New deep learning tool is designed to help radiologists evaluate chest X-rays regardless of where they work.
A new deep learning model could help radiologists in any facility interpret chest X-rays.
In a new study published in The Lancet Digital Health, investigators from Australia outlined their new tool. It is designed to alleviate heavy workloads and make it easier for providers who do not have specialty thoracic training to read these scans while reducing errors.
Chest X-rays are already the most common imaging study worldwide, and that number is growing, said the team from annalise.ai, the company that created the AI model. Developing a tool to help shoulder the weight and process the workload will be critical.
“The ability of the AI model to identify findings on chest X-rays is very encouraging,” said Catherine Jones, MBBS, thoracic radiologist, chest lead at annalise.ai, and lead study author. “Radiologists and non-radiology clinicians incorporate clinical factors into decision-making, but ultimately rely on perception of findings to underpin our clinical interpretation.”
For their study, Jones’ team trained their deep learning model on 821,681 chest X-ray images taken from 520,014 studies in 284,649 patients. They evaluated radiologist performance alone and compared it to how the same radiologist performed with the model.
Overall, 20 radiologists assessed 2,568 chest X-rays both with and without the tool. Based on the team’s assessment, the tool significantly helped radiologists improve their classification for 102 of 127 clinical findings (80 percent), and it was statistically non-inferior for 19 findings (15 percent).
Additionally, assisted radiologists had an average area under the curve of 0.808 compared with 0.713 for unassisted radiologists. The model alone had an area under the curve of 0.957, and model classification by itself was distinctly more accurate than unassisted radiologists for 117 of 125 clinical findings (94 percent). It was also non-inferior to unassisted radiologist for all clinical findings.
“Radiologist accuracy improved across a large number of clinical chest X-ray findings when assisted by the deep-learning model,” the team said. “Effective implementation of the model has the potential to augment clinicians and improve clinical practice.”
Further research, they said, is needed to confirm the efficacy and accuracy of their tool in real-world settings.
For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.
MRI-Based Deep Learning Algorithm Shows Comparable Detection of csPCa to Radiologists
May 8th 2024In a study involving over 1,000 visible prostate lesions on biparametric MRI, a deep learning algorithm detected 96 percent of clinically significant prostate cancer (csPCa) in comparison to a 98 percent detection rate for an expert genitourinary radiologist.
FDA Clears AI-Powered Qualitative Perfusion Mapping for Cone-Beam CT
May 6th 2024Reportedly validated in more than 10 clinical trials, the AngioFlow perfusion imaging software enables timely identification of brain regions with cerebral blood flow reduction and those with significant hypoperfusion.
New Literature Review Finds ChatGPT Effective in Radiology in 84 Percent of Studies
April 29th 2024While noting a variety of pitfalls with the chatbot ranging from hallucinations to improper citations, the review authors found the use of ChatGPT in radiology demonstrated “high performance” in 37 out of 44 studies.