Thyroid nodule recognition method based on generative adversarial network

Abstract

The invention discloses a thyroid nodule recognition method based on a generative adversarial network. The thyroid nodule recognition method comprises the following steps: screening the data of a patient suffering from a thyroid disease, building a thyroid nodule database, and carrying out circling marking on thyroid nodules and thyroid parenchyma; semantic segmentation: taking U-net as a condition of a generative network to generate an adversarial network model, and inputting an annotated image to realize semantic segmentation; benign and malignant classification input: after multiple convolution, activation and pooling, using a full connection layer to integrate the extracted features to realize image benign and malignant judgment by the convolutional neural network; after the lesion isinput into the convolutional neural network, automatically classifying the lesion and outputting a benign and malignant judgment result of the lesion by the convolutional neural network; and giving areference diagnosis report of the illness state of the patient according to the training result. The thyroid nodule recognition method based on a generative adversarial network can improve thyroid nodule benign and malignant judgment accuracy of doctors, can reduce the thyroid screening time of ultrasound, can reduce the working intensity of medical staff, and can increase the satisfaction degreeof patients.

Description

technical field [0001] The invention relates to the field of image processing and medical aided diagnosis, in particular to a method for identifying thyroid nodules based on a generative confrontation network. Background technique [0002] The thyroid is located in the lower part of the neck and its function is to produce hormones that provide energy to the body's cells. Currently, the incidence of thyroid cancer is on the rise. Although the incidence of thyroid cancer is very high, in more than 50% of adults, only about 7% of most thyroid nodules are malignant and the diagnosis methods of benign and malignant nodules are very different. It plays an important clinical guiding significance for the diagnosis and treatment of thyroid diseases. [0003] Ultrasound (US) technology is a non-invasive, portable and safe imaging mode, widely used in the diagnosis and follow-up of thyroid nodules. In the United States, thyroid ultrasonography is already the best diagnostic tool for...

AI Technical Summary

Problems solved by technology

The whole ultrasonic diagnostic process depends entirely on the visual evaluation of the first-time doctor, and the entire diagnostic process lacks repeatability. Therefore, how to quantify the ultrasonic data so that doctors can qualitatively and quantitatively analyze the characteristics of thyroid nodules is the key to solving the problem.

Images