LGESynthNet: Controlled Scar Synthesis for Improved Scar Segmentation in Cardiac LGE-MRI Imaging
This work addresses the challenge of data scarcity for medical imaging researchers and clinicians by enabling synthetic data generation with fine-grained control, though it is incremental as it builds on existing diffusion and ControlNet methods.
The paper tackled the problem of limited annotated data for scar segmentation in cardiac LGE-MRI by developing LGESynthNet, a latent diffusion-based framework for controllable enhancement synthesis, which improved downstream segmentation and detection performance by up to 6 and 20 points respectively.
Segmentation of enhancement in LGE cardiac MRI is critical for diagnosing various ischemic and non-ischemic cardiomyopathies. However, creating pixel-level annotations for these images is challenging and labor-intensive, leading to limited availability of annotated data. Generative models, particularly diffusion models, offer promise for synthetic data generation, yet many rely on large training datasets and often struggle with fine-grained conditioning control, especially for small or localized features. We introduce LGESynthNet, a latent diffusion-based framework for controllable enhancement synthesis, enabling explicit control over size, location, and transmural extent. Formulated as inpainting using a ControlNet-based architecture, the model integrates: (a) a reward model for conditioning-specific supervision, (b) a captioning module for anatomically descriptive text prompts, and (c) a biomedical text encoder. Trained on just 429 images (79 patients), it produces realistic, anatomically coherent samples. A quality control filter selects outputs with high conditioning-fidelity, which when used for training augmentation, improve downstream segmentation and detection performance, by up-to 6 and 20 points respectively.