Fast Automatic Multi-atlas Segmentation of the Prostate from 3D MR Images.
ABSTRACT A fast fully automatic method of segmenting the prostate from 3D MR scans is presented, incorporating dynamic multi-atlas label fusion. The diffeomorphic demons method is used for non-rigid registration and a comparison of alternate metrics for atlas selection is presented. A comparison of results from an average shape atlas and the multi-atlas approach is provided. Using the same clinical dataset and manual contours from 50 clinical scans as Klein et al. (2008) a median Dice similarity coefficient of 0.86 was achieved with an average surface error of 2.00mm using the multi-atlas segmentation method.
SourceAvailable from: Fahmi Khalifa
Dataset: PhD Dissertation
[Show abstract] [Hide abstract]
ABSTRACT: Prostate cancer is the second most common cause of cancer deaths in males. Accurate prostate segmentation from magnetic resonance (MR) images is critical to the diagnosis and treatment of prostate cancer. Automated prostate segmentation is challenging due to the variety in shapes and sizes of the prostate. Furthermore, the expected boundaries of ROIs are often indistinct, while heterogeneity concurrently exists within the ROIs. To address these challenges, we propose an automated approach that incorporates the local intensity features by random walker (RW) algorithm and global probability knowledge from an atlas to better describe unique characteristics of the prostate in MR images. We formulated a new RW weight function to take into account atlas probabilities and intensity differences. The prior knowledge from the atlas probability map not only reflects the statistical shape approximation of the prostate but also provides confinement and guidance for RW segmentation. Our approach was validated and compared with the conventional RW algorithm on segmenting 30 3-T prostate MR volumes. The experimental results indicated that our approach with an average DSC of 80.7±5.1%, outperformed that of the conventional RW (average DSC = 71.9±9.1%) and several other reported methods in terms of DSC accuracy and robustness.Digital Image Computing: Techniques and Applications (DICTA), 2013 International Conference on; 01/2013
[Show abstract] [Hide abstract]
ABSTRACT: Early detection of human organ diseases is of great importance for the accurate diagnosis and institution of appropriate therapies. This can potentially prevent progression to end-stage disease by detecting precursors that evaluate organ functionality. In addition, it also assists the clinicians for therapy evaluation, tracking diseases progression, and surgery operations. Advances in functional and contrast-enhanced (CE) medical images enabled accurate noninvasive evaluation of organ functionality due to their ability to provide superior anatomical and functional information about the tissue-of-interest. The main objective of this dissertation is to develop a computer-aided diagnostic (CAD) system for analyzing complex data from CE magnetic resonance imaging (MRI). The developed CAD system has been tested in three case studies: (i) early detection of acute renal transplant rejection, (ii) evaluation of myocardial perfusion in patients with ischemic heart disease after heart attack; and (iii), early detection of prostate cancer. However, developing a noninvasive CAD system for the analysis of CE medical images is subject to multiple challenges, including, but are not limited to, image noise and inhomogeneity, nonlinear signal intensity changes of the images over the time course of data acquisition, appearances and shape changes (deformations) of the organ-of-interest during data acquisition, determination of the best features (indexes) that describe the perfusion of a contrast agent (CA) into the tissue. To address these challenges, this dissertation focuses on building new mathematical models and learning techniques that facilitate accurate analysis of CAs perfusion in living organs and include: (i) accurate mathematical models for the segmentation of the object-of-interest, which integrate object shape and appearance features in terms of pixel/voxel-wise image intensities and their spatial interactions; (ii) motion correction techniques that combine both global and local models, which exploit geometric features, rather than image intensities to avoid problems associated with nonlinear intensity variations of the CE images; (iii) fusion of multiple features using the genetic algorithm. The proposed techniques have been integrated into CAD systems that have been tested in, but not limited to, three clinical studies. First, a noninvasive CAD system is proposed for the early and accurate diagnosis of acute renal transplant rejection using dynamic contrast-enhanced MRI (DCE-MRI). Acute rejection–the immunological response of the human immune system to a foreign kidney–is the most sever cause of renal dysfunction among other diagnostic possibilities, including acute tubular necrosis and immune drug toxicity. In the U.S., approximately 17,736 renal transplants are performed annually, and given the limited number of donors, transplanted kidney salvage is an important medical concern. Thus far, biopsy remains the gold standard for the assessment of renal transplant dysfunction, but only as the last resort because of its invasive nature, high cost, and potential morbidity rates. The diagnostic results of the proposed CAD system, based on the analysis of 50 independent in-vivo cases were 96% with a 95% confidence interval. These results clearly demonstrate the promise of the proposed image-based diagnostic CAD system as a supplement to the current technologies, such as nuclear imaging and ultrasonography, to determine the type of kidney dysfunction. Second, a comprehensive CAD system is developed for the characterization of myocardial perfusion and clinical status in heart failure and novel myoregeneration therapy using cardiac first-pass MRI (FP-MRI). Heart failure is considered the most important cause of morbidity and mortality in cardiovascular disease, which affects approximately 6 million U.S. patients annually. Ischemic heart disease is considered the most common underlying cause of heart failure. Therefore, the detection of the heart failure in its earliest forms is essential to prevent its relentless progression to premature death. While current medical studies focus on detecting pathological tissue and assessing contractile function of the diseased heart, this dissertation address the key issue of the effects of the myoregeneration therapy on the associated blood nutrient supply. Quantitative and qualitative assessment in a cohort of 24 perfusion data sets demonstrated the ability of the proposed framework to reveal regional perfusion improvements with therapy, and transmural perfusion differences across the myocardial wall; thus, it can aid in follow-up on treatment for patients undergoing the myoregeneration therapy. Finally, an image-based CAD system for early detection of prostate cancer using DCE-MRI is introduced. Prostate cancer is the most frequently diagnosed malignancy among men and remains the second leading cause of cancer-related death in the USA with more than 238,000 new cases and a mortality rate of about 30,000 in 2013. Therefore, early diagnosis of prostate cancer can improve the effectiveness of treatment and increase the patient’s chance of survival. Currently, needle biopsy is the gold standard for the diagnosis of prostate cancer. However, it is an invasive procedure with high costs and potential morbidity rates. Additionally, it has a higher possibility of producing false positive diagnosis due to relatively small needle biopsy samples. Application of the proposed CAD yield promising results in a cohort of 30 patients that would, in the near future, represent a supplement of the current technologies to determine prostate cancer type. The developed techniques have been compared to the state-of-the-art methods and demonstrated higher accuracy as shown in this dissertation. The proposed models (higher-order spatial interaction models, shape models, motion correction models, and perfusion analysis models) can be used in many of today’s CAD applications for early detection of a variety of diseases and medical conditions, and are expected to notably amplify the accuracy of CAD decisions based on the automated analysis of CE images.03/2014, Degree: PhD, Supervisor: Ayman S. El-Baz and Karla Conn Welch