FlexiMIND: Combined Pretrained Vision Features and Local Structure for Training-Free Deformable Medical Image Registration
Authors: Mojtaba Safari, Mingzhe Hu, Shansong Wang, Suyang Xi, Joshua Lorenz, Sean Pitroda, Aytekin Oto, Ralph R. Weichselbaum, Xiaofeng Yang
Organizations: Department of Radiation and Cellular Oncology, The University of Chicago, Chicago, IL, United States. · Department of Computer Science and Informatics, Emory University, Atlanta, GA, United States. · Department of Radiology, The University of Chicago, Chicago, IL, United States.
Objective. Deformable image registration remains challenging because corresponding anatomy may exhibit different intensity relationships. Hand-crafted modality-robust descriptors such as Modality-Independent Neighborhood Descriptor with Self-Similarity Context (MIND-SSC) and features from pretrained vision encoders provide distinct representations for correspondence estimation, but how they should be combined remains unclear. We investigate whether tighter integration improves registration over a sequential, scale-separated design. Approach. We introduce FlexiMIND, a training-free and label-free two-stage framework in which frozen FlexiCT-2D encoder features establish contextual correspondence and MIND-SSC refines the deformation within ConvexAdam. Five alternative feature-coupling strategies were evaluated under a protocol across CT-CT (DIR-Lab 4DCT 4DCT, 10 cases), MR-MR (ACDC, 50 patients), and MR-CT (Learn2Reg Abdomen MR-CT, 8 pairs) registration. Performance was assessed using target registration error (TRE), Dice similarity coefficient (Dice), SDlogJ, and local folding. Main results. FlexiMIND achieved a TRE of 1.27 mm on DIR-Lab 4DCT, a DSC of 0.807 on ACDC, and an organ-averaged DSC of 0.825 on Abdomen MR-CT. Across 15 coupling-strategy dataset comparisons, only four outperformed the reference method, with a maximum absolute improvement of 0.0043 in Dice, whereas degradation reached 20.6 % in TRE and 37.9 % in Dice. Changes in registration accuracy were closely associated with changes in deformation-field regularity, and strategies that better preserved regularity generally remained closest to the reference performance. Landmark-level analysis showed that the pretrained encoder stage primarily reduced large correspondence errors rather than uniformly improving local precision. Code is available upon acceptance at https://github.com/mosaf/FlexiMIND.
Figures & tables
Figure 1: Overview of the reference FlexiMIND registration framework. For DIR-Lab 4DCT and ACDC, a shared rigid-plus-affine pre-alignment φaffine is applied to all configurations, whereas for Abdomen MR-CT, φaffine is the identity transformation. (a) Frozen encoder features are extracted slice-wise, reduced to q=24 channels using PCA, and used by ConvexAdam to estimate the initial deformation u1 . (b) The moving image is warped by u1 , after which MIND-SSC features are computed and a second ConvexAdam stage estimates the residual deformation u2 . (c) The affine and deformable transformations are composed to obtain the final registration.
Strategy
Modification
Random projection
Replace the PCA projection W with a seeded random projection R .
Confidence gate
Weight the first-stage displacement as u1→γ⊙u1 , where γ is derived from the sharpness of CFq .
Cost-volume fusion
Combine normalized encoder and MIND-SSC costs during the first stage as CFq/CFq+CM12/CM12 .
Composed fields
Estimate an additional encoder-derived displacement u1b and compose it with u1 before MIND-SSC refinement.
MIND on the token grid
Apply a local self-similarity representation to Fq on the encoder token grid before correspondence estimation.
Table 1: Feature-coupling strategies evaluated as modifications of the reference FlexiMIND cascade. W denotes the PCA projection in Eq. 5 , R is a seeded random projection of the same dimensions, CFq and CM12 denote the encoder and MIND-SSC cost volumes, and γ(x)∈[0,1] denotes a voxel-wise confidence weight. Components not explicitly modified were held fixed relative to the reference configuration.
Figure 2: Qualitative registration result for DIR-Lab 4DCT in the sagittal plane. Columns show the fixed image, unregistered moving image, and moving image warped using random projection, ConvexAdam with MIND-SSC, DINO-Reg, and FlexiMIND. Rows show the registered image, absolute intensity difference relative to the fixed image, and Jacobian determinant of the estimated transformation. Voxels with non-positive Jacobian determinant are indicated in green. Inset values report the case-specific TRE in millimeters.
