Colorectal Cancer Detection Using Pre-trained Ensemble Algorithms

Abstract

Colorectal Cancer Is The Second Leading Cause Of Cancer-related Deaths Globally And The Most Common Cancer After Lung And Breast Cancer. If Caught Early, Many Lives Can Be Saved. On The Other Hand, Conventional Diagnosis Requires Doctors To Manually Analyze Histopathological Images, Which Can Be A Lengthy Process, Expensive, And Sometimes Even Lead To Human Mistakes. This Creates A Strong Need For Smarter And More Reliable Systems That Can Support Medical Professionals In Their Work. This Paper Presents A Colorectal Cancer Detection System Built On Pre-trained Deep Learning Models. Our System Employs Popular Architectures Such As Inspection V3, Resnet50, And EfficientNetB0 That Are Not Only Capable Of Identifying Key Features In Medical Images But Also Help In Enhancing The Accuracy Of The Predictions. The Model Was Built And Evaluated Using The LC25000 Dataset, Which Comprises Histopathological Images Of Both Cancerous And Non-cancerous Tissues. We Do Not Need To Depend On A Single Model But Instead Use An Ensemble Approach, Where The Strengths Of Multiple Models Are Combined Into One. This Helps Us Achieve Better Results In Terms Of Accuracy, Precision, Recall, And F1-score Compared To Traditional Methods Such As Random Forest And Naive Bayes. Our Findings Also Indicate That The System Can Detect Cancer Faster And More Effectively. Overall, This Approach Provides A Simple, Reliable, And Cost-effective Solution To Assist Doctors In Diagnosing Colorectal Cancer. By Supporting Early Detection, It Helps Improve Treatment Outcomes And Gives Patients A Better Chance Of Survival.

Keywords