Randomized trial demonstrates efficient skull stripping in mouse brains, suggesting applicability across imaging conditions.
Key Points
This research aims to improve skull stripping techniques in mouse brain MRI by utilizing template-based label propagation for automation and efficiency.
Developed a high throughput skull stripping pipeline using template-based label propagation.
Constructed a population average ex vivo T2-weighted mouse brain MRI template, manually annotating a single brain mask in template space.
Trained an attention-based 3D U-Net segmentation model using the propagated labels and evaluated its performance on in vivo datasets.
Compared to conventional methods, the proposed pipeline achieved competitive segmentation performance with reduced manual annotation effort.
Training with propagated labels alone resulted in robust segmentation performance, highlighting the importance of label consistency.
Applying the full pipeline to in vivo datasets resulted in strong segmentation performance, indicating adaptability across imaging conditions.