| Popis: |
The incidence of thyroid nodules is increasing year by year. Accurate determination of benign and malignant nodules is an important basis for formulating treatment plans. Ultrasonography is the most widely used in the diagnosis of benign and malignant nodules, but manual diagnosis is highly subjective, and the rate of missed diagnosis and misdiagnosis is high. To improve the accuracy of clinical diagnosis, this paper proposes a new diagnostic model based on deep learning. The diagnostic model adopts the diagnosis strategy of Localization-Classification. First, the distribution law of nodule size and nodule aspect ratio is obtained through data statistics, the multi-scale localization network structure is a priori designed, and the nodule aspect ratio is obtained from the positioning results. Then, the uncropped ultrasound image and the nodule area image are correspondingly input into the two-way classification network, and the improved attention mechanism is used to enhance the feature extraction, and finally, the deep features, the shallow features, and the nodule aspect ratio are fused, input the fully connected layer to complete the classification of benign and malignant nodules. The experimental data set is 4021 ultrasound images, each image is marked under the guidance of doctors, and the ratio of the training set, validation set, and test set is close to 3:1:1. The experimental results show that the accuracy of the multi-scale localization network reaches 93.74%, and the accuracy, specificity, and sensitivity of the classification network reach 86.34%, 81.29%, and 90.48%, respectively. Compared with the champion model of the TNSCUI 2020 classification competition, the accuracy rate is 1.52 points higher. Therefore, the network model proposed in this paper can effectively diagnose benign and malignant thyroid nodules. |
