Lightweight echocardiogram standard section identification method and device and medium

Abstract

The invention relates to a lightweight echocardiogram standard section recognition method and device and a medium, and the method comprises the following steps: S1, collecting data of a two-dimensional echocardiogram standard section, and forming a data set; s2, classifying the data set, and dividing the data set into a training set, a test set and a verification set; s3, carrying out preprocessing and data enhancement on the data set; s4, building a deep learning network model as a training model, and training the training model by using the training set; s5, verifying the training model by using the verification set, adjusting hyper-parameters of the training model according to an evaluation result, and testing the generalization ability of the training model by using the test set to obtain an echocardiogram classification model; and S6, inputting the echocardiogram into the trained echocardiogram classification model to obtain an echocardiogram classification recognition result. According to the method, model parameters are reduced on the basis of lightweight network design in a Mobilene-Echo network structure. Belongs to the technical field of detection.

Description

technical field [0001] The invention relates to the technical field of detection, in particular to a method, a device and a medium for identifying a standard slice of a lightweight echocardiogram. Background technique [0002] At present, heart disease not only has a significant impact on the quality of human life, but also threatens human health. In this regard, early diagnosis and diagnosis and treatment of heart disease are very important. In recent years, with the continuous progress of computer technology and artificial intelligence medical imaging technology, artificial intelligence medical imaging has gradually developed from an auxiliary examination method to the most important clinical diagnosis and differential diagnosis method in modern medicine. [0003] Echocardiography can visually display and quantitatively measure important information on cardiac anatomy, function and hemodynamics, so as to reflect the configurational characteristics of the heart under physi...

AI Technical Summary

Problems solved by technology

[0007] The above-mentioned existing image recognition technology has not been developed for embedded devices, and cannot solve the challenges brought about by the large number of model parameters and the long recognition time in this scenario.

The first patent uses DenseNet101 to classify the standard slices of echocardiography. This network has a large amount of calculation and has high requirements for equipment performance.

The second patent uses FPN (image pyramid) to extract the features of each layer of the image, but the feature extraction of each layer has obvious limitations, the inference time will rise sharply, and the memory usage of end-to-end training on FPN is relatively large

[0008] Echocardiography standard slice recognition based on deep learning will encounter the following challenges: the existing classification model has too many parameters, resulting in too slow running speed in embedded devices, which cannot well meet the real-time requirements

Images