Res-Caps: A Modified Capsule Network for Ultra Sonic Image Classification and Detection of Chronic Cysts

Abstract

persons with renal stones have significant health issues when they are not treated properly in time. They face a potential pain and many complications in their daily routine. Early detection in time prevents the disease and circumstances and some time even more complicated. These study outcomes the automated detection methodologies for renal stones through analyzing the medical images which undergo various image processing methods. Irregular, small uneven shaped stones can be well determined. Res-Caps, a innovative enhanced capsule network, applied for classifying cysts characteristics in ultrasonic images, demonstrating improved classification accuracy compared to standard capsule networks. The analyses described here provide insight into the functioning of capsule networks and demonstrate their potential advantages over traditional convolution neural networks. The capacity of capsule networks to represent and encode data across vector components. Object embodiment parameters via optical transformations represent a significant advancement over current models in networks. Our implementation methodology gained accuracy of 96.3%, precision 95.8%, recall 96.7%, and an F1 score of 96.2%. The best part is that it can accurately change normal renal anatomy when renal stones are present. These metrics prove that the system could be a useful diagnostic tool in medical imaging.