Diffusion-Based Approaches in Medical Image Generation and Analysis

TL;DR

The paper tackles data scarcity in medical image analysis by leveraging diffusion models to generate synthetic, realistic images for training CNNs across three domains: Brain Tumor MRI, ALL, and SARS-CoV-2 CT scans. It trains diffusion models to produce labeled synthetic data and trains eight CNN architectures on this data, evaluating on unseen real data and applying LIME to inspect decision regions. The results show competitive performance across datasets (e.g., up to 91.38% accuracy on ALL) and demonstrate that diffusion-generated data can help overcome limited patient data while providing interpretable insights through XAI. This approach offers a practical pathway to reduce dependence on patient-derived data while maintaining performance and interpretability in medical image analysis.

Abstract

Data scarcity in medical imaging poses significant challenges due to privacy concerns. Diffusion models, a recent generative modeling technique, offer a potential solution by generating synthetic and realistic data. However, questions remain about the performance of convolutional neural network (CNN) models on original and synthetic datasets. If diffusion-generated samples can help CNN models perform comparably to those trained on original datasets, reliance on patient-specific data for training CNNs might be reduced. In this study, we investigated the effectiveness of diffusion models for generating synthetic medical images to train CNNs in three domains: Brain Tumor MRI, Acute Lymphoblastic Leukemia (ALL), and SARS-CoV-2 CT scans. A diffusion model was trained to generate synthetic datasets for each domain. Pre-trained CNN architectures were then trained on these synthetic datasets and evaluated on unseen real data. All three datasets achieved promising classification performance using CNNs trained on synthetic data. Local Interpretable Model-Agnostic Explanations (LIME) analysis revealed that the models focused on relevant image features for classification. This study demonstrates the potential of diffusion models to generate synthetic medical images for training CNNs in medical image analysis.

Figures (7)

• Figure 1: This figure showcases different generative models and provides an overview of their underlying principles kazerouni2023diffusion
• Figure 2: Generative learning trilemma kazerouni2023diffusion
• Figure 3: Architecture of Diffusion Model rink2021memory
• Figure 4: Flow diagram of our proposed methodology.
• Figure 5: Sample images from the datasets a. Brain Tumor MRI b. Acute Lymphoblastic Leukemia (ALL) c. SARS-CoV-2 CT-Scans