Pulmonary nodule determining method and device and realization device

Abstract

The invention provides a pulmonary nodule determining method and device and a realization device. The method comprises that a CT image and position information of a suspected area in the CT image areobtained; the CT image is enhanced; according to the position information, image data of the suspected area is extracted from the enhanced CT image; the image data of the suspected area is normalized;and the normalized image data is input to a pre-established pulmonary nodule determining model which is trained via a 3D convolutional neural network, and whether the suspected area refers to a pulmonary nodule is determined according to an output result. According to the invention, the distinguishing characteristic between the pulmonary nodule and non-nodules can be amplified by image enhancement and normalization, the 3D convolutional neural network can be used determine whether the suspected area refers to the pulmonary nodule, and the pulmonary nodule can be identified more accurately.

Description

technical field [0001] The present invention relates to the technical field of medical imaging, in particular to a determination method, device and implementation device for pulmonary nodules. Background technique [0002] Pulmonary nodule is a multi-system multi-organ granulomatous disease of unknown etiology. It often invades the lungs, bilateral hilar lymph nodes, eyes, skin and other organs, and the chest invasion rate is as high as 80% to 90%. Existing medical screening methods usually use naked eye recognition, image recognition, or neural network recognition, etc., but these methods are less accurate, and the suspected areas screened out may be pulmonary nodules, blood vessels, inflammatory lesions, etc. . [0003] Aiming at the problem of low accuracy of the identification result of the above-mentioned existing identification methods of pulmonary nodules, no effective solution has been proposed yet. Contents of the invention [0004] In view of this, the object o...

AI Technical Summary

Problems solved by technology

Existing medical screening methods usually use naked eye recognition, image recognition, or neural network recognition, etc., but these methods are less accurate, and the suspected areas screened out may be pulmonary nodules, blood vessels, inflammatory lesions, etc.

[0003] Aiming at the problem of low accuracy of the identification results of the above-mentioned existing identification methods of pulmonary nodules, no effective solution has been proposed

Images