Kakao Brain (CEOs Kim Byeong-hak and Kim Il-du) announced on the nineteenth that that they had released the 'Labeler Project', a medical image evaluation specialist that extracts specific disease names from chest X-ray evaluation, on GitHub.
The Labeler project is a research and development project that extracts specific disease names from various forms of natural language readings. It’s characterised by with the ability to extract a complete of 13 diseases that could be diagnosed based on chest
The reason is that when a user inputs a reading written in natural language, the evaluation is performed and the user is notified whether the 13 diseases are positive or negative.
Kakao Brain explained that as a way to contribute to improving medical diagnosis, it has begun research on a labeler project that extracts specific disease names more accurately and efficiently than other firms' models. This project has also been used to check the interior performance of Kakao Brain's chest X-ray draft interpretation technology.
In-house, we measured the accuracy of extracting disease names for 10 actual diseases, including fracture, pneumothorax, and pulmonary edema, and recorded the next accuracy (90.39%) than other firms' models (about 76%).
The project was also released on GitHub on the sixteenth. At the top of January, a paper titled ‘Improved Chest
Meanwhile, Kakao Brain announced that, just like the open source project 'Honeybee' released in January, this labeler was also released to revitalize the AI open source ecosystem.
Kim Il-du, CEO of Kakao Brain, said, “We plan to release a test set we created in order that many researchers can use the Labeler project as a test benchmark.” He added, “We’ll use the Kakao Brain language model to enhance performance by moreover learning chest X-ray data.” “We plan to enhance it further,” he said.
Reporter Jang Se-min semim99@aitimes.com
