Study Co-Authored by SriniVas R. Sadda, MD, Shows Accurate Prediction of Disease-Risk Factors Using Artificial Intelligence

On October 1, Nature Biomedical Engineering published a study that showed a new deep-learning computer framework using artificial intelligence (AI), called SLIViT (SLice Integration by Vision Transformer), that can teach itself to analyze and diagnose 3D medical images with accuracy. This application has the potential to save valuable clinician hours, reduce data costs and time, and help expedite medical research and clinical applications to improve patient outcomes.

Typically, reviewing 3D scans can take more time and requires more skill for a clinician to review and interpret, and current AI models are being trained to focus on a single imaging technique, a single part of the body, or one specific disease. What makes SLIViT so unique is that, compared to other models that analyze 3D images, this framework can be adapted for a variety of imaging techniques, such as retinal scans, ultrasound videos, MRI scans, and CT scans to identify various disease risk biomarkers.

The paper was co-authored by SriniVas R. Sadda, MD, director of Artificial Intelligence & Imaging Research at Doheny Eye Institute and professor of Ophthalmology at the UCLA Geffen School of Medicine, Eran Halperin, PhD, adjunct professor of computer science at UCLA Samueli and a professor at the Computational Medicine Department, which is affiliated with both UCLA Samueli and the UCLA David Geffen School of Medicine, and Oren Avram, PhD, a postdoctoral researcher at UCLA Computational Medicine.

Dr. Sadda noted, “What thrilled me most was SLIViT’ s remarkable performance under real-life conditions, particularly with low-number training datasets. SLIViT thrives with just hundreds – not thousands – of training samples for some tasks, giving it a substantial advantage over other standard 3D-based methods in almost every practical case related to 3D biomedical imaging annotation.”

Read more in Nature Biomedical Engineering at this link.