METGAN: Generative Tumour Inpainting and Modality Synthesis in Light Sheet Microscopy
This work addresses the lack of annotations in preclinical research using light sheet microscopy, enabling better tumor analysis through improved synthetic data generation.
The paper tackles the problem of insufficient annotated data for deep learning in multimodal imaging by introducing METGAN, a generative method that uses real anatomical information to create realistic image-label pairs of tumors, resulting in significant quantitative improvements over existing methods and substantially enhancing segmentation performance when used for data augmentation.
Novel multimodal imaging methods are capable of generating extensive, super high resolution datasets for preclinical research. Yet, a massive lack of annotations prevents the broad use of deep learning to analyze such data. So far, existing generative models fail to mitigate this problem because of frequent labeling errors. In this paper, we introduce a novel generative method which leverages real anatomical information to generate realistic image-label pairs of tumours. We construct a dual-pathway generator, for the anatomical image and label, trained in a cycle-consistent setup, constrained by an independent, pretrained segmentor. The generated images yield significant quantitative improvement compared to existing methods. To validate the quality of synthesis, we train segmentation networks on a dataset augmented with the synthetic data, substantially improving the segmentation over baseline.