Validation of the algorithm output.

Two WISH experts marked each of 200 processed images as good, intermediate or bad (1.0, 0.5 or 0.0) on three potentially subjective qualities: whether the embryo outline had been captured correctly, whether the stained regions were delineated correctly, and whether the pigmented regions were delineated correctly. For each quality we correlated the experts’ scores over all 200 images.两位WISH专家针对三项具有潜在主观性的质量维度，为200张经预处理的图像进行评分，将图像分为合格、中等、不合格三个等级，对应分值分别为1.0、0.5和0.0。三项质量维度具体为：胚胎轮廓是否被正确捕捉、染色区域是否被正确勾勒，以及色素沉着区域是否被正确勾勒。针对每一项质量维度，我们对全部200张图像的专家评分开展了相关性分析。