Fig. 4: Qualitative results of the missing-modality MRI brain experiment.
From: Mining multi-center heterogeneous medical data with distributed synthetic learning
a Three examples of real brain images from three data centers respectively (Real-TCIA, Real-CBICA, Real-OTHER), and the corresponding synthetic images from our method (Heteo-MM-DSL). The red dash boxes indicate the missing modality and the yellow solid boxes indicate the completed synthetic image generated by DSL. b Two examples of segmentation results for different methods vs the ground truth Label. The segmentation model learned from Hetero-MM-DSL’s synthetic data obtains more accurate results than Hetero-Real-FedSeg and does not have a clear performance drop compared to the complete-modality model (MM-DSL).
