Method and device for fetal brain age estimation and abnormal detection based on deep learning

Abstract

The invention discloses a method and device for estimating fetal brain age and abnormality detection based on deep learning. In the method of brain age estimation and abnormality detection, firstly, a data set of normal fetal brain T2-weighted magnetic resonance images is established by using T2-weighted images of pregnant women collected routinely in clinic. Secondly, the U-shaped network is used to segment the fetal brain from the uterus, and then the deep residual network based on the attention mechanism is used to predict the fetal brain age, and the uncertainty of the brain age and the reliability of the fetal brain age estimation are generated. Finally, a classifier is constructed based on the differences between actual gestational age and predicted brain age, uncertainty, and reliability to determine whether fetal brain development is abnormal. The present invention can estimate the age of the fetal brain at the same time, and generate indicators such as uncertainty and estimation reliability to detect fetuses with abnormal brain development, and has high accuracy and precision as well as high clinical application prospects and value .

Description

Technical field[0001]This application relates to the field of brain magnetic resonance image processing, in particular to the application of deep learning and brain segmentation and brain age estimation.Background technique[0002]The age of the brain based on magnetic resonance neuroimaging is widely used to describe the development process of the normal brain. The degree of deviation from the normal brain development trajectory can be used as a sign and indicator of abnormal brain. Research in the past ten years has shown that the predicted age difference (PAD) can measure the abnormal development of the brain of premature children, the degree of brain atrophy in patients with Alzheimer’s disease and traumatic brain injury, and schizophrenia. The degree of accelerated aging of the patient. Fetal brain imaging has gradually become an important tool to evaluate the normal development of the early brain. However, brain age prediction methods have not been applied to fetal neuroimaging....

AI Technical Summary

Problems solved by technology

However, the use of PAD as a single indicator in the brain age model to detect brain abnormalities has obvious shortcomings. On the one hand, PAD is a regression residual, and the change of the model will directly affect the size of PAD, resulting in inconsistent calculation results of different models; on the other hand, The brain age prediction model is basically based on the data of the normal population, while the abnormally developed brain contains characteristics that are not in the normal development of the brain, which makes it difficult to detect using a single indicator of PAD and reduces the explainability of the PAD mechanism

Images