Intelligent pulmonary tuberculosis detection method and system based on images and clinical data

Abstract

The invention discloses an intelligent pulmonary tuberculosis detection method and system based on images and clinical data. The method comprises the following steps: acquiring a medical image and clinical data of a to-be-processed target; the clinical data comprises medical history data and biochemical data; encoding the medical history data and the biochemical data to obtain a medical history data vector and a biochemical data vector; performing multi-modal data coupling on the medical image, the medical history data vector and the biochemical data vector to obtain a multi-dimensional feature map sequence group; inputting the multi-dimensional feature map sequence group into a self-attention network, and superposing the output of the self-attention network to obtain feature data; and according to the feature data, obtaining a pulmonary tuberculosis detection result of the to-be-processed target. According to the method, the features of the image data and the clinical data are well fused, the method is used for achieving intelligent detection of the pulmonary tuberculosis, the detection precision of the pulmonary tuberculosis can be remarkably improved, diagnosis work can be assisted to be completed, and the workload of doctors is relieved. The method can be widely applied to the technical field of medical image processing.

Description

technical field [0001] The present application relates to the technical field of medical image processing, in particular to an intelligent detection method and system for pulmonary tuberculosis based on images and clinical data. Background technique [0002] Pulmonary tuberculosis is a chronic infectious disease transmitted through the respiratory tract. Chest X-ray (CXR) imaging is a non-invasive medical imaging method for clinical first-line screening of pulmonary tuberculosis. In clinical practice, CXR-based tuberculosis screening mainly relies on clinicians to read images to judge the image characteristics of lesions. The accuracy of discrimination is highly dependent on doctors' subjective experience and image quality. Large workload, increased working hours, and discrepancies in the reading results of doctors with different qualifications. The above problems are mainly caused by the following two points. First, the CXR image is a single 2-dimensional image. The image ...

AI Technical Summary

Problems solved by technology

[0003] However, there are still the following problems that need to be solved urgently in the application of artificial intelligence technology in the clinical screening of pulmonary tuberculosis in related technologies: (1) In terms of the acquisition and utilization of active pulmonary tuberculosis CXR data, it is still difficult to use artificial intelligence technology to distinguish with high accuracy Pulmonary tuberculosis and other pulmonary lesions; (2) It is impossible to simulate the process of doctors' diagnosis and reasoning based on information features in the clinical diagnosis process. The above is the bottleneck of the current application of artificial intelligence technology in pulmonary tuberculosis screening, which is not conducive to the realization of artificial intelligence technology embedded in clinical practice process

Images