Task 3: Segmentation of Lacunes
On this page we introduce and explain task 3 in more detail. Please see the description page, data page and the evaluation page for more information about this task.
Goal
The goal of this task is to develop an automated method to segment lacunes (of presumed vascular origin) in MRI scans. The training set contains for every case 2 segmentations masks of lacunes in the full brain, because all cases have been annotated by 2 raters. The submitted automated methods will be applied on the hidden test set and should output lacunes segmentations in the full brain as well as an uncertainty indication. The desired output is a segmentation mask and a corresponding uncertainty map. The predicted segmentation mask should be continuous and will be thresholded at 0.5 during evaluation to obtain a binary segmentation mask (0: background, 1: lacune).
Data
See the data page.
Segmentations
Evaluation
Methods will be evaluated on segmentation, volume estimation, detection and uncertainty. See the evaluation page.
Expected Method Output
The expected output per case is:
- 3D prediction image with an intensity range between 0 (=background) and 1 (=lacune=foreground). This predicted image will be binarised for all metrics except the measures of volume differences (for which a clipped version will be used for assessment)
- 3D uncertainty map with an intensity range of 0 (=certain) and 1 (=uncertain). This uncertainty map will be thresholded at 0.5 when computing for the uncertainty metrics.
Both output images should have the same shape as the input images. For more information see the evaluation page and the challenge design document.