【摘要】 近年来,利用有监督学习进行医疗图像的自动识别应用越来越广泛,但是在该领域中无法忽视的问题是专业医疗图像的精细标注的获取成本非常巨大。因此,多实例学习由于其将单个图像的多个实例视为一个包,在学习过程中只需要包的标签,而不需要精确地针对每个实例的标签的特性,为解决医疗图像识别缺乏精准标注的问题提供了解决方案。特别是随着近十年来深度学习的飞速发展,大量基于神经网络的多实例学习方法被提出,使得多实例学习领域焕发了新的生机。更多还原

【Abstract】 In recent years, the application of supervised learning in medical image automatic recognition is more and more widely. However, the problem that cannot be ignored in this field is that the cost of obtaining the fine annotation of professional medical image is very huge. Therefore, multi instance learning, which regards multiple instances of a single image as a bag, only needs the label of the bag in the learning process, and does not need the label for each instance accurately. It provides a solution to solve the problem of lack of accurate annotation in medical image recognition. Especially with the rapid development of deep learning in recent ten years, a large number of multi instance learning methods based on neural network have been proposed, which makes the multi instance learning field full of new vitality.更多还原