Studies from Department of Stomatology Further Understanding of Jaw Cysts (MFI-Net: Multi-Level Feature Integration Network With SE-Res2Conv Encoder for Jaw Cyst Segmentation).

A new study from the Department of Stomatology focuses on the accurate segmentation of jaw cysts from medical images using deep-learning techniques. The researchers propose a multi-level feature integration network (MFI-Net) with an SE-Res2Conv encoder to address the challenges of blurred boundaries and chaotic backgrounds. The MFI-Net model demonstrates superior results compared to other methods, with Dice, IoU, and Jaccard values reaching up to 93.47% in the original database. The computational efficiency of MFI-Net is also impressive, with a speed of over 100 frames per second on a NVIDIA RTX6000 graphics card. [Extracted from the article]