CNN-based periocular organ segmentation method and device, and storage medium

Abstract

The invention belongs to the technical field of medical images and computers, and relates to a convolutional neural network-based periocular organ segmentation method and device, and a storage medium,and the method comprises the following steps: inputting a to-be-segmented medical image into a trained convolutional neural binary classification network, and obtaining a cross section containing eyes in the medical image; positioning an approximate eye area in the cross section of the eye; finding out the center of the head in the cross section of the eye after segmenting the head; according tothe anatomical size of the human body, positioning an approximate eye area through the center of the skull; segmenting an eye delineation eyeball at the positioned eye approximate position by using the trained convolutional neural network for delineating the eyeball; respectively positioning crystalline lenses, optic nerves and pituitary gland according to the positions of the eyes and the anatomical structure of the human body; and respectively drawing a crystalline lens, an optic nerve and a pituitary gland through corresponding convolutional neural networks.

Description

technical field [0001] The invention belongs to the field of medical imaging and computer technology, and relates to a convolutional neural network (CNN)-based segmentation method, device and storage medium for periocular organs (including eyes, lenses, optic nerves, and pituitary glands). Background technique [0002] The delineation of OARs (organs at risk) plays a key role in radiotherapy planning, but now OARs delineation is mostly done manually by doctors, assisted by registration. For example, it takes about 3 to 4 hours to outline each set of CT, and may need to be revised, which not only delays the treatment time of the patient, but also increases the workload of the doctor. [0003] At present, doctors use registration to assist in delineating OARs. However, the registration algorithm is not only time-consuming, but also unstable, and requires a lot of modifications by doctors. With the development of artificial intelligence, deep learning is gradually applied to t...

AI Technical Summary

Problems solved by technology

The entire segmentation process requires multiple binary classification networks, the segmentation effect is not good, and the eye organs need to be post-processed through the conditional random field, so this method is not only cumbersome, but also poor in robustness

Images