A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

Luoyan Wang; Xiaogen Zhou; Xingqing Nie; Xingtao Lin; Jing Li; Haonan Zheng; Ensheng Xue; Shun Chen; Cong Chen; Min Du; Tong Tong; Qinquan Gao; Meijuan Zheng

doi:10.3389/fnins.2022.878718

A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

Front Neurosci. 2022 May 19:16:878718. doi: 10.3389/fnins.2022.878718. eCollection 2022.

Authors

Luoyan Wang^{1

2}, Xiaogen Zhou^{1

2}, Xingqing Nie^{1

2}, Xingtao Lin^{1

2}, Jing Li^{1

2}, Haonan Zheng^{1

2}, Ensheng Xue^{3

4}, Shun Chen³, Cong Chen³, Min Du^{1

2}, Tong Tong^{1

2}, Qinquan Gao^{1

2}, Meijuan Zheng³

Affiliations

¹ College of Physics and Information Engineering, Fuzhou University, Fuzhou, China.
² Fujian Key Lab of Medical Instrumentation & Pharmaceutical Technology, Fuzhou University, Fuzhou, China.
³ Fujian Medical University Union Hospital, Fuzhou, China.
⁴ Fujian Medical Ultrasound Research Institute, Fuzhou, China.

Abstract

Automated thyroid nodule classification in ultrasound images is an important way to detect thyroid nodules and to make a more accurate diagnosis. In this paper, we propose a novel deep convolutional neural network (CNN) model, called n-ClsNet, for thyroid nodule classification. Our model consists of a multi-scale classification layer, multiple skip blocks, and a hybrid atrous convolution (HAC) block. The multi-scale classification layer first obtains multi-scale feature maps in order to make full use of image features. After that, each skip-block propagates information at different scales to learn multi-scale features for image classification. Finally, the HAC block is used to replace the downpooling layer so that the spatial information can be fully learned. We have evaluated our n-ClsNet model on the TNUI-2021 dataset. The proposed n-ClsNet achieves an average accuracy (ACC) score of 93.8% in the thyroid nodule classification task, which outperforms several representative state-of-the-art classification methods.

Keywords: deep convolutional neural network; densely connection; hybrid atrous convolution; multi-scale; the thyroid nodule classification.