AI model for Diagnosis of Usual Interstitial Pneumonia
- Machine learning Other Applications
- National Institute of Advanced Industrial Science and Technology, Artificial Intelligence Research Center (AIRC)
Nagasaki University and National Institute of Advanced Industrial Science and Technology have developed a new method for creating an AI model that can diagnose usual interstitial pneumonia with high accuracy, while maintaining transparency in the basis for diagnostic decisions, by combining the expertise of physicians with efficient feature extraction techniques using AI. In this method, medical knowledge can be applied to the decision of the model by excluding features that do not affect the diagnosis, and integrating features that represent the same phenomenon based on medical doctor's experience and expertise. It helps to clarify the relationship between medical knowledge and the decision of AI, and to confirm whether the decision of the model is appropriate or not by medical doctors.
Using this method, we have developed an AI model that can make a pathological diagnosis with high accuracy for usual interstitial pneumonia, which is difficult to diagnose and has a high mortality rate.
The results of this research have been accepted for publication of in "Modern Pathology" in the Nature Publishing Group, on February 3, 2022 (Japan time).
For more information, click here.