How to train lung disease diagnoses

Heatmaps the AI assessed to make lung disease diagnoses.
(ABOVE) Heatmaps the AI assessed to make lung disease diagnoses.
COVIDx-US

Researchers at Charles Darwin University (CDU) in Australia have developed an artificial intelligence (AI) tool that can diagnose lung diseases with an accuracy of 96.57%. The study, “Automated Diagnosis of Respiratory Diseases From Lung Ultrasound Videos Ensuring XAI: An Innovative Hybrid Model Approach,” was published in Frontiers in Computer Science.

The research team, which also included members from United International University in Bangladesh and Australian Catholic University, trained the model using videos from lung ultrasound procedures.

The technology works by analyzing videos frame-by-frame to identify important features of the lungs. With enough data, it was able to assess the order of the video frames to understand lung patterns and diagnose respiratory diseases, including pneumonia and COVID-19.

Niusha Shafiabady, PhDNiusha Shafiabady, PhDCDUAccording to Niusha Shafiabady, PhD, study co-author and CDU adjust associate professor, they built the model with an explainable AI feature, which allows it to describe diagnostic results and makes it more reliable to use.

“The system shows doctors why it made certain decisions using visuals like heatmaps,” Dr. Shafiabady said in a university news release. “This interpretation technique will aid a radiologist in localizing the focus area and improve clinical transparency substantially.”

Ultimately, she said the AI technology will improve outcomes for patients with lung disease by identifying and initiating treatment early.

“This model helps doctors diagnose lung diseases quickly and accurately, supports their decision-making, saves time and serves as a valuable training tool.”

Dr. Shafiabady said the model can also be trained to identify other respiratory conditions, such as asthma, tuberculosis, pulmonary fibrosis or lung cancer.

More in Pulmonary
Page 1 of 16
Next Page