15 August 2020>: Clinical Research
Visual Interpretability in Computer-Assisted Diagnosis of Thyroid Nodules Using Ultrasound Images
Xi Wei 1ADEFG , Jialin Zhu 1ACDEF* , Haozhi Zhang 2CDF , Hongyan Gao 3BCF , Ruiguo Yu 4BCD , Zhiqiang Liu 4BCD , Xiangqian Zheng 2BF , Ming Gao 2AF , Sheng Zhang 1ABDEF*DOI: 10.12659/MSM.927007
Med Sci Monit 2020; 26:e927007
Figure 1 Corresponding visualization results are shown for experiments 1–3. The areas of the orange frames represent the visualization results on the datasets. Each column shows the same ultrasound image with different degrees of segmentation. (A) Experiment 1: The model was trained and tested using full images. (B) Experiment 2: The model was trained and tested using images after the removal of peripheral interference. (C) Experiment 3: The model was trained and tested using the region of interest (ROI).