LungMate: AI-Driven Lung Cancer Diagnosis with Quantitative Cancer Assessment

Lung cancer continues to be among the leading cause of mortality, thereby requiring early detection and accurate treatment. Current diagnostic methods including 3D imaging are expensive and complex, thus delaying appropriate intervention in resource-limited settings. This paper discusses the use of a hybrid deep learning model, EfficientNet-B3, Swin Transformer, CBAM, and YOLOv11, to facilitate the detection of lung cancer across four classes, namely adenocarcinoma, small cell carcinoma, large cell carcinoma, and squamous cell carcinoma. The model trained with 6,789 annotated CT scans which improved robustness through aggressive data augmentation and test-time augmentation (TTA). It measures the cancer metrics: diameter, area, volume, weight and perimeter based on bounding box prediction and a pixel-to-cm conversion factor, proving to be a computationally economical alternative for 3D imaging. The model achieves a mean average precision (mAP) of 91.05%, precision of 90.12%, and accuracy of 84% at an inference time of 0.00235 seconds, signifying real-time capability. The PyQt5 GUI allows interactive visualization including image enhancement and multi-mode processing. The integration of convolution and transformer-based features enhances the detection of multi-class cancers and allows for an economical deployment, which will improve diagnostic workflow and aid clinicians in tumor assessment.