High-Accuracy Detection of Cancer, COVID-19, and Tuberculosis in Chest CT Scans Using Fine-Tuned InceptionV3

Revolutionizing disease diagnosis across a variety of imaging modalities have been brought on by the rapid advances in deep learning for medical image analysis. Our study introduces a novel approach to simultaneously detecting three significant diseases--cancer, COVID-19, and tuberculosis--using chest CT scans. Employing a solitary multi-class classifier, we used an optimized InceptionV3 CNN to achieve high accuracy in disease detection. For training and assessment, we employed a diverse dataset including confirmed cases of these diseases and healthy samples that were augmented for robustness. When the Chest CT dataset was used to fine-tune InceptionV3 for transfer learning, model performance improved further. Our results showed outstanding accuracy (99.43% precision) in distinguishing diseases from healthy scans. This both confirms the fidelity of the model and supports its use as a tool for diagnosis. The substantial clinical implications of this model's sensitivity and specificity allows for timely detection and best treatment of patients. By offering a precise method for recognizing critical conditions in chest CTs, this research contributes significantly to medical image processing and could lead to better healthcare outcomes and more efficient resource utilization.