Hash sample balance cancer labeling method for histopathologic image

Abstract

The invention discloses a Hash sample balance cancer labeling method for histopathologic images, which comprises the following steps: extracting features of various types of images by using a convolutional auto-encoder, extracting image block features by using an unsupervised convolutional auto-encoder CAE, and using a supervised convolutional neural network (CNN) for a final classification task; mapping an image in a high-dimensional feature space to a low-dimensional binary space by using a Hash method, forming Hash codes for all multi-class image samples, and wherein each Hash code corresponds to a subspace in an original feature space, also called a Hash bucket; and calculating the selection proportion of the extracted samples in each hash bucket, and selecting representative samples. The invention provides an efficient resampling method HBU for the class imbalance problem in a large-scale pathological histological image classification task. The pairwise distance between the samples does not need to be calculated, and the invention has high efficiency and high expansibility.

Description

technical field [0001] The invention relates to the technical field of image analysis, in particular to a hash sample balanced cancer labeling method for histopathological images. Background technique [0002] Early diagnosis of cancer based on ultra-high resolution pathological images of patients plays an important role in medicine. Existing methods mainly divide the original histopathological image into a large number of image blocks, and then judge whether the image blocks are cancer tissue images, so as to realize accurate judgment and location of cancer lesions. However, the number of normal tissue image blocks in reality often far exceeds that of cancer tissue block images, resulting in an imbalance problem in the dataset, making it difficult to effectively train a cancer tissue image classifier. Therefore, this paper proposes a hash sample balanced cancer labeling method for histopathological images. Based on the hash method, a balanced training set is extracted to t...

AI Technical Summary

Problems solved by technology

However, when the imbalance rate is high, the ensemble method needs to train enough base classifiers to achieve better results. At the same time, multiple models also bring greater time overhead and machine cost when deployed.

Another reason for the slowness of ensemble methods is the use of distance-based resampling methods in each iteration to obtain balanced data

Images