Pancreatic Tumor Segmentation as Anomaly Detection in CT Images Using Denoising Diffusion Models

TL;DR

This work tackles pancreatic tumor detection in CT images under limited pixel-level annotations by adopting a weakly supervised anomaly-detection framework based on denoising diffusion models (DDPM/DDIM) with classifier guidance. The method trains on both healthy and diseased samples to learn to translate diseased CT images into healthy representations, enabling detailed anomaly maps without relying on segmentation masks, using forward steps $N$ and guided reverse sampling. Evaluations on the MSD pancreas dataset and a real-world OUHSC test set demonstrate that the anomaly localization is meaningful and tunable via the classifier guidance scale $S$ and noise level $N$, though Dice scores lag behind fully supervised approaches. Overall, the diffusion-based approach reduces labeling requirements and provides a viable, flexible pathway for medical anomaly detection in pancreatic imaging, with clear avenues for improving robustness and multi-modal integration in future work.

Abstract

Despite the advances in medicine, cancer has remained a formidable challenge. Particularly in the case of pancreatic tumors, characterized by their diversity and late diagnosis, early detection poses a significant challenge crucial for effective treatment. The advancement of deep learning techniques, particularly supervised algorithms, has significantly propelled pancreatic tumor detection in the medical field. However, supervised deep learning approaches necessitate extensive labeled medical images for training, yet acquiring such annotations is both limited and costly. Conversely, weakly supervised anomaly detection methods, requiring only image-level annotations, have garnered interest. Existing methodologies predominantly hinge on generative adversarial networks (GANs) or autoencoder models, which can pose complexity in training and, these models may face difficulties in accurately preserving fine image details. This research presents a novel approach to pancreatic tumor detection, employing weak supervision anomaly detection through denoising diffusion algorithms. By incorporating a deterministic iterative process of adding and removing noise along with classifier guidance, the method enables seamless translation of images between diseased and healthy subjects, resulting in detailed anomaly maps without requiring complex training protocols and segmentation masks. This study explores denoising diffusion models as a recent advancement over traditional generative models like GANs, contributing to the field of pancreatic tumor detection. Recognizing the low survival rates of pancreatic cancer, this study emphasizes the need for continued research to leverage diffusion models' efficiency in medical segmentation tasks.

Figures (4)

• Figure 1: (a) Subject's CT image and (b) its corresponding segmentation mask from the MSD dataset, after pre-processing steps
• Figure 2: Anomaly map of the subject, with different noise levels (N) and classifier guidance scales (S)
• Figure 3: Segmentation map of the subject, with different noise levels (N) and classifier guidance scales (S), and a fixed threshold of 35%
• Figure 4: The anomaly heat map results of the OUHSC test set acquired from the diffusion model accompanied by the cross marked approximate tumor locations, and the corresponding classifier confidence levels of 0.9971, 0.9960, and 0.4645 respectively.