Medical imaging is one of the most promising areas for using AI tools as pattern recognition and large datasets enhance and compete with human diagnosis (by radiologists). A review of AI methodology was conducted in 2023 and concluded that such techniques would be useful for early disease detection and personalized diagnosis and treatment.
In March 2025, the Nature article discussed multimodal generation AI for medical image interpretation. Although such techniques can help clinicians, he said, “There are still some horrific obstacles to verifying the accuracy of the model, ensuring transparency, and bringing out subtle impressions.”
Researchers also focus on specific areas such as rib fracture imaging. Recent RIB research focuses on criteria for assessing AI technologies, such as the use of existing industry standards, and “The MAIC-10 checklist appears to be a valid evaluation tool for assessing the quality of such research.” In another recent study, Polish researchers reviewed the application of AI in medical imaging for cancer detection.
Some models focus on capturing uncertainty to assist clinicians and researchers, as shown in this research paper.
Given these studies and reviews, it appears that researchers and the medical device industry are trying to apply some of the principles that the FDA provided in 2024 for superior machine learning practices. Such principles include focusing on “human team performance” and testing within “clinically relevant conditions.” However, the resource reviews discussed here reveal that AI tools often provide complex output, so Quest continues to apply another of these principles. Provides “clear and essential information” to users.
[View source.]
