Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Inpainting Pathology in Lumbar Spine MRI with Latent Diffusion

Colin Hansen, Simas Glinskis, Ashwin Raju, Micha Kornreich, JinHyeong Park, Jayashri Pawar, Richard Herzog, Li Zhang, Benjamin Odry

TL;DR

MRI data scarcity and class imbalance impede automated radiology diagnostics. The authors propose a latent diffusion model–based inpainting approach that uses a voxelwise Gaussian‑weighted noise schedule to insert pathologies into healthy lumbar spine MRI, trained on pathology‑labeled FSUs and evaluated on disc herniation and central canal stenosis. The method yields superior Fréchet Inception Distance and MS‑SSIM compared with two LDM baselines, and radiologist validation shows high realism and substantial accuracy of inserted pathology. This approach provides a computationally efficient, anatomically coherent data augmentation tool to balance datasets and improve robustness in medical imaging.

Abstract

Data driven models for automated diagnosis in radiology suffer from insufficient and imbalanced datasets due to low representation of pathology in a population and the cost of expert annotations. Datasets can be bolstered through data augmentation. However, even when utilizing a full suite of transformations during model training, typical data augmentations do not address variations in human anatomy. An alternative direction is to synthesize data using generative models, which can potentially craft datasets with specific attributes. While this holds promise, commonly used generative models such as Generative Adversarial Networks may inadvertently produce anatomically inaccurate features. On the other hand, diffusion models, which offer greater stability, tend to memorize training data, raising concerns about privacy and generative diversity. Alternatively, inpainting has the potential to augment data through directly inserting pathology in medical images. However, this approach introduces a new challenge: accurately merging the generated pathological features with the surrounding anatomical context. While inpainting is a well established method for addressing simple lesions, its application to pathologies that involve complex structural changes remains relatively unexplored. We propose an efficient method for inpainting pathological features onto healthy anatomy in MRI through voxelwise noise scheduling in a latent diffusion model. We evaluate the method's ability to insert disc herniation and central canal stenosis in lumbar spine sagittal T2 MRI, and it achieves superior Frechet Inception Distance compared to state-of-the-art methods.

Inpainting Pathology in Lumbar Spine MRI with Latent Diffusion

TL;DR

MRI data scarcity and class imbalance impede automated radiology diagnostics. The authors propose a latent diffusion model–based inpainting approach that uses a voxelwise Gaussian‑weighted noise schedule to insert pathologies into healthy lumbar spine MRI, trained on pathology‑labeled FSUs and evaluated on disc herniation and central canal stenosis. The method yields superior Fréchet Inception Distance and MS‑SSIM compared with two LDM baselines, and radiologist validation shows high realism and substantial accuracy of inserted pathology. This approach provides a computationally efficient, anatomically coherent data augmentation tool to balance datasets and improve robustness in medical imaging.

Abstract

Data driven models for automated diagnosis in radiology suffer from insufficient and imbalanced datasets due to low representation of pathology in a population and the cost of expert annotations. Datasets can be bolstered through data augmentation. However, even when utilizing a full suite of transformations during model training, typical data augmentations do not address variations in human anatomy. An alternative direction is to synthesize data using generative models, which can potentially craft datasets with specific attributes. While this holds promise, commonly used generative models such as Generative Adversarial Networks may inadvertently produce anatomically inaccurate features. On the other hand, diffusion models, which offer greater stability, tend to memorize training data, raising concerns about privacy and generative diversity. Alternatively, inpainting has the potential to augment data through directly inserting pathology in medical images. However, this approach introduces a new challenge: accurately merging the generated pathological features with the surrounding anatomical context. While inpainting is a well established method for addressing simple lesions, its application to pathologies that involve complex structural changes remains relatively unexplored. We propose an efficient method for inpainting pathological features onto healthy anatomy in MRI through voxelwise noise scheduling in a latent diffusion model. We evaluate the method's ability to insert disc herniation and central canal stenosis in lumbar spine sagittal T2 MRI, and it achieves superior Frechet Inception Distance compared to state-of-the-art methods.
Paper Structure (17 sections, 3 equations, 1 figure, 2 tables)

This paper contains 17 sections, 3 equations, 1 figure, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Image Synthesis
  4. Diffusion Models
  5. Inpainting
  6. Method
  7. Latent Diffusion
  8. Weighted Inpainting
  9. Inpainting Pathology
  10. Experiments
  11. Data
  12. Implementation Details
  13. Baseline Methods
  14. Comparison
  15. Discussion
  16. ...and 2 more sections

Figures (1)

  • Figure 1: Left, a lumbar spine sagittal T2 weighted MRI is shown with L1-2 through L5-S1 disc centers localized by a deep reinforcement learning model. Associated FSUs are cropped to 8 $\times$ 8 $\times$ 5 cm. Right, a FSU with a disc herniation landmark (top) has noise added in a spherical ROI (middle) as well as Guassian weighted noise (bottom).