arXiv reaDer
Improve Cross-Modality Segmentation by Treating MRI Images as Inverted CT Scans
Computed tomography (CT) segmentation models frequently include classes that are not currently supported by magnetic resonance imaging (MRI) segmentation models. In this study, we show that a simple image inversion technique can significantly improve the segmentation quality of CT segmentation models on MRI data, by using the TotalSegmentator model, applied to T1-weighted MRI images, as example. Image inversion is straightforward to implement and does not require dedicated graphics processing units (GPUs), thus providing a quick alternative to complex deep modality-transfer models for generating segmentation masks for MRI data.
updated: Sat May 04 2024 14:02:52 GMT+0000 (UTC)
published: Sat May 04 2024 14:02:52 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト