ROBUST COLON CANCER DETECTION FRAMEWORK USING GOV COCANET TECHNIQUE COMBINED WITH AG FDA ALGORITHM

Abstract

Municipal health authorities continue to place a high priority on early detection of colon cancer because late-stage diagnosis frequently results in higher treatment costs, higher patient mortality, and significant strain on regional healthcare services. In order to overcome this difficulty, the current study presents Governance-Driven Colon Cancer Network (Gov-CoCaNet), a governance-aligned deep learning architecture designed to provide quick and reliable colon cancer detection using histopathological images from well-known colorectal cancer datasets such as CRC-100K and GlaS. This work's main contribution is the integration of the Adaptive Governance Feature Distillation Algorithm (AG-FDA) with the Gov-CoCaNet framework to create an integrated pipeline that improves diagnostic accuracy while preserving operational efficiency. By focusing on highly discriminative tissue characteristics, AG-FDA reduces computational demand and makes the framework appropriate for implementation in municipal diagnostic facilities with limited resources. The Municipal Policy Learning Optimizer (MPLO) is used to increase training stability and speed up convergence by enabling dynamic hyperparameter adaptation through feedback mechanisms inspired by governance. Gov-CoCaNet outperforms a number of cutting-edge methods in important performance metrics such as accuracy, recall, precision, and F1-score according to experimental evaluations carried out on standard datasets. This framework supports prompt clinical decision-making and is in line with public health governance goals to strengthen early detection capabilities within regional healthcare systems by enabling automated analysis and decreasing reliance on manual assessment.