature extraction, fitting, and classification. The third strategy is based on signal analysis (CARSsa). The output of all the three paradigms provide tracing of the far adventitial (AD(F)). The authors validated CARSgd, CARSia, and CARSsa on a dataset of 365 longitudinal B-Mode carotid images, acquired by different sonographers. Performance evaluation of the carotid recognition process was done in three ways: (1) visual inspection by experts; (2) by measuring the Hausdorff distance (HD) between the automatic far adventitial (AD(F)) and the manually traced AD(F), and (3) by measuring the HD between AD(F) and the lumen-intima (GT(LI)) and media-adventitia (GT(MA)) borders of the arterial walls.
The average HD between AD(F) and the manual AD(F) was 1.53 ± 1.51 mm for CARSgd, 1.82 ± 3.08 mm for CARSia, and 2.56 ± 2.89 mm for CARSsa. The average HD between GT(LI) and AD(F) for CARSgd, CARSia, and CARSsa were 2.16 ± 1.16 mm, 2.71 ± 2.89 mm, and 2.66 ± 1.52 mm, respectively. The average HD between AD(F) and GT(MA) for CARSgd, CARSia, and CARSsa were 1.54 ± 1.19 mm, 1.86 ± 2.66 mm, and 1.95 ± 1.64 mm, respectively. Considering a maximum distance of 50 pixels (about 3 mm), CARSgd showed an identification accuracy of 100%, CARSia of 92%, and CARSsa of 96%. These identification accuracies were confirmed by visual inspection. All the three systems work on MATLAB, Windows OS, and on a PC based cross platform medical application written in Java called ATHEROEDGE™ with 1 s per image.
CARSgd showed very accurate AD(F) profiles coupled with a low computational burden and without the need for specific tuning. It can be thought of as a reference technique for carotid localization, to be used in automated intima-media thickness measurement strategies.
Medical Physics 01/2012; 39(1):378-91. · 2.83 Impact Factor
-
Journal of Mechanics in Medicine and Biology 01/2012; 12(4):1240013-1-1240013-14. · 0.47 Impact Factor
-
Atti Congresso Nazionale di Bioingegneria 2012; 01/2012
-
[show abstract]
[hide abstract]
ABSTRACT: Ultrasound-based thyroid nodule characterization into benign and malignant types is limited by subjective interpretations. This paper presents a Computer Aided Diagnostic (CAD) technique that would present more objective and accurate classification and further would offer the physician a valuable second opinion. In this paradigm, we first extracted the features that quantify the local changes in the texture characteristics of the ultrasound off-line training images from both benign and malignant nodules. These features include: Fractal Dimension (FD), Local Binary Pattern (LBP), Fourier Spectrum Descriptor (FS), and Laws Texture Energy (LTE). The resulting feature vectors were used to build seven different classifiers: Support Vector Machine (SVM), Decision Tree (DT), Sugeno Fuzzy, Gaussian Mixture Model (GMM), K-Nearest Neighbor (KNN), Radial Basis Probabilistic Neural Network (RBPNN), and Naive Bayes Classifier (NBC). Subsequently, the feature vector-classifier combination that results in the maximum classification accuracy was used to predict the class of a new on-line test thyroid ultrasound image. Two data sets with 3D Contrast-Enhanced Ultrasound (CEUS) and 3D High Resolution Ultrasound (HRUS) images of 20 nodules (10 benign and 10 malignant) were used. Fine needle aspiration biopsy and histology results were used to confirm malignancy. Our results show that a combination of texture features coupled with SVM or Fuzzy classifiers resulted in 100% accuracy for the HRUS dataset, while GMM classifier resulted in 98.1% accuracy for the CEUS dataset. Finally, for each dataset, we have proposed a novel integrated index called Thyroid Malignancy Index (TMI) using the combination of FD, LBP, LTE texture features, to diagnose benign or malignant nodules. This index can help clinicians to make a more objective differentiation of benign/malignant thyroid lesions. We have compared and benchmarked the system with existing methods.
Ultrasonics 11/2011; 52(4):508-20. · 1.84 Impact Factor
