ConvexAdam (MIND)

convex_adam_MIND with the repository's defaults: MIND-SSC features, a coupled-convex coarse search, then Adam instance optimisation.

RegistrationGPUversion b229e52every datasetApache-2.0

Get the app Source

Run it

With its KonfAI app, at the benchmark's settings (the fidopt app repositories are private for now: a Hugging Face token with access opens them):

pip install konfai-apps

configuration		Command
ConvexAdam (MIND)	Get the app	`konfai-apps infer fideus-labs/fidopt-registration:ConvexAdam_MIND -i fixed.nii.gz -i moving.nii.gz -o out --gpu 0`
the authors' Learn2Reg 2021 LungCT entry		No app for this configuration yet.

Where it works

Its place on each kind of problem it has results on, among the methods ranked there. A method that leads one problem can trail another: read the one closest to yours.

Problem	Place	Ranked	Datasets
Brain MR, same patient	1	of 2	1 of 1
Heart SPECT, inter-patient	1	of 1	1 of 1
Hand MR, two poses, same patient	1	of 2	1 of 1
Tumours under treatment, same patient	1	of 2	1 of 1
Abdomen MR, inter-patient	1	of 2	2 of 2
Fluorescence microscopy, between animals	1	of 1	1 of 1
Mouse micro-CT, same animal	1	of 1	1 of 1
Light-sheet microscopy of whole mice, between animals	1	of 2	1 of 1
Abdomen CT, same patient, files without their voxel size	1	of 2	2 of 2
Mouse brain MR, between animals	1	of 2	1 of 1
Marmoset MRI before and after fixation	1	of 1	1 of 1
Temporal bone CBCT, between donors	1	of 4	1 of 1
Pediatric T1 MRI with automatic regional labels	1	of 4	1 of 1
Thorax and abdomen MR, breathing, same patient	2	of 2	1 of 1
Light-sheet whole mouse brains, two channels, known warp (out of core)	2	of 2	1 of 1
Repeat scans, same patient	3	of 5	1 of 1
Simulated thoracic CBCT, known motion	3	of 4	1 of 1
Repeat chest CT, same patient, tumour truth	4	of 5	1 of 1
Abdomen CT, same patient		not ranked yet	1 of 3
Abdomen CT, inter-patient		not ranked yet	1 of 3
Brain MR, inter-subject		not ranked yet	2 of 4
Two modalities, same patient		not ranked yet	3 of 5
Learn2Reg official validation pairs		not ranked yet	1 of 4
Chest and body CT, inter-patient		not ranked yet	1 of 2
Head, neck and spine, inter-patient		not ranked yet	2 of 4
Known deformation, two images of one session		not ranked yet	3 of 5
Same patient, truth made by a program		not ranked yet	5 of 8

