← Back to Publications List

LungNet: A Novel Deep Learning-based Lung Disease Classification Approach with Explainable AI

Students & Supervisors

Student Authors
Aritra Das
BSc in Computer Science & Engineering, FST
Jamin Rahman Jim
BSc in Computer Science & Engineering, FST
Supervisors
Prof. Dr. Kamruddin Nur
Professor, Faculty, FST
Dr. Muhammad Firoz Mridha
Associate Professor, Faculty, FST

Abstract

The global spread of lung diseases like COVID-19 and pneumonia has challenged medical specialists in accurate di- agnosis and management. Chest X-ray (CXR) images are essential diagnostic tools, but subtle differences in disease imaging characteristics make identification difficult. Traditional methods depend on radiologists’ expertise, leading to interpretation vari- ations and potential delays. To enhance accuracy and efficiency, this study introduces "LungNet," a lightweight deep learning architecture for classifying epidemic lung diseases. The model was trained and evaluated on a diverse CXR dataset, including images of COVID-19, pneumonia, and normal lungs. Preprocessing techniques and data augmentation were applied to optimize performance and address data variability and class imbalance. LungNet achieved a training accuracy of 100% and a testing ac- curacy of 96%. Additionally, Gradient-weighted Class Activation Mapping (GRAD-CAM) was used to improve interpretability. This research advances healthcare outcomes by addressing challenges in disease identification from CXR images.

Keywords

Lung Cancer Deep Learning Explainable AI

Publication Details

  • Type of Publication: Conference
  • Conference Name: ETCoS-Grace 2024 Renewable Energy, Environment, and Technology for Sustainable Tranformation
  • Date of Conference: 22/08/2024 - 22/08/2024
  • Venue: Universitas Muhammadiyah Yogyakarta (UMY)
  • Organizer: Universitas Muhammadiyah Yogyakarta (UMY)
View on Publisher