Radiologist-level stroke classification on non-contrast CT scans with Deep U-Net
This addresses stroke diagnosis for medical imaging, offering a tool that can assist or match radiologists in clinical settings.
The paper tackled the problem of classifying ischemic stroke and intracranial hemorrhage on non-contrast CT scans, achieving results that significantly outperformed 7 out of 10 radiologists and matched the remaining 3.
Segmentation of ischemic stroke and intracranial hemorrhage on computed tomography is essential for investigation and treatment of stroke. In this paper, we modified the U-Net CNN architecture for the stroke identification problem using non-contrast CT. We applied the proposed DL model to historical patient data and also conducted clinical experiments involving ten experienced radiologists. Our model achieved strong results on historical data, and significantly outperformed seven radiologist out of ten, while being on par with the remaining three.