Results

Dataset	#	Read by	Score	Change	SDlogJ	Failed	Time s
ACDC	3	Dice	0.583	+0.025	0.004	9	7
Pediatric T1 MRI with and without previous ADHD diagnosis	1	Dice	0.609	+0.197	0.060	0	12
Mouse whole-body light-sheet (AIMOS)	1	Dice	0.443	+0.231	0.099	0	15
AIND SmartSPIM whole mouse brains, two channels	2	TRE µm	61.18	−91.52	0.087	0	8
AMOS MR/MR	1	Dice	0.426	+0.161	0.122	0	14
Abdominal-DIR-QA9/10 pairs		Dice	0.765	+0.333	0.097	0	31
FDG-PET-CT-Lesions (autoPET)7/30 pairs		Dice	0.707	−0.062	0.178	0	61
PSMA-PET-CT-Lesions (autoPET III and IV)29/30 pairs		Dice	0.628	−0.149	0.168	0	58
Baby Open Brains (BOBs)	1	Dice	0.762	+0.166	0.071	0	9
Human Bony Labyrinth	1	Dice	0.927	+0.148	0.123	0	6
BraTS-Reg	1	TRE mm	2.23	−4.10	0.030	0	11
Breast-MRI-NACT-Pilot	1	Dice	0.656	+0.147	0.089	0	9
Colorectal-Liver-Metastases, liver and veins	2	Dice	0.281	+0.065	0.127	0	20
CT Colonography (ACRIN 6664)8/10 pairs		Dice	0.713	+0.235	0.133	0	77
CT-ORG, pairs of patients3/10 pairs		Dice	0.509	+0.099	0.149	0	42
CC-Tumor-Heterogeneity	1	TRE mm	8.94	−0.21	0.090	0	9
Head Neck CBCT CT (HUST)	1	Dice	0.699	+0.028	0.040	0	7
HNTS-MRG 2024	1	TRE mm	3.52	−0.70	0.067	0	19
HQColon6/11 pairs		Dice	0.586	+0.069	0.133	0	78
PIANO hand MRI	1	TRE mm	33.00	−2.18	0.165	0	18
OAIZIB-CM	1	Dice	0.667	+0.262	0.103	0	21
ThoraxCBCT (Learn2Reg 2023), validation	2	Dice	0.594	+0.250	0.117	0	15
LCTSC	1	Dice	0.729	+0.276	0.121	0	30
LUMIR (Learn2Reg 2024), training images	1	Dice	0.620	+0.159	0.049	0	9
LUNA25	2	TRE mm	0.63	−6.45	0.090	0	34
Liver-DIR-QA	1	Dice	0.743	+0.303	0.117	0	30
CT follow-up of liver metastases	1	Dice	0.863	+0.129	0.113	0	12
PLC-CECT	1	Dice	0.964	+0.058	0.046	0	25
Longitudinal-CT21/29 pairs		TRE mm	5.19	−9.81	0.101	0	40
Lung-DIR-QA5/10 pairs		TRE mm	4.46	−8.53	0.089	0	34
M-CRIB 2.0 and M-CRIB-WM	1	Dice	0.580	+0.220	0.038	0	10
MSLesSeg9/10 pairs		Dice	0.778	−0.021	0.026	0	8
Brain/MINDS marmoset in vivo and ex vivo T2 MRI	1	Dice	0.693	+0.092	0.155	0	6
MMB-MRI	1	Dice	0.775	+0.051	0.035	0	5
Mouse whole-body micro-CT	1	Dice	0.627	+0.199	0.078	0	6
Multimodal-HC, whole-body MR/CT	1	Dice	0.645	+0.210	0.211	0	24
Myocardial perfusion SPECT	1	Dice	0.828	+0.108	0.067	0	5
NeuroPAL atlas (C. elegans heads)	1	TRE µm	13.02	−0.19	0.101	0	7
OpenEar, inter-specimen temporal-bone CBCT	1	Dice	0.596	+0.114	0.128	0	54
PDDCA	1	Dice	0.519	+0.318	0.141	0	34
PROMISE12	3	Dice	0.657	+0.041	0.099	0	12
PSMA-PET-CT-Lesions follow-up (Learn2Reg 2026 PSMAReg)	4	TRE mm	5.27	−2.31	0.096	0	60
Pancreas-CT, 8 organs9/10 pairs		Dice	0.382	+0.099	0.121	0	49
Pediatric-CT-SEG14/17 pairs		Dice	0.423	+0.140	0.134	0	45
Pediatric thoracic dynamic MRI	2	Dice	0.889	+0.045	0.030	0	8
Pelvic-Reference-Data	2	TRE mm	2.19	−0.53	0.140	0	23
Pelvic landmarks CT/CT (LUMC)	2	Dice	0.751	+0.136	0.138	0	33
Pelvic landmarks MR/CT (LUMC)	2	Dice	0.604	+0.074	0.090	0	11
Prostate-MRI-US-Biopsy	1	Dice	0.621	+0.002	0.253	0	7
LCTSC CT and simulated CBCT, known deformation	3	TRE mm	0.27	−0.96	0.030	0	6
QIN-PROSTATE-Repeatability	3	Dice	0.688	+0.045	0.082	0	7
QUADRA_HC, whole-body CT test-retest	3	Dice	0.838	+0.287	0.090	0	26
RIDER Lung CT	4	Dice	0.784	+0.048	0.081	0	30
RibSeg v215/30 pairs		Dice	0.301	+0.258	0.133	0	52
SPIDER	2	Dice	0.748	+0.163	0.036	6	5
Spine-generic multi-subject	1	Dice	0.853	+0.523	0.107	0	5
SynthRAD2023 CBCT/CT, brain11/18 pairs		Dice	0.737	+0.073	0.083	0	6
SynthRAD2023 MR/CT, brain8/18 pairs		Dice	0.635	+0.061	0.075	0	6
SynthRAD2023 CBCT/CT, pelvis8/18 pairs		Dice	0.529	+0.084	0.118	0	6
SynthRAD2023 MR/CT, pelvis8/18 pairs		Dice	0.749	+0.054	0.154	0	8
Tongue musculature MRI	1	Dice	0.707	+0.209	0.030	0	8
TopBrain 2025/2026	1	Dice	0.632	+0.184	0.131	0	29
TopCoW 2024	1	Dice	0.730	+0.118	0.131	0	38
TotalSegmentator MRI	1	Dice	0.331	+0.146	0.123	0	8
UCSD-BMETS-Longitudinal21/30 pairs		Dice	0.792	+0.001	0.020	0	26
UltraCortex	2	Dice	0.662	+0.165	0.060	0	17
VerSe6/11 pairs		Dice	0.406	+0.215	0.131	0	21
Vestibular-Schwannoma-SEG	1	Dice	0.700	+0.445	0.072	0	8
µ-RegPro	1	Dice	0.370	+0.152	0.112	0	4

Source: github.com/multimodallearning/convexAdam
Reference: Siebert et al., IEEE Transactions on Medical Imaging 2024