HRCT image ossicle automatic segmentation method based on deep learning

Abstract

The invention discloses an HRCT image ossicle automatic segmentation method based on deep learning. The method comprises the following steps: S1, screening n cases of normal person ossicle chain HRCT image data, and carrying out the region labeling on an image; S2, obtaining an HRCT image of a patient in the detection process, and preprocessing the image; S3, inputting the preprocessed ossicular HRCT images at the left and right positions into an improved U-net model; and S4, performing slice segmentation on the ossicular HRCT image by using a U-net model, and performing test and re-modeling on image data after slice segmentation. According to the method, full-automatic ossicle segmentation can be completed on the HRCT image data of the individual patient by applying a deep learning technology, post-processing reconstruction of the ossicle is achieved, manual operation is not needed in the whole process, disease diagnosis is assisted, the working efficiency is improved, misdiagnosis and missed diagnosis can be reduced, and the method has the advantages of being high in working efficiency and capable of effectively reducing misdiagnosis and missed diagnosis.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to an automatic segmentation method of auditory ossicles in HRCT images based on deep learning. Background technique [0002] Head CT examination is a routine screening method for head diseases, in which the examination of the ossicles is particularly important on the CT image; at present, it is necessary to use the doctor's diagnostic experience to locate and dig out the position of the ossicles in the image; because the ossicles have signs Small features, the inspection process is time-consuming and laborious, which brings great pressure and challenges to doctors in the context of the current universal medical care; [0003] Hearing impairment is a common disease, and ossicular chain disease is one of the common causes. Changes in the shape, position, and density of the auditory ossicle structure may lead to different degrees of hearing impairment, which brings in...

AI Technical Summary

Problems solved by technology

[0004] The parts of the ossicular chain in conventional transverse CT images are not on the same plane, and the structure of the ossicular chain is so small that it cannot show the complete structure of the ossicular chain, so it is difficult to make an accurate diagnosis, and post-processing reconstruction such as CPR and MPR is often required to display the whole picture , this process is extremely dependent on the experience of radiologists, and it is also very time-consuming. Most general hospitals lack ENT radiologists, resulting in low work efficiency and even missed or misdiagnosed patients.

Images