Ultrasound image-based deep learning to assist in diagnosing gross extrathyroidal extension thyroid cancer: a retrospective multicenter study

Abstract

Background: The presence of gross extrathyroidal extension (ETE) in thyroid cancer will affect the prognosis of patients, but imaging examination cannot provide a reliable diagnosis for it. This study was conducted to develop a deep learning (DL) model for localization and evaluation of thyroid cancer nodules in ultrasound images before surgery for the presence of gross ETE.

Methods: From January 2016 to December 2021 grayscale ultrasound images of 806 thyroid cancer nodules (4451 images) from 4 medical centers were retrospectively analyzed, including 517 no gross ETE nodules and 289 gross ETE nodules. 283 no gross ETE nodules and 158 gross ETE nodules were randomly selected from the internal dataset to form a training set and validation set (2914 images), and a multitask DL model was constructed for diagnosing gross ETE. In addition, the clinical model and the clinical and DL combined model were constructed. In the internal test set [974 images (139 no gross ETE nodules and 