OneSide：一种新的婴儿MRI数据增强方式

MRI scan is a common examination modality for the diagnosis of brain-related diseases in clinical practice. Through MRI, doctors can observe the tissue structure of the patient\'s brain more visually and accurately, which in turn facilitates the diagnosis of diseases and the formulation of treatment plans, so being able to accurately segment the brain MRI is of great value in clinical practice. Due to the process of myelin formation and maturation, the white matter and gray matter in the brain of 6-month-old infants exhibit similar intensities in MRI, which poses a greater challenge for segmentation. Meanwhile, the traditional data enhancement approach, which does not take into account the symmetry of brain tissues to a certain extent, causes an exponential increase in the amount of data and makes the model training inefficient and ineffective. For this reason, this paper proposes the OneSide method based on the flipped data enhancement approach, which makes the model focus more on the tissue characteristics of the half-brain. We also show that this method can improve the training efficiency of the model and have better training effect than the existing flipping method.