Medical image diagnostic method

Abstract

The present invention provides a medical image diagnostic method. The method comprises the steps of: collection of medical images, medical image preprocessing, classification analysis, obtaining of aprecious medical image and comparison assessment. A double convolutional neural network is employed to perform sketching of the medical images in batches, and the medical image diagnostic method can be widely applied to a small image data set to reduce the consumption of the calculation resources and obtain very high accuracy; a FPGA (Field Programmable Gate Array) is employed to achieve the wholedesign of a medical deagnostic tool, the power consumption of the FPGA is low to meet the energy conservation and emission reduction green economy development concept; the medical images with different types of diseases, different area crowds and different medical technical devices are employed to obtain corresponding parameters of a second-level neural network model so as to greatly facilitate doctors' usage and improve the diagnosis effect; and the economy cost of the neural network calculation unit based on the FPGA is low, so that medical image diagnostic method is suitable for each department of a general hospital and is conveniently applied to township-level medical institutions.

Description

technical field [0001] The invention relates to a medical image diagnosis method, which belongs to the cross technical field of artificial intelligence, medical image processing and customized calculation. Background technique [0002] Medical images are an important tool for diagnosing diseases. However, there is still a lack of reliable quantitative analysis tools for medical images at this stage, and most medical diagnoses still rely on doctors' experience and human judgment. Therefore, an accurate and reliable method and tool for automatic quantitative analysis of medical images is urgently needed. However, to achieve automated quantitative analysis, the first thing to achieve is the contouring of organs and tissues in medical images. In this regard, the current mainstream methods include traditional contour finding methods and machine learning methods. [0003] Traditional contour drawing methods include Active Contour Selection (ACS), discrete active contour, prior e...

AI Technical Summary

Problems solved by technology

Due to the high power consumption of this type of technology, the consumption of a large amount of computing resources and high economic costs, its large-scale field deployment and application are severely limited.

Images