← Back to Publications List

TelLungNet - Enabling Telemedicine Utilizing an Improved U-Net Lung Image Segmentation

Students & Supervisors

Student Authors
Rifat Al Mamun Rudro
Master of Science in Computer Science, FST
Api Alam
BSc in Computer Science & Engineering, FST
Shafin Talukder
BSc in Computer Science & Engineering, FST
Tanvir Ahmed
BSc in Computer Science & Engineering, FST
Nayma Islam
BSc in Computer Science & Engineering, FST
Supervisors
Prof. Dr. Kamruddin Nur
Professor, Faculty, FST

Abstract

This study presents a novel approach to web-based telemedicine services by utilizing and improving the U-Net deep learning architecture. Here we present a user-friendly web application designed for medical professionals to diagnose chest X-ray images easily using the service provided by our proposed TelLung-Net architecture. By using the web application a user can upload their X-ray images, the proposed architecture then segments the image, and instant segmentation results provide visual aids during remote consultation for both patients and medical professionals. In this research, we use 1228 X-Ray images from Mendeley data for training and testing our proposed architecture. Our experiments demonstrate that the proposed TelLung-Net achieves 96.26% accuracy with a 0.95 f1 score in chest X-ray segmentation. Incorporating image segmentation into U-Net deep learning architecture significantly improves the precision and accuracy of identifying issues in X-ray images. It also improves system reliability, and reduces time by identifying chest abnormalities.

Keywords

Image Segmentation U-Net architecture Biomedical Image Processing Telehealth

Publication Details

  • DOI: 10.1109/CAI59869.2024.00247
  • Type of Publication: Conference
  • Conference Name: 2024 IEEE Conference on Artificial Intelligence (CAI)
  • Date of Conference: 25/06/2024 - 27/06/2024
  • Venue: Marina Bay Sands, Singapore
  • Organizer: CMA International, National University of Singapore, Nanyang Technology University
View on Publisher