Participate in UCSF ci2, MICCAI 2022 Challenge: K2S K-Space2Segmentation

TweetShareEmail
By Valentina Pedoia, PhD

Each year, the Medical Image Computing and Computer Assisted Intervention Society (MICCAI) hosts competitive challenges at their international conference. This year, Valentina Pedoia, PhD, UCSF Radiology faculty and member of the UCSF Center for Intelligent Imaging (ci2), and her team will host “K2S: from undersampled K-space to Automatic Segmentation.”  

MICCA
Advance the state-of-the-art in accelerated acquisition and reconstruction of 3D knee MRI using novel algorithms to directly segment anatomical structures from 8x undersampled multi-channel k-space data. Access our uniquely curated dataset with raw k-space data and precise segmentations for 350 patients along with fully-sampled images as auxiliary information for training. Top submissions will receive cash prizes and invited to publish.

This challenge is related to applications of deep learning for knee MRI which can lead to faster MRI acquisition along with enhanced image post-processing applications such as tissue segmentation. While deep learning is changing the landscape of MRI accelerated imaging with results never obtained before, the implications for down-stream tasks such as tissue segmentation using convolutional neural networks (CNNS) are not well-characterized.

“Efficient segmentation of key anatomical structures from undersampled data is an open question that has clinical relevance,” says Dr. Pedoia. “The goal of this challenge, therefore, is to train segmentation models directly from 8x undersampled knee MRI.”

Dr. Pedoia and her team have curated a dataset of high-resolution 3D knee MRI including raw k-space data and post-processing annotations with masks for tissue segmentation. The 8x under sampled, multi-channel k-space data of 300 fat-suppressed 3D FSE Cube sequences along with segmentation masks of cartilage, meniscus, and bone will be shared as part of the challenge. Fully-sampled images will be shared for training as auxiliary information, but will not be available for inference on the test set.

Submissions will be evaluated in an end-to-end fashion from under sampled k-space data to segmentation on a test set of 200 fat-suppressed FSE-Cube sequences. “We want the challenge participants to try to solve the problem end-to-end, K2S: K-space2Segmentation” says Dr Pedoia. 

Participants can visit the challenge’s website to learn more and participate. The 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2022) takes place from September 18 through 22, 2022. See you In Singapore!