Results showed that the performance of the X-ray image reading generation AI (artificial intelligence) solution developed by a domestic company was significantly superior to OpenAI’s ChatGPT.
Deepnoid, a medical AI specialist company, conducted a joint study with Inha University Department of Radiology professor Lee Roun’s team and published ‘A study on the comparison of chest It was announced on the 27th that it was published in the Journal of Clinical Medicine.
M4CXR, developed by Deepnoid, is a cloud-based medical AI solution designed with LLM (Large Language Model) that analyzes chest X-ray images and creates a draft interpretation. Deepnoid explains that this study proved that M4CXR is suitable as an auxiliary tool that strengthens clinicians’ capabilities in actual medical settings when diagnosing chest X-rays.
According to Deepnoid, Professor Loon Lee’s team, experts in medical AI, evaluated the reading results of M4CXR and ChatGPT based on five indicators: accuracy, false positives, location errors, count errors, and hallucinations. M4CXR recorded superior performance than ChatGPT in all evaluation items. In particular, in the location accuracy category, M4CXR recorded an accuracy of 76-77.5%, while ChatGPT recorded only 36-36.5%, and in the diagnostic accuracy category, M4CXR (60-62%) significantly surpassed ChatGPT (42-45%).
Professor Lee Ro-won, the first author of this paper, said, “Through this study, we were able to confirm the potential of M4CXR as a diagnostic aid tool in the medical field.” He added, “In particular, the excellent performance shown in location accuracy and diagnostic accuracy is of great value in actual clinical settings. “It will be demonstrated,” he said. The explanation is that M4CXR is not only suitable for use in a professional medical environment, but is also significantly superior to ChatGPT, a general-purpose AI, so it will be able to complement the capabilities of medical experts by optimizing diagnostic work in actual medical settings.
Woosik Choi, CEO of Deepnoid, said, “This study proves that M4CXR has excellent performance that surpasses ChatGPT in the professional medical field,” and added, “We are preparing to enter the image reading market with the goal of commercialization in the first half of 2025.” . He also emphasized, “The global video reading market is expected to grow at an average annual rate of 23.7% by 2030 to approximately 56 trillion won,” and added, “We will lead the market through M4CXR.”
Reporter Hwang Guksang [email protected]
ⓒ Money Today & mt.co.kr, unauthorized reproduction and redistribution prohibited
The categories of this article follow the classification of media outlets.
The category an article belongs to is classified by the media company.
News organizations can classify an article into two or more categories.
