Brain image processing method and system for stroke patients

Abstract

The invention discloses a brain image processing method and system of a stroke patient. The original brain structure image of a stroke patient is obtained by using a nuclear magnetic resonance imager, and image conversion preprocessing is performed to identify each lobe sub-region of the brain, extract Fuzzy sulcus collection and brain contour repair processing and other methods, so as to obtain brain repair images with high recognition accuracy. The brain image processing method and system can correct the blurred shadow area in the original brain structure image, make the brain structure outline more clear and accurate, and improve the recognition accuracy of the brain structure outline in the MRI brain image.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a brain image processing method and system for stroke patients. Background technique [0002] Stroke is an acute cerebrovascular disease with high morbidity and mortality. It mainly includes ischemic stroke and hemorrhagic stroke. The pathogenic mechanism is brain tissue damage due to local cerebral blood vessel rupture or obstruction. With the help of medical imaging technology, real artifact images of brain tissue of stroke patients can be obtained, which can effectively provide more accurate physiological information to help doctors make initial clinical diagnosis. At present, clinical brain imaging techniques such as Computed Tomography (CT), Positron Emission Tomography (PET), Magnetic Resonance Imaging (MRI) and so on are widely used. Because MRI can obtain brain anatomy with high spatial resolution, and does not require injection of radioactive substances, it can ...

AI Technical Summary

Problems solved by technology

However, these two methods only use pixel-level features for image processing, and ignore the structural features of the brain when they are applied to brain image processing, which will lead to a certain degree of limitation in the accuracy of brain contour recognition.

However, the extraction of brain image features based on deep learning algorithms requires a large number of training set data samples, and the individual differences among stroke patients are obvious. The available medical image data cannot satisfy a large number of data samples, which is likely to cause overfitting problems. training results

In addition, due to the variability of brain atrophy or local vascular rupture in stroke patients, the brain image is small and there are many shadow areas. risk of misdiagnosis

Images