2018/08/23 by Mohsen Hajabdollahi, Reza Esfandiarpoor, Hajabdollahi, Mohsen +11
Medicine · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Gastrointestinal Bleeding Diagnosis and Treatment
paper · pdf · doi:10.48550/arxiv.1808.07746
openalex publication_date 2018/08/23 · openalex created_date 2022/08/04 · openalex updated_date 2026/07/28
Wireless capsule endoscopy (WCE) is an effective mean for diagnosis of\ngastrointestinal disorders. Detection of informative scenes in WCE video could\nreduce the length of transmitted videos and help the diagnosis procedure. In\nthis paper, we investigate the problem of simplification of neural networks for\nautomatic bleeding region detection inside capsule endoscopy device. Suitable\ncolor channels are selected as neural networks inputs, and image classification\nis conducted using a multi-layer perceptron (MLP) and a convolutional neural\nnetwork (CNN) separately. Both CNN and MLP structures are simplified to reduce\nthe number of computational operations. Performances of two simplified networks\nare evaluated on a WCE bleeding image dataset using the DICE score. Simulation\nresults show that applying simplification methods on both MLP and CNN\nstructures reduces the number of computational operations significantly with\nAUC greater than 0.97. Although CNN performs better in comparison with\nsimplified MLP, the simplified MLP segments bleeding regions with a\nsignificantly smaller number of computational operations. Concerning the\nimportance of having a simple structure or a more accurate model, each of the\ndesigned structures could be selected for inside capsule implementation.\n