Automated Pneumonia Based Lung Diseases Classification with Robust Technique Based on a Customized Deep Learning...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• Infectious lung diseases (ILD) pose a significant global health burden, with COVID-19 exacerbating hospital intensive care unit (ICU) occupancy.
• Early and accurate diagnosis of ILD is vital for reducing treatment durations and healthcare system strain.

Purpose of the Study:

• To develop and evaluate a novel deep learning model for the early classification of healthy individuals, COVID-19, and pneumonia from chest X-ray (CX-R) images.
• To enhance classification performance and model reliability in detecting ILD.

Main Methods:

• A customized deep learning model (ACL model) was developed, integrating Convolutional Neural Networks (CNN) with attention and Long Short-Term Memory (LSTM) mechanisms.
• The marker-controlled watershed (MCW) segmentation algorithm was employed to highlight significant features in CX-R images.
• The ACL model was trained and validated using various training-test ratios (90-10%, 80-20%, 70-30%).

Main Results:

• The ACL model achieved high accuracy rates: 100% for the 90-10% training-test ratio, and 96% for both 80-20% and 70-30% ratios.
• The proposed model demonstrated superior performance compared to existing methods for ILD classification.
• Detailed analysis confirmed the contribution of implemented strategies to the model's classification accuracy.

Conclusions:

• Deep learning models, such as the proposed ACL model, show promise as effective decision support tools for physicians in the early diagnosis of ILD.
• Further validation with diverse datasets is essential to ensure the reliability and generalizability of these AI-driven diagnostic applications.