Improved swin transformer-based thorax disease classification with optimal feature selection using chest X-ray.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer-Aided Diagnosis

Background:

• Thoracic diseases pose significant health risks, necessitating accurate and timely diagnosis.
• Current diagnostic methods for thorax diseases can be time-consuming and may lack precision.

Purpose of the Study:

• To propose a deep learning model for accurate thorax disease classification using Chest X-rays.
• To enhance feature extraction and selection for improved diagnostic performance.

Main Methods:

• Pre-processing of Chest X-ray images including resizing, normalization, and data augmentation.
• Feature extraction using an Enhanced Auto-Encoder (EnAE) with an attention module.
• Feature selection via the Chaotic Whale Optimization (ChWO) Algorithm.
• Classification using an Improved Swin Transformer (IMSTrans) model.

Main Results:

• The proposed EnAE + ChWO + IMSTrans model achieved high performance metrics.
• Key performance indicators included Accuracy (0.964), Precision (0.977), Recall (0.9845), and F-Score (0.964).
• The model demonstrated reliable and efficient thorax disease classification.

Conclusions:

• The developed deep learning model offers a reliable and efficient solution for thorax disease classification.
• The integration of EnAE, ChWO, and IMSTrans significantly improves diagnostic accuracy.
• This approach holds promise for advancing computer-aided diagnosis in thoracic imaging.