K2S Challenge Goals¶

Task¶

To develop segmentation models for anatomical structures (cartilage, bone) from 8x undersampled , multi-channel k-space data of 3D FSE-Cube MRI sequences.

Relevance¶

The fat-suppressed FSE cube images are intended for general use in a clinical setting for 3D knee MRI evaluation. These sequences provide excellent resolution in comparison with standard 2D FSE sequences currently used in clinics (slice thicknesses of 0.6mm as opposed to 3-6mm), but its relatively longer acquisition times has in part limited clinical adoption to this point. This challenge pushes cube sequences towards the shorter acquisition times that will be required for more widespread clinical adoption in pertinent settings such as tissue segmentation.