Artificial Intelligence (AI) might change into a radiologist’s finest pal, with researchers coaching the expertise to precisely diagnose pneumonia, COVID-19 and different lung illnesses.
The new examine by researchers from Charles Darwin University (CDU), United International University, and Australian Catholic University (ACU) describes creating and coaching an AI mannequin to research lung ultrasound videos and diagnose respiratory illnesses.
The research is published within the journal Frontiers in Computer Science.
The mannequin works by inspecting every video body to search out necessary options of the lungs and assessing the order of the video frames to know the patterns of the lungs over time.
The mannequin then identifies particular patterns indicating completely different lung illnesses and, primarily based on this data, classifies the ultrasound right into an analysis class resembling regular, pneumonia, COVID-19 and different lung illnesses.
Co-author and CDU adjunct Associate Professor Niusha Shafiabady mentioned the mannequin had an accuracy of 96.57%, with the AI analyses verified by medical professionals.
“The mannequin additionally makes use of AI methods to indicate radiologists why it made sure choices, making it simpler for them to belief and perceive the outcomes,” Associate Professor Shafiabady mentioned.
The mannequin makes use of explainable AI, a way which permits human customers to know and belief the outcomes created by machine studying algorithms.
“The explainability of the proposed mannequin goals to extend the reliability of this method,” Associate Professor Shafiabady mentioned.
“The system reveals docs why it made sure choices utilizing visuals like heatmaps. This interpretation approach will support a radiologist in localizing the main target space and enhance scientific transparency considerably.
“This mannequin helps docs diagnose lung illnesses shortly and precisely, helps their decision-making, saves time, and serves as a invaluable coaching instrument.”
Associate Professor Shafiabady mentioned if fed the suitable knowledge, the mannequin might be skilled to determine extra illnesses resembling tuberculosis, black lung, bronchial asthma, most cancers, persistent lung illness, and pulmonary fibrosis.
The research was led by researchers at United International University in Bangladesh, alongside CDU researchers Dr. Asif Karim, Dr. Sami Azam, Dr. Kheng Cher Yeo, Professor Friso De Boer and Associate Professor Niusha Shafiabady, who can also be a researcher at ACU.
Potential avenues for analysis embrace coaching the mannequin to evaluate different imaging, resembling CT scans and X-rays.
More data:
Arefin Ittesafun Abian et al, Automated analysis of respiratory illnesses from lung ultrasound movies making certain XAI: an modern hybrid mannequin method, Frontiers in Computer Science (2024). DOI: 10.3389/fcomp.2024.1438126
Citation:
Researchers practice AI to diagnose lung illnesses with 96.57% accuracy (2025, January 24)
25
ai-lung-diseases-accuracy.html
.
. The content material is offered for data functions solely.