Abstract | ||
---|---|---|
Rotator cuff lesions are very frequent events. The diagnosis of these lesions is challenging and requires experience. The goal of this study is to develop a computer aided diagnosis (CAD) system based on Capsule Network (CapsNet) to classify rotator cuff lesions as normal, degenerated or torn in a new dataset of 1006 shoulder proton density (PD) weighted MRIs. Increasing the number of primary capsules and adding two cascaded convolution layers before capsule layer provided the CapsNet model to extract discriminative features for the better recognition of rotator cuff pathologies. The overall success rate of proposed Capsnet model was 94.75%, compared to custom designed CNN, AlexNet, GoogLenet and the gray level co-occurrence matrix (GLCM) which provided overall success rates of 93.21%, 88.45%, 87.63% and 85.20%, respectively. CapsNet performs better than CNNs on the augmented dataset as well, and robustly handles classification difficulties of rotator cuff pathologies from MRI. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1016/j.compeleceng.2019.106480 | Computers & Electrical Engineering |
Keywords | Field | DocType |
Capsule network,Convolutional neural network,Rotator cuff pathologies,PD weighted MRI,Image classification | CAD,Proton density,Pattern recognition,Convolution,Computer science,Rotator cuff,Computer-aided diagnosis,Capsule,Real-time computing,Artificial intelligence,Gray level,Discriminative model | Journal |
Volume | ISSN | Citations |
80 | 0045-7906 | 0 |
PageRank | References | Authors |
0.34 | 0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Aysun Sezer | 1 | 5 | 3.51 |
Hasan Basri Sezer | 2 | 0 | 0.34 |