CT lumbar spine automatic locating method and device and storage medium

Abstract

The embodiment of the invention discloses a CT lumbar spine automatic locating method and device and a storage medium The method comprises the following steps of: the training step: acquiring the initial CT stereoscopic image of the spine; processing the initial CT stereoscopic image of the spine so as to obtain the locating parameters of the lumbar spine position; the prediction step: acquiring the CT stereoscopic image of the spine to be measured; performing projection processing on the CT stereoscopic image of the spine to be measured by using the digital reconstruction radiography and thegeometric relationship so as to obtain the DRR image to be measured; predicting the DRR image to be measured by using the convolutional neural network and locating parameters so as to obtain the position information of the lumbar spine in the DRR image to be measured; and mapping the position information of the lumbar spine in the DRR image to be measured to the CT stereoscopic image of the spineto be measured by using the digital reconstruction radiography technology so as to obtain the position information of the lumbar spine in the CT stereoscopic image of the spine to be measured. With application of the CT lumbar spine automatic locating method and device and storage medium, the robustness, the real-time performance and the accuracy of lumbar spine automatic locating can be enhanced.

Description

technical field [0001] The invention relates to the technical field of image processing of applied medical images, in particular to a method, device and storage medium for automatic positioning of lumbar spine in CT. Background technique [0002] Medical CT (Computed Tomography) uses X-rays to scan the human body inspection site layer by layer with a certain thickness to obtain a series of images. Clinical application Through CT observation of bone tissue, liver and other tissues for pathological analysis. As a relatively obvious and fixed tissue in the human body, the spine can be used as the best reference point during the operation. Therefore, the fast and accurate positioning of the spine in the CT volume data becomes the key to the automatic analysis and diagnosis of liver pathology. At present, automatic lumbar spine positioning can be divided into three types: 1. Manual positioning; 2. Spine positioning based on CT volume data based on traditional image processing; ...

AI Technical Summary

Problems solved by technology

Manual positioning through software interaction is time-consuming and labor-intensive, increasing the workload of doctors

Due to the individual differences of the human body, traditional image processing uses technologies such as binarization and simple data statistics to realize positioning, and its robustness needs to be improved; To locate it, the feature selection of this method has a decisive impact on the accuracy of position regression. However, selecting one or more features cannot fully describe the CT volume data, and its robustness needs to be improved. The amount of volume data is large, and the speed of positioning needs to be improved

Images