Abstract

Esophageal adenocarcinoma is the worst cancer that may strike humans. According to the World Health Organization's predictions for 2020, it ranks sixth in global deaths and seventh in illness. Because of this, this effort aims to create a computer system that uses cuttingedge image analysis techniques and algorithms to work on the early identification of esophageal cancer. To lower the fatality rate by assisting qualified medical professionals in spotting it in an early stage. For esophageal cancer detection and classification, such as malignant vs. nonmalignant, the Upgraded Mixed Convolutional and Adaptation Residual Network (IMCARNet) method is presented in this study. The images captured during an Esophageal Endoscopy (EE) procedure are first preprocessed using an Enhanced Kaun Filter to reduce noise and smooth the image (IKF). To improve the expected exactness, particular characteristics are then extracted. IMCARNet has finally been used to classify the tissue. The results of the experiments demonstrate that the anticipated IMCARNet achieves superior execution with a higher degree of deference than the accuracy rate of 98.5%, the sensitivity rate of 97.5%, and the specificity rate of 95.5% when compared to existing algorithms such as the Deep Neural Network (CNN) and quicker Regional Multi-Layer perceptron (R-CNN) schemes