Comparison and Analysis of Ultrasound Diagnosis Networks for Thyroid Nodules Based on Different Computer Vision Task Types

Abstract

This paper reviews the research of deep learning algorithms in the classification of benign and malignant thyroid nodules in ultrasound diagnosis, then summarizes the current research progress on the invasiveness of thyroid nodules. Based on the analysis of the existing classification algorithms, segmentation algorithms, and the characteristics of thyroid ultrasound image data, the paper clarifies limitations and reasons of deep learning in thyroid nodule diagnosis, then proposes the solutions includes the introduction of the clinical experience, pre-training of ultrasound data. The importance of invasive prediction of thyroid nodules and the shortcomings of deep learning in invasive prediction are pointed out, and the prospect of deep learning in the field of invasive prediction of thyroid nodules is prospected.

Ähnliche Arbeiten