SymmDiff: Symmetry-Guided Diffusion for Fracture Localization in Pelvic Radiographs

Abdul Rahman¹, Afnan Ghafoor², Bumshik Lee^1,*

¹Energy AI Track, Korea Institute of Energy Technology (KENTECH), ²Information and Communication Engineering Department, Chosun University

²Information and Communication Engineering Department, Chosun University

^*Corresponding author

Accepted at MICCAI 2026

Paper Code BibTeX

SymmDiff couples a flipped contralateral reference into the reverse diffusion trajectory, producing a pseudo-healthy reconstruction whose discrepancy with the input precisely localizes pelvic fractures.

Abstract

Unsupervised anomaly detection (UAD) offers a promising direction for fracture analysis where abnormal annotations are scarce, yet pelvic X-ray fracture detection remains exceptionally challenging due to overlapping anatomy, projection artifacts, and the inherently asymmetric presentation of fracture patterns.

We introduce SymmDiff, a symmetry-aware diffusion framework that embeds contralateral anatomical consistency directly into the generative correction process, mirroring the clinical practice of bilateral comparison. By pairing anatomically corresponding left–right regions and enforcing symmetry-consistent reconstruction during diffusion, SymmDiff generates a normal-corrected radiograph that selectively suppresses fracture-induced asymmetries while preserving global anatomical structure, enabling precise fracture localization through residual discrepancies between the input image and its symmetry-corrected counterpart.

Extensive evaluations on pelvic radiographs spanning subtle to severe fractures show that SymmDiff consistently outperforms existing anomaly-detection baselines, highlighting the effectiveness of symmetry-guided generative modeling for reliable fracture localization in clinical imaging.

Key Contributions

A symmetry-conditioned latent diffusion model that establishes a patient-specific representation of normal pelvic anatomy through contralateral correspondence, providing a structural prior for fracture localization.
A symmetry-conditioned reverse diffusion formulation in which contralateral information guides the reverse transition at each step, stabilizing the anatomical trajectory during iterative denoising.
An ROI-guided reverse diffusion scheme with contralateral latent-swap initialization that softly restricts updates to regions exhibiting elevated bilateral discrepancy, reducing projection-induced over-correction while preserving global anatomy.

Method

SymmDiff is a symmetry-guided latent diffusion framework. The input radiograph x and its horizontally flipped reference x_sym are encoded by a pelvic-finetuned VAE to latents z, z_sym. Because pelvic anatomy is bilaterally correspondent, conditioning on z_sym reduces uncertainty (H(Z | R) < H(Z)) and provides a patient-specific structural anchor.

Training applies random patch masking to the noisy latent while channel-concatenating z_sym as guidance, forcing the U-Net ε_θ to inpaint masked regions using bilateral context rather than memorizing local textures.

Inference derives a soft ROI mask from a side-to-side discrepancy score and initializes the reverse trajectory via a latent swap inside the ROI. Deterministic DDIM (η = 0) updates only inside the ROI, making the reverse process the identity outside it — imperfect mask estimation cannot induce global structural drift.

Multi-scale anomaly fusion combines four complementary residuals — pixel, gradient, perceptual (ResNet features), and latent — into a single anomaly map, capturing thin cortical breaks as well as broader deformities.

Results

Quantitative Comparison

We evaluate on a retrospective Chosun University Hospital pelvic X-ray dataset (1,268 normal radiographs split 80/10/10 train/val/test; 534 abnormal cases reserved for evaluation, with pixel-level masks). SymmDiff consistently outperforms recent diffusion-based UAD baselines. Relative to the second-best method (THOR), image-level AUROC / AUPRC / F1_max improve by +4.9 / +2.5 / +2.8, and pixel-level by +3.9 / +7.7 / +5.5.

Method	Image-level			Pixel-level
Method	AUROC	AUPRC	F1_max	AUROC	AUPRC	F1_max
DDPM	84.5	83.0	80.1	76.5	23.3	24.6
AnoDDPM	87.5	85.4	83.0	79.5	26.1	27.8
AutoDDPM	89.0	85.9	84.2	80.6	27.6	27.0
THOR	89.3	89.1	88.3	84.5	36.2	37.2
SymmDiff (ours)	94.2	91.6	91.1	88.4	43.9	42.7

Table 1. Quantitative comparison on the Chosun University Hospital pelvic X-ray dataset. Best results in bold.

Qualitative Localization

Across minor-to-large fractures (Samples 1–3), SymmDiff yields anatomically plausible reconstructions and localizes anomalies consistent with the masks. For the implant case (Sample 4), responses concentrate around the implant with minimal spillover; for the healthy case (Sample 5), activations remain low.

Qualitative pelvic-radiograph localization results

Figure 2. Qualitative localization results on pelvic X-rays. Columns show five different test samples. Rows: input radiograph, reconstructed corrected image, binary ground-truth mask, and predicted anomaly heatmap.

Ablations

Each SymmDiff component contributes. Removing reference (symmetry) conditioning causes the largest drop, confirming the side-to-side prior as the strongest stabilizer; ROI-constrained denoising prevents global drift, and multi-scale fusion is necessary for thin fractures that are weak in raw intensity residuals.

Variant	Image-level			Pixel-level
Variant	AUROC	AUPRC	F1_max	AUROC	AUPRC	F1_max
SymmDiff (full)	94.2	91.6	91.1	88.4	43.9	42.7
w/o reference conditioning	89.4	88.2	86.2	79.9	32.4	30.6
w/o random patch masking	93.7	90.8	89.7	86.3	36.0	35.7
w/o ROI-constrained denoising	91.8	90.3	89.8	83.2	35.7	35.1
w/o latent swap	92.6	90.5	89.8	84.5	38.5	37.7
w/o multi-scale fusion (pixel only)	93.9	91.2	90.7	86.6	36.8	35.9

Table 2. Ablation study of SymmDiff components on the Chosun University Hospital pelvic X-ray dataset.

Insights

Insight 1. Contralateral consistency is a strong inductive bias for pelvic UAD — it stabilizes "normality" where global density alone fails.

Insight 2. Coupling symmetry into the reverse trajectory — not post-hoc — prevents anatomical drift during iterative denoising.

Insight 3. ROI-gated updates with identity-outside-mask make the pipeline robust to imperfect contralateral references.

Insight 4. Multi-scale residual fusion is essential for thin cortical breaks that vanish in any single residual view.

BibTeX

@inproceedings{rahman2026symmdiff,
  title     = {SymmDiff: Symmetry-Guided Diffusion for Fracture
               Localization in Pelvic Radiographs},
  author    = {Rahman, Abdul and Ghafoor, Afnan and Lee, Bumshik},
  booktitle = {Medical Image Computing and Computer-Assisted
               Intervention -- MICCAI 2026},
  year      = {2026},
  publisher = {Springer},
}