Figure 3: Qualitative registration result for ACDC in the short-axis plane. Columns follow Fig. 2 . Fixed anatomical contours are shown in blue and propagated moving contours in red. Rows show the registered image, absolute intensity difference relative to the fixed image, and Jacobian determinant of the estimated transformation. Inset values report the subject-averaged Dice for the displayed patient.
Figure 4: Qualitative registration result for Learn2Reg Abdomen MR-CT pair 0008 in the axial plane. Columns follow Fig. 2 . Fixed organ contours are shown in blue and propagated moving contours in red. The middle row shows 1−NMI to account for the multimodal image pair, and the bottom row shows the Jacobian determinant of the estimated transformation. Inset values report subject-averaged Dice for the displayed pair.
Dataset / metric
Configuration
Accuracy
95% CI
SDlogJ
Runtime (s)
DIR-Lab 4DCT CT-CT TRE (mm) ↓
Unregistered
8.4612±3.1596
6.6858 to 10.7230
–
–
ConvexAdam
1.5300±0.7348
1.2402 to 2.3064
0.0430
29.2±10.8
DINO-Reg
2.2992±0.8165
1.8995 to 2.9754
0.0633
608.8±148.8
Encoder only
1.6954±0.3628
1.4508 to 1.8954
0.0634
153.6±45.3
FlexiMIND
1.2677±0.1950
1.1305 to 1.3769
0.0643
159.0±47.1
Published reference values
Table 2: Registration accuracy, deformation-field regularity, and runtime across the three modality pairings. Configurations evaluated in this study are reported as mean ± standard deviation with 95% bootstrap confidence intervals. Published reference values report registration accuracy only and are shown separately within each dataset block. For Abdomen MR-CT, subject-averaged Dice is reported with organ-averaged Dice in parentheses. Arrows indicate the direction of better registration accuracy. Runtime was measured on an NVIDIA RTX 6000 Ada GPU.
Figure 5: Per-case distributions of the registration metric for the configurations evaluated under the common study protocol. Individual cases are overlaid on the corresponding box plots to show the variability within each dataset.
Figure 6: Landmark-level comparison on DIR-Lab 4DCT using all 3000 annotated landmarks. (a) Per-landmark error of FlexiMIND versus ConvexAdam with MIND-SSC; the dashed line indicates equal error and dotted lines mark 5mm . (b) Fraction of landmarks with error exceeding the indicated threshold for the evaluated configurations. Both panels use unrounded landmark errors and are therefore not directly comparable with the voxel-rounded TRE values in Table 2 .
DIR-Lab 4DCT
ACDC
Abdomen MR-CT
Configuration
TRE (mm)
SDlogJ
DSC
SDlogJ
DSC
SDlogJ
FlexiMIND (reference)
1.2677
0.0643
0.8067
0.0960
0.8129
0.1448
Random projection
1.4854
0.0824
0.7948
0.1267
0.5047
0.4022
Confidence gate
1.2898
0.0653
0.8082
0.0987
0.8133
0.1451
Cost-volume fusion
1.2838
0.0651
0.8082
0.0973
0.8104
0.1451
Composed fields
1.3655
0.0825
0.8110
0.1187
0.8082
0.1863
Table 3: Ablation results for the five feature-coupling strategies and the reference FlexiMIND cascade. Registration accuracy is reported as TRE in millimeters for DIR-Lab 4DCT and subject-averaged Dice for ACDC and Abdomen MR-CT. SDlogJ characterizes deformation-field regularity, with lower values indicating smoother fields.
Figure 7: Relative change in registration accuracy for the five feature-coupling strategies compared with the reference FlexiMIND cascade. The sign is normalized across metrics so that positive values indicate improved registration and negative values indicate reduced registration accuracy.
Figure 8: Relationship between deformation-field regularity and registration accuracy across the evaluated configurations. Open circles denote other measured configurations, red triangles denote the five feature-coupling strategies, and the black diamond denotes FlexiMIND. Coupling strategies are numbered as: (1) random projection, (2) confidence gate, (3) cost-volume fusion, (4) composed fields, and (5) MIND on the token grid. The vertical line indicates the SDlogJ of FlexiMIND. Larger deviations from the regularity of the reference configuration were generally associated with larger changes in registration accuracy.
