Medical image processing system, model training method and training device

Abstract

The invention provides a medical image processing system, a model training method and a training device. The medical image processing system includes an image acquisition unit, a virtual DWI image generation unit, and a display unit. After acquiring the CTA image obtained by CTA scanning, the CTA image is converted into a corresponding virtual DWI image through a deep learning model. DWI images are used for stroke diagnosis. Compared with CTA images, virtual DWI images can provide more accurate and comprehensive display of stroke information, so it greatly facilitates doctors to diagnose early stroke, and it is helpful for the control and treatment of stroke. It provides a reliable and comprehensive diagnosis basis, reduces the rate of missed diagnosis and misdiagnosis, and allows patients to receive timely intervention and treatment. The fusion of inspection methods can also save medical expenses. The deep learning model in the medical image processing system can be trained through the model training method and device of the present invention.

Description

technical field [0001] The invention relates to the technical field of medical equipment, in particular to a medical image processing system, a model training method and a training device. Background technique [0002] Stroke is a major disease that seriously threatens the health of the population and hinders social and economic development. Among them, acute ischemic stroke (Acute Ischemic Stroke, AIS) accounts for 60%-80% of strokes. Due to the limited diagnostic methods, the early misdiagnosis rate of AIS is extremely high, so it is very important for its early diagnosis and determination of infarct size. Important, it is the decisive factor for timely intervention and treatment to reduce disability and mortality. [0003] Diffusion-Weighted Image (DWI) obtained by magnetic resonance imaging (MRI) is currently the most sensitive and specific diagnostic method for early stroke, with high signal-to-noise ratio, diagnostic specificity and The sensitivity can reach 100%, an...

AI Technical Summary

Problems solved by technology

[0003] Diffusion-Weighted Image (DWI) obtained by magnetic resonance imaging (MRI) is currently the most sensitive and specific diagnostic method for early stroke, with high signal-to-noise ratio, diagnostic specificity and The sensitivity can reach 100%, and it can accurately judge the extent and degree of ischemic brain tissue. However, MRI imaging takes a long time, requires a high degree of patient cooperation, and has a strong magnetic field that is not convenient for emergency patients. Therefore, it is a non-emergency examination. project

In addition, the waiting period for MRI examinations in large hospitals is generally longer and the cost is higher, while small hospitals are often not equipped with MRI equipment. Poor sex

[0004] CTA (CT angiography, CT angiography), also known as non-invasive angiography, has a fast detection process and is currently a feasible early diagnosis method for stroke. The infarction caused by blood vessels and small blood vessels is not sensitive enough, and the specific extent of the infarcted brain tissue cannot be determined. The diagnosis of stroke based on the results of CTA examination is limited, and there are many missed diagnoses and misdiagnoses, especially for inexperienced and junior doctors. In other words, the probability of missed diagnosis is higher, but the night shift radiologists in the clinical emergency department are mainly junior doctors

Images