Cloud-based large-scale pathological image cooperation annotation method and system

Abstract

The invention provides a cloud-based large-scale pathological image cooperation annotation method and a system. The method comprises the steps of through an image preprocessing module, preprocessing apathological picture at cloud through super-pixel segmentation, and furthermore storing the boundary in a binary image; through a data recording module, performing data recording of multiple users sothat the user can perform cooperative image annotation; and through a frontend interaction module, finishing capturing of a user action and displaying of the pathological picture, and through a backend processing module, in response to a request which is transmitted from the frontend interaction module, finishing data annotation and reminding of a possible annotation area of the user. Through theeffects realized by the modules, the method and the system realize purposes of multiple-person cooperation and pixel-grade convenient annotation on the pathological picture at a webpage end.

Description

technical field [0001] The invention relates to the field of digital pathology, in particular to a cloud-based large-scale pathology image collaborative annotation method and system. Background technique [0002] At present, most of the digital pathology software mainly display and analyze the pathological images whose data has been annotated, and lack software for users to annotate data. As for software that supports user annotation data, there are only local image annotation software represented by QuPath, and online annotation software represented by The Digital Slide Archive, but both have their own advantages and disadvantages. Stand-alone labeling software based on QuPath can perform superpixel segmentation in small areas, and users can select areas and label data. However, its shortcomings are also particularly obvious. There are the following four points: (1) When marking, each computer needs to install configuration software and copy pathological pictures. Usually...

AI Technical Summary

Problems solved by technology

When selecting a large area or the whole image for superpixel segmentation, it takes a long time to wait

At the same time, the results of each superpixel segmentation cannot be saved, so each time the program is restarted, the image needs to be processed again

(3) Currently there is only a stand-alone version, which does not support simultaneous labeling by multiple users, resulting in low labeling efficiency

(4) The pictures marked with data by the software are not machine-readable and can only be opened with their own software, making it impossible to analyze the marked data with other software

Although the online annotation software represented by The Digital Slide Archive can annotate online, in addition to the fourth shortcoming mentioned above, the function is relatively single at the same time. It only has annotations similar to graffiti or custom polygon shapes, which cannot provide users with Accurate and convenient labeling

Images