Medical image registration aligns scans across time, patients, or imaging modalities to track anatomy and support diagnosis, treatment, and image-guided procedures.
Deformable registration is crucial in medical imaging. Several existing applications include lesion tracking, probabilistic atlas generation, and treatment response evaluation. However, current methods often lack robustness and generalizability across two key factors: spatial resolution and differences in anatomical coverage. We jointly optimize a registration field and a learnable dimensionality reduction module so that compressed VFM embeddings remain registration-relevant, and fuse these global semantic cues with MIND local descriptors. GLIDE-Reg achieves average dice similarity coefficients (DSC) across 6 anatomical structures of 0.859, 0.862, and 0.901 in two public cohorts (Lung250M and NLST) and one institution cohort (UCLA5DCT), and outperforms the state-of-the-art DEEDS (0.834, 0.858, 0.900) with relative improvements of 3.0%, 0.5%, and 0.1%. For target registration errors, GLIDE-Reg achieves 1.58 mm on Lung250M landmarks (compared to 1.25 mm on corrField and 1.91 mm on DEEDS) and 1.11 mm on NLST nodule centers (compared to 1.11 mm on DEEDS). The substantiated performance on the nodule centers also demonstrates its robustness across challenging downstream tasks, such as nodule tracking, which is an essential prior step for early-stage lung cancer diagnosis.
Yunzheng Zhu, Aichi Chien, Kimaya kulkarni +5
Medical & Imaging Informatics, Department of Radiological Sciences, David Geffen School of Medicine at UCLA, USA · Magnetic Resonance Research Labs, Department of Radiological Sciences, David Geffen School of Medicine at UCLA, USA · Department of Radiation Oncology, David Geffen School of Medicine at UCLA, Los Angeles, USA
Medical image registration is a fundamental task for aligning anatomical structures across images. Exhaustively searching for point-wise correspondences is prohibitively expensive in both computation and memory, particularly for high-resolution 3D medical images. Most recent learning-based methods directly predict correspondences in a coarse-to-fine manner, but often propagate intermediate estimates across scales without explicitly weighting correspondence evidence, which can allow coarse-scale errors to persist during refinement. To address these challenges, we propose ACorD, an adaptive correspondence discovery framework designed to mitigate error propagation with an attention-guided process. ACorD first predicts multiple potential correspondence candidates through Correlation-guided Cross-scale Offset Prediction (CCOP), which combines correlation maps as spatial matching priors with cross-scale features as global deformation context. Then, Multi-Head Deformable Aggregation (MHDA) aggregates these candidate correspondences using attention, producing a learned, non-uniform weighting over competing matches rather than treating them equally. The multi-head design further captures complementary deformation patterns, ensuring both global alignment and local precision. By integrating adaptive offset prediction with selective, attention-based aggregation of correspondence evidence, ACorD enables accurate deformation refinement. Extensive experiments on four public medical registration benchmarks demonstrate that adaptive correspondence discovery within a multi-scale refinement framework improves registration robustness across diverse anatomical structures.
Multi-modal image registration plays a critical role in precision medicine but faces challenges from non-linear intensity relationships and local optima. While deep learning models enable rapid inference, they often suffer from generalization collapse on unseen modalities. To address this, we propose Search-MIND, a training-free, iterative optimization framework for instance-specific registration. Our pipeline utilizes a coarse-to-fine strategy: a hierarchical coarse alignment stage followed by deformable refinement. We introduce two novel loss functions: Variance-Weighted Mutual Information (VWMI), which prioritizes informative tissue regions to shield global alignment from background noise and uniform regions, and Search-MIND (S-MIND), which broadens the convergence basin of structural descriptors by considering larger local search range. Evaluations on CARE Liver 2025 and CHAOS Challenge datasets show that Search-MIND consistently outperforms classical baselines like ANTs and foundation model-based approaches like DINO-reg, offering superior stability across diverse modalities.
Boya Wang, Ruizhe Li, Chao Chen +1
Intelligent Modelling & Analysis Group (IMA), School of Computer Science, University of Nottingham, UK · Lab for Uncertainty in Data and Decision Making (LUCID), Nottingham Biomedical Research Centre (BRC), School of Medicine, University of Nottingham, UK