Chen Zhao1, Haipeng Tang2, Daniel McGonigle2, Zhuo He1, Chaoyang Zhang2, Yu-Ping Wang3, Hong-Wen Deng3, Robert Bober4, Weihua Zhou1,5. 1. Michigan Technological University, Department of Applied Computing, Houghton, Michigan, United States. 2. University of Southern Mississippi, School of Computing Sciences and Computer Engineering, Hattiesburg, Mississippi, United States. 3. Tulane University School of Public Health and Tropical Medicine, Tulane Center of Bioinformatics and Genomics, New Orleans, Louisiana, United States. 4. Ochsner Medical Center, Department of Cardiology, New Orleans, Louisiana, United States. 5. Michigan Technological University, Institute of Computing and Cybersystems, and Health Research Institute, Center of Biocomputing and Digital Health, Houghton, Michigan, United States.
Abstract
Purpose: In stable coronary artery disease (CAD), reduction in mortality and/or myocardial infarction with revascularization over medical therapy has not been reliably achieved. Coronary arteries are usually extracted to perform stenosis detection. As such, developing accurate segmentation of vascular structures and quantification of coronary arterial stenosis in invasive coronary angiograms (ICA) is necessary. Approach: A multi-input and multiscale (MIMS) U-Net with a two-stage recurrent training strategy was proposed for the automatic vessel segmentation. The proposed model generated a refined prediction map with the following two training stages: (i) stage I coarsely segmented the major coronary arteries from preprocessed single-channel ICAs and generated the probability map of arteries; and (ii) during the stage II, a three-channel image consisting of the original preprocessed image, a generated probability map, and an edge-enhanced image generated from the preprocessed image was fed to the proposed MIMS U-Net to produce the final segmentation result. After segmentation, an arterial stenosis detection algorithm was developed to extract vascular centerlines and calculate arterial diameters to evaluate stenotic level. Results: Experimental results demonstrated that the proposed method achieved an average Dice similarity coefficient of 0.8329, an average sensitivity of 0.8281, and an average specificity of 0.9979 in our dataset with 294 ICAs obtained from 73 patients. Moreover, our stenosis detection algorithm achieved a true positive rate of 0.6668 and a positive predictive value of 0.7043. Conclusions: Our proposed approach has great promise for clinical use and could help physicians improve diagnosis and therapeutic decisions for CAD.
Purpose: In stable coronary artery disease (CAD), reduction in mortality and/or myocardial infarction with revascularization over medical therapy has not been reliably achieved. Coronary arteries are usually extracted to perform stenosis detection. As such, developing accurate segmentation of vascular structures and quantification of coronary arterial stenosis in invasive coronary angiograms (ICA) is necessary. Approach: A multi-input and multiscale (MIMS) U-Net with a two-stage recurrent training strategy was proposed for the automatic vessel segmentation. The proposed model generated a refined prediction map with the following two training stages: (i) stage I coarsely segmented the major coronary arteries from preprocessed single-channel ICAs and generated the probability map of arteries; and (ii) during the stage II, a three-channel image consisting of the original preprocessed image, a generated probability map, and an edge-enhanced image generated from the preprocessed image was fed to the proposed MIMS U-Net to produce the final segmentation result. After segmentation, an arterial stenosis detection algorithm was developed to extract vascular centerlines and calculate arterial diameters to evaluate stenotic level. Results: Experimental results demonstrated that the proposed method achieved an average Dice similarity coefficient of 0.8329, an average sensitivity of 0.8281, and an average specificity of 0.9979 in our dataset with 294 ICAs obtained from 73 patients. Moreover, our stenosis detection algorithm achieved a true positive rate of 0.6668 and a positive predictive value of 0.7043. Conclusions: Our proposed approach has great promise for clinical use and could help physicians improve diagnosis and therapeutic decisions for CAD.
Authors: Peter Cram; John A House; John C Messenger; Robert N Piana; Phillip A Horwitz; John A Spertus Journal: Am Heart J Date: 2012-02 Impact factor: 4.749
Authors: Julian Guerrero; Septimiu E Salcudean; James A McEwen; Bassam A Masri; Savvakis Nicolaou Journal: IEEE Trans Med Imaging Date: 2007-08 Impact factor: 10.048
Authors: Stephan D Fihn; Julius M Gardin; Jonathan Abrams; Kathleen Berra; James C Blankenship; Apostolos P Dallas; Pamela S Douglas; Joanne M Foody; Thomas C Gerber; Alan L Hinderliter; Spencer B King; Paul D Kligfield; Harlan M Krumholz; Raymond Y K Kwong; Michael J Lim; Jane A Linderbaum; Michael J Mack; Mark A Munger; Richard L Prager; Joseph F Sabik; Leslee J Shaw; Joanna D Sikkema; Craig R Smith; Sidney C Smith; John A Spertus; Sankey V Williams Journal: J Am Coll Cardiol Date: 2012-11-19 Impact factor: 24.094