Optimizing CT Image Quality through AI-based Reconstruction and Deep Learning Models for Enhanced Diagnostic Accuracy

Background: Early diagnosis of chest cancers and infections is crucial for effective treatment, and Chest CT scans play a vital role in detecting such conditions. However, quality of raw CT images often limits diagnostic accuracy. This research investigates AI-based reconstruction techniques with deep learning models to enhance CT image quality for improved chest cancer diagnostics. Methodology: Medical CT images from open-source databases undergo preprocessing, including PIL’s Image Enhance, normalization, image size adjustments, and Contrast-Limited Adaptive Histogram Equalization (CLAHE) for contrast enhancement. One-hot encoded data is split into training and testing sets, and MobileNetV2 is used for classification. Performance is assessed using confusion matrix, F1-score, precision, recall, and accuracy. Results: Experimental results show MobileNetV2 achieves 91.07% accuracy, 91.14% precision, and 91% recall, indicating strong classification capability. The high F1-scores further validate its effectiveness. Conclusion: AI-based enhancement techniques significantly improve medical image quality, facilitating faster and more accurate chest cancer diagnosis.