| Abstrakt: |
A study conducted by researchers at China Medical University and Hospital in Taichung, Taiwan, has developed a deep learning model to automatically detect and diagnose acute ischemic stroke lesions on non-contrast CT brain scans. The model, called SwinUNETR, achieved a mean Dice score of 46.7% in predicting lesion volume compared to diffusion-weighted imaging verified by experts. The model was able to complete the analysis in approximately 30 seconds and showed superior performance in CT scans with lesion volumes greater than 40 ml and onset-to-CT time between 3 and 24 hours. The researchers concluded that this model could aid in timely and accurate diagnosis of acute ischemic stroke in emergency settings and low-resourced hospitals. [Extracted from the article] |
