Smart Detection of Lung Cancer via Image Processing and Symptom Integration

This research presents a smart lung cancer detection system that combines medical image processing with clinical symptom analysis to improve early diagnosis and risk assessment. The proposed framework integrates a Convolutional Neural Network (CNN) for analyzing lung scan images and an XGBoost-based machine learning model for evaluating important clinical parameters such as nodule diameter, spiculation score, lobulation, smoking exposure (pack years), age factor, and calcification. A fusion-based prediction mechanism combines outputs from both models using weighted averaging to generate a more reliable malignancy risk score. The backend system is developed using FastAPI, enabling efficient API-based prediction and real-time analysis. Image inputs are preprocessed through grayscale conversion, resizing, normalization, and tensor transformation before deep learning inference. Experimental implementation demonstrates that integrating imaging features with patient symptoms and risk factors enhances prediction capability compared with standalone methods. The proposed system offers a scalable, interpretable, and deployment-ready solution for assisting healthcare professionals in early lung cancer screening, clinical decision support, and improved patient management.