New AI system LILAC efficiently analyzes changes in medical imaging

In the field of medicine, the analysis of imaging data has always been a complex and cumbersome process. Recently, researchers at Will Cornell Medical School have developed a new artificial intelligence system called LILAC (Learning-based Longitudinal Image Change Inference) that can efficiently and accurately analyze and detect medical images over time. The study was published February 20 in the Proceedings of the National Academy of Sciences and demonstrates the wide application potential of LILAC in multiple medical scenarios.

Traditional medical imaging analysis methods often require a lot of customization and pre-processing. Taking brain MRI data as an example, researchers usually need to spend a lot of time adjusting and correcting images in order to focus on a specific area, and even eliminate the effects of different angles and size differences. The LILAC system greatly simplifies this process, automating these complex preprocessing steps, allowing researchers to more easily perform long-time series image analysis.

The flexibility of LILAC is reflected in its ability to adapt to a variety of medical images. The research team trained LILAC through microscopic images of hundreds of sets of in vitro fertilized embryos to test their ability to judge chronological order in random image pairs. The results show that the accuracy rate of LILAC is as high as 99%. In other experiments, the system also successfully detected differences in wound healing and changes in the brain of older people and accurately predicted cognitive scores.

Dr. Kim Hee-Jung, the study’s chief designer, said the goal of LILAC is to support those who do not yet fully understand the research process, especially when there is a large variation among individuals. This technology is not only suitable for current image data, but also can flexibly respond to unknown changes in the future.

Currently, the research team plans to apply LILAC to real-world clinical scenarios, especially to predict treatment responses in prostate cancer patients through MRI scans. The launch of this innovative system undoubtedly brings new hope and possibilities to medical imaging analysis.