MR-linac cine target motion

0 datasets, 0 test cases. A method is ranked here when it has a result on every case of every one of them.

Read before comparing

Ranked apart: one small target per pair, the pair chosen on the truth as the widest motion, and a pool restricted to targets that move.
1 dataset of this kind is shown below and not ranked on the official board: its licence does not allow a public leaderboard. The research board (Research, above the ranking) ranks it too.

Ranking Official Research

No method has a complete result on any dataset here yet.

1 place in that table, 1 is best2/10 cases done, not ranked yetn/a does not apply not run some cases failed

Under each dataset, the method's place in that dataset's table: 1 is best, the darker the better. The mean rank averages its places over the datasets.

The scores, dataset by dataset

Each score is followed by its change against no registration: green when the method did better, red when it did worse.

TrackRAD2025, two frames of a sagittal cine MR sequence

19 pairs · ordered on Dice and SDlogJ research only

#	Method	Dice	Change	SDlogJ
1	Demons	0.773	+0.238	0.051
2	greedy	0.744	+0.209	0.083
3	ANTs SyN	0.726	+0.191	0.083
4	NiftyReg	0.711	+0.176	0.060
4	vismatch matchers · ALIKED + LightGlue	0.629	+0.094	0.025
6	vismatch matchers · DISK + LightGlue	0.624	+0.089	0.009
7	vismatch matchers · SIFT + LightGlue	0.616	+0.081	0.008
8	vismatch matchers · RoMa	0.581	+0.046	0.006
9	vismatch matchers · XFeat	0.561	+0.026	0.006
10	vismatch matchers · LoFTR	0.546	+0.012	0.002
	No registration	0.535		0.000

MR-linac cine target motion

Read before comparing

Ranking OfficialResearch

The scores, dataset by dataset

Ranking Official Research