An Automatic Detection and Correction Method of Head Tilt in CT Image

Abstract

The invention belongs to the technical field of radiotherapy, and relates to an automatic detection and correction method, device and storage medium for head tilt in CT images. The method includes the following steps: performing binarization operation on the whole body CT image; searching for the maximum connected region, deleting all non-maximally connected regions, and removing the non-main body parts in each layer of CT images, and only keeping the main body parts; judging shoulder The junction layer between the head and the neck; determine the best jaw bone layer; perform ellipse fitting on the jaw bone area in the best jaw bone layer; record the centroid of the jaw bone and the centroid of the fitting ellipse; determine the body position and implement head correction . The invention can quickly and effectively detect the inclination and orientation of the head in the CT image, and then provide accurate parameter information for the inclination correction of the head image; it has the advantages of low computational complexity, short time consumption, high noise resistance, and good data adaptation advantages; this method can also be extended to correct human body tilt in different scenarios, and has good scalability.

Description

technical field [0001] The invention belongs to the technical field of radiotherapy, and relates to an automatic detection and correction method, device and storage medium for head tilt in CT images. Background technique [0002] Automated computed tomography (CT) medical image analysis and diagnosis technology usually has certain requirements for the image morphology to be analyzed. Taking the algorithm for automatically detecting eyes in head CT images as an example, in order to reduce the space search time and error probability of the algorithm, it is usually assumed that the scanned object is in a specific scanning position (such as a supine position). However, in clinical practice, the scanned object often participates in the scan with the head tilted due to various reasons (such as coma, doctor's order, etc.). The resulting CT images obtained from different scanning body positions will greatly increase the error probability of the image analysis algorithm. In order t...

AI Technical Summary

Problems solved by technology

[0003] At present, the method of tilt correction for medical images mainly serves medical image registration. When it is applied to the tilt detection and correction of the head in CT images, the existing methods based on image geometric moment, image inertia matrix, and singular value decomposition Other medical image tilt correction methods have problems such as high computational complexity, long time consumption, poor noise immunity, and poor data adaptability, and their performance and robustness in practical applications are not high.

Images