Mohammad Ghassemi highlights a novel AI method to improve medical image segmentation.
Friends, if you (or someone you care about) ever face a serious medical issue, obtaining a medical image (CT/MRI) is often the first critical step in assessing the situation and planning treatment. However, the human body is incredibly complex, and medical images don’t always make it clear where one structure ends and another begins.
Thankfully, under the leadership of my brilliant PhD student Niloufar Eghbali and in collaboration with my exceptional colleagues Tuka Alhanai and Hassan Bagher-Ebadian, we’ve developed a novel AI method called GLoG-CSUnet. Our approach enhances AI models (Vision Transformers) by integrating Adaptable Radiomic Features, outperforming all models we tested, and significantly improving medical image segmentation.
We’re honored to be presenting this work at the International Conference on Speech and Signal Processing, widely recognized as the most prestigious signal processing conference in the world.
You can learn more about the work here: https://lnkd.in/g4qWCMGQ. If you’re curious about our work or have questions, feel free to reach out to me, or the talented Ms. Eghabli.