Ultrasonic image hybrid training method based on deep learning

Abstract

The invention relates to the field of medical image processing, and aims to provide an ultrasonic image hybrid training method based on deep learning. The method comprises the following steps: preparing a training set, a verification set and a test set by utilizing ultrasonic image data of different examination parts in a database; preprocessing each data set; constructing and training a convolutional neural network, adopting multi-channel output during training, but only enabling a single channel to participate in back propagation; and testing the trained convolutional neural network. According to the method, ultrasonic data sets of multiple different disease types are combined for training together, so that the convolutional neural network model can be in contact with more samples, the problems of few data set samples and single case are relieved, and the performance of the model obtained through training on a single task is improved. According to the invention, a plurality of ultrasonic tasks are completed through the same convolutional neural network, the training and deployment cost can be reduced, and the user experience is improved.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to an ultrasound hybrid training method based on deep learning. Background technique [0002] Ultrasound imaging is a non-invasive examination method. Ultrasound imaging examination has the advantages of cheap, non-destructive, repeatable, and high sensitivity, and is the preferred imaging examination method for disease screening. However, affected by the visual fatigue of medical workers and the level of diagnosis, there are many subjective factors in the results of ultrasound diagnosis, and the diagnosis process is laborious and time-consuming. [0003] Deep learning can directly process raw data (such as ultrasound images) and automatically learn mid-level and high-level abstract features from it. It can perform various automatic analysis tasks of ultrasound images, such as lesion / nodule classification, tissue segmentation and target detection, etc. Using deep learning ...

AI Technical Summary

Problems solved by technology

However, the current ultrasound data set is difficult to meet the needs of deep learning

On the one hand, the number of public data sets is often limited; on the other hand, although the hospital has a large amount of ultrasound image data, it is difficult for the outside to obtain a large amount of data from the hospital due to the "data island"

Images