Enhanced Pulmonary Pattern Classification in HRCT Images Using Advanced Data Augmentation Strategies and Fine-Tuned CNN Models

High-resolution computed tomography (HRCT) images are among the most essential and fundamental imaging modalities to diagnose and assess various pulmonary condition types. The current research focuses on classifying HRCT images into seven significant categories: ground-glass, fibrosis, micronod-ules, consolidation, healthy, reticulation, and emphysema. We first preprocess and clean the collected HRCT data and then fine-tune the state-of-the-art 2D pre-trained models like Inception-v3 and EfficientNet on the category-classified HRCT images. Further, we applied various data augmentation techniques that will improve the performance of these models. The results we obtained showed that this combination of models produced significant improvements in classification accuracy, as demonstrated by our experiment. This is expected to help doctors make more confident decisions when diagnosing pulmonary conditions. This approach not only enhances diagnostic precision but also paves the way for further research in automated medical image analysis.