Mobile Based Android Application for Identification and Reduction of Intussusception

Abstract

Intussusception is abnormality in intestine when a invasion of one portion of intestine invades into adjoining part of another. Diagnosing and classification intussusception from images of ultrasonography and computed tomography can be challenging task for radiologist and pediatrician. Reduction of the intussusception is achieved by surgical and non-surgical methodologies. The non-surgical reduction methodologies are air enema and liquid enema techniques. The objective of this study is to identify and reduce intussusception by classifying intestinal CT scans and ultrasonography images to differentiate between healthy and diseased subjects. To concentrate on the challenge of identification and reduction of intussusception, an android application titled “Successful Detection and Reduction of Intussusception” (SDRI), has been designed and developed. To improve the efficacy of treatment and increase patient survival, an accurate diagnosis of an intestinal intussusception is crucial. Nevertheless, it is challenging to manually assess many Computed Tomography (CT) and Ultrasonography pictures captured in a hospital. As a result, more accurate computer-based intussusception identification techniques are needed. Numerous initiatives have looked at traditional Machine Learning (ML) techniques to automate this procedure in recent years. Recently, interests in utilizing deep learning approaches to achieve better reliably and effectively diagnose intussusception has increased. This leads to the development in this research of an improved Convolutional Neural Network (CNN) for precise categorization of intestinal CT images and Ultrasound images. Hyper-parameters are optimized for CNN layer training via the Resnet Reduction of the intussusception by non-surgical methodologies can be achieved by the same android app SDRI via Bluetooth way of communication. To treat infantile intussusception, novel equipments have been developed, the air enema technique and liquid enema technique has been used. Through these techniques, intussusception can be reduced, and the rate of reduction is faster compared to other existing methods. The equipments incorporate a microcontroller which is programmed in Embedded C to achieve operation. The deliverable equipment executes a function that regulates outflow of the air and liquid. These equipments offer a secure and forth right technique of treating intussusception. The proposed solution is a reliable, unsophisticated, and reasonably priced piece of equipment that may be employed in Out-Patient Departments, Clinics, and Hospitals. The developed systems can be implemented practically after the successful clinical trials. . The suggested model is evaluated and deployed to SDRI application using the cancer imaging archive dataset. With 89% accuracy, 93.0% sensitivity, and 75% specificity, the improved CNN model achieves the acceptable range of accuracy.