Abstract :

Lung Cancer Remains The Leading Cause Of Cancerrelated Mortality Worldwide, Accounting For Approximately 18% Of All Cancer Deaths. Early And Accurate Detection Is Critical To Improving Patient Outcomes, Yet Traditional Centralized Deep Learning Approaches Face Fundamental Barriers In Healthcare: Patient Privacy Regulations Restrict Data Sharing, Institutions Are Reluctant To Contribute Proprietary Data, And Siloed Datasets Limit Model Generalization. This Paper Presents A Federated Learning Framework For Lung Cancer Detection That Enables Multiple Hospitals To Collaboratively Train A Shared EfficientNetB3 Classification Model Without Exchanging Sensitive Patient Data. The Proposed System Implements The Federated Averaging (FedAvg) Algorithm Across Simulated Client Nodes, Each Training On Private Subsets Of The LC25000 Histopathology Dataset Spanning Five Diagnostic Classes: Lung Adenocarcinoma, Lung Squamous Cell Carcinoma, Lung Benign Tissue, Colon Adenocarcinoma, And Colon Benign Tissue. Under IID Data Distribution, The Federated Global Model Achieves 94.2% Accuracy, Within 1% Of A Centrally Trained Baseline, While Under Non-IID Conditions Accuracy Reaches 87.5% Due To Statistical Heterogeneity. The Trained Model Is Deployed In A Flask-based Web Application That Supports Multi-image Case Upload, Per-image EfficientNetB3 Inference At 248ms Average Latency, Caselevel Aggregation With Consensus (majority Voting), Consistency Metrics, Confidence Scores, And Prediction Distribution Visualized As Interactive Pie Charts. Plainlanguage Explanations And Downloadable HTML Case Reports Bridge The Gap Between Technical Model Outputs And Clinical Decision Support. Comprehensive Evaluation Including Unit, Integration, System, And User Acceptance Testing Demonstrates 100% Test Pass Rate And An Overall User Satisfaction Score Of 4.5/5. This Work Advances Privacypreserving Medical AI By Demonstrating That Federated Learning Enables Collaborative Intelligence Without Compromising Patient Confidentiality. Keywords—Federated Learning, Lung Cancer Detection, EfficientNetB3, Deep Learning, Privacy-Preserving AI, Medical Imaging, FedAvg, Histopathology, Flask Web Application, Multi-Image Case Analysis

Published:

02-4-2026

Issue:

Vol. 26 No. 4 (2026)

Page Nos:

155-160

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite