System and method for automatically screening and labeling helicobacter pylori

Abstract

The invention provides a system and method for automatically screening and labeling helicobacter pylori, and the method mainly comprises the following steps: (1) selecting gastric tissue slices with expert diagnosis information, including helicobacter pylori positive and negative slices; (2) scanning and converting conventional slice into a digital pathological slices by using a slice scanner to serve as a digital pathological image database; (3) enabling a pathologist to mark the helicobacter pylori in the digital slices and take the marked helicobacter pylori as an helicobacter pylori database; (4), analyzing and learning the digital pathological image database and the helicobacter pylori database by using a deep learning algorithm and an image processing method, and establishing a digital pathological image automatic diagnosis model; (5) diagnosing a digital pathological slice of an unknown lesion by using the digital pathological image automatic diagnosis model established in the step (4), automatically labeling and identifying the helicobacter pylori, and feeding back the labeled diagnosis result to the pathologist to assist the pathologist in diagnosis.

Description

technical field [0001] The invention relates to the application of deep learning in medical diagnosis, in particular to a system and method for automatic screening and labeling of Helicobacter pylori. Background technique [0002] Deep learning is a new field in machine learning research. It establishes and simulates a neural network for analysis and learning of the human brain, that is, establishes a convolutional neural network to automatically extract features from externally input data, so that the machine can understand the learning data. , get information and output. Deep learning algorithms have been applied to intelligent recognition of images, sounds, and texts. Medical pathology mainly relies on tissue images to make a diagnosis, and the application of deep learning in pathology will promote the development of pathological diagnosis. [0003] Helicobacter pylori (H. pylori) is a microaerophilic Gram-negative bacillus isolated from gastric mucosal biopsy tissue. ...

AI Technical Summary

Problems solved by technology

[0008] Risk of missed detection: When the amount of H. pylori infected is very small, because H. pylori is small and similar to tissue staining, it needs to be viewed under each high-power field of view, resulting in some easy missed detection

[0009] Poor diagnostic consistency: The diagnostic results are affected by subjective factors such as the pathologist's own experience and working status, so the result consistency is poor

[0010] Difficulty in follow-up visits: When using conventional glass sections for diagnosis, they cannot be marked and preserved; when the amount of H. pylori is small, it is difficult for other doctors to find H. pylori for follow-up visits

[0011] Uneven distribution of pathologist resources: pathologists need long-term professional training and practice, and the training cycle is long; experts are mostly in well-known tertiary hospitals, and there is a shortage of pathologists in county and town hospitals

Images