Editorial: Weakly supervised deep learning-based methods for brain image analysis.

Abstract

elderly. Early and accurate diagnoses are significant for effective prevention and treatment of AD. Li et al. proposed to build a brain network for each subject for AD prediction and analysis, which assembled several commonly used neuroimaging data simply and reasonably, including structural MRI, diffusion-weighted imaging (DWI), and amyloid positron emission tomography (PET). Accurate labeling is essential for supervised deep learning methods. However, it is almost impossible to accurately and manually annotate thousands of images, which results in many labeling errors for most datasets. Liu et al. proposed a local label point correction (LLPC) method to improve annotation quality for edge detection and image segmentation tasks. U-Net is widely used in medical image segmentation in recent years. Lu et al. analyzed the effects of different parts of the U-Net in the experiment of image segmentation, and proposed a more efficient architecture, called Half-UNet. Shi et al. proposed a deep learning network with subregion partition for predicting metastatic origins and EGFR/HER2 status in patients with brain metastasis.In conclusion, these 10 papers included in this Research Topic provide new ideas on how to effectively use limited data to build reliable machine learning models for brain image analysis.