Re-DiffiNet: Modeling discrepancies in tumor segmentation using diffusion models
This work addresses the challenge of insufficient accuracy and generalizability in deep learning-based tumor segmentation for clinical applications, representing an incremental advancement.
The paper tackled the problem of improving tumor segmentation accuracy for glioblastoma patients by modeling discrepancies between segmentation model outputs and ground truth using diffusion models, resulting in an average improvement of 0.55% in Dice score and 16.28% in HD95 over a state-of-the-art U-Net model.
Identification of tumor margins is essential for surgical decision-making for glioblastoma patients and provides reliable assistance for neurosurgeons. Despite improvements in deep learning architectures for tumor segmentation over the years, creating a fully autonomous system suitable for clinical floors remains a formidable challenge because the model predictions have not yet reached the desired level of accuracy and generalizability for clinical applications. Generative modeling techniques have seen significant improvements in recent times. Specifically, Generative Adversarial Networks (GANs) and Denoising-diffusion-based models (DDPMs) have been used to generate higher-quality images with fewer artifacts and finer attributes. In this work, we introduce a framework called Re-Diffinet for modeling the discrepancy between the outputs of a segmentation model like U-Net and the ground truth, using DDPMs. By explicitly modeling the discrepancy, the results show an average improvement of 0.55\% in the Dice score and 16.28\% in HD95 from cross-validation over 5-folds, compared to the state-of-the-art U-Net segmentation model.