Dr. Rafat Damseh

Contemporary Artificial Intelligence in Medical Imaging

Part of Dr. Damseh's research projects aim to advance the use of contemporary artificial intelligence in medical imaging with special emphasis on brain microvascular examination. His latest work introduced generative adversarial learning models to the challenge of converting optical coherence tomography (OCT) data into high-quality two-photon microscopy (TPM) angiographies. The technique improves vascular image resolution and depth for enhanced microvascular structure analysis, which is important in studying micro-strokes and small vessel disease. On a parallel line of work, his team developed a machine learning-based MRI response prediction model directly from OCT microvascular models. With the use of deep 3D neural networks, their method predicts MRI signals with less than 1% error without the need for burdensome and time-consuming simulation pipelines. Together, these developments are an important step towards scalable, non-invasive, and AI-enabled diagnostic imaging, with novel applications for precision neuroimaging and biomarker discovery in cerebrovascular health.