SynSeg-Net: Synthetic Segmentation Without Target Modality Ground Truth
This addresses the need for generalizable medical image segmentation across modalities, reducing manual annotation efforts, but it is incremental as it builds on existing CycleGAN and DCNN techniques.
The paper tackles the problem of training segmentation networks for new imaging modalities without manual labels by proposing SynSeg-Net, which uses unpaired images and source modality labels, achieving comparable performance to methods using target labels in some cases.
A key limitation of deep convolutional neural networks (DCNN) based image segmentation methods is the lack of generalizability. Manually traced training images are typically required when segmenting organs in a new imaging modality or from distinct disease cohort. The manual efforts can be alleviated if the manually traced images in one imaging modality (e.g., MRI) are able to train a segmentation network for another imaging modality (e.g., CT). In this paper, we propose an end-to-end synthetic segmentation network (SynSeg-Net) to train a segmentation network for a target imaging modality without having manual labels. SynSeg-Net is trained by using (1) unpaired intensity images from source and target modalities, and (2) manual labels only from source modality. SynSeg-Net is enabled by the recent advances of cycle generative adversarial networks (CycleGAN) and DCNN. We evaluate the performance of the SynSeg-Net on two experiments: (1) MRI to CT splenomegaly synthetic segmentation for abdominal images, and (2) CT to MRI total intracranial volume synthetic segmentation (TICV) for brain images. The proposed end-to-end approach achieved superior performance to two stage methods. Moreover, the SynSeg-Net achieved comparable performance to the traditional segmentation network using target modality labels in certain scenarios. The source code of SynSeg-Net is publicly available (https://github.com/MASILab/SynSeg-Net).