Pulmonary nodule labeling system

Abstract

The invention discloses a Pulmonary nodule labeling system comprising the following parts: an import module used for importing a CT image; a labeling module used for labeling the CT image and obtaining a first CT image; the labeling module comprises an image display unit, a drawing unit, a data forming unit, and a label filling unit; the image display unit is used for displaying each slice image of the CT image; the drawing unit is used for drawing a Pulmonary nodule frame on the slice image; the data forming unit is used for forming size data of the Pulmonary nodule frame, and forming a framenumber; the label filling module is used for displaying the frame number, and allowing a labeler to fill in the Pulmonary nodule information according to the frame number; the system also comprises adatabase used for storing data labels corresponding to the first CT image, and an export module used for exporting the first CT image and corresponding data labels. The Pulmonary nodule marking system can accurately and effectively label the Pulmonary nodule on the CT image, thus providing accurate training samples for a Pulmonary nodule detection system.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a pulmonary nodule labeling system. Background technique [0002] Lung cancer is one of the malignant tumors with the fastest-growing morbidity and mortality and the greatest threat to the health and life of the population. Early imaging of lung cancer manifests as pulmonary nodules, and its inspection method mainly relies on chest CT tomography, and each inspection will have as many as hundreds of tomographic images. According to statistics, there are hundreds of millions of patients with pulmonary nodules in my country, and for each case of CT images, the accuracy of manual reading is about 50%-70%, which largely depends on the professionalism of doctors; It usually takes about a week to issue a report after reading the film, which takes a long time. [0003] In recent years, AI-based pulmonary nodule detection systems have gained widespread attention. However, in or...

AI Technical Summary

Problems solved by technology

However, this tool is not suitable for annotating CT data in DICOM format

CT image data is sequence data, and each case contains 60-500 tomographic images, so it is difficult to label with LableMe; the other is the ITK-SNAP tool, but this tool is mainly used to separate and extract regions of interest as images, The specific coordinate values ​​of the region of interest cannot be provided and saved, therefore, it cannot be used to label pulmonary nodules on CT images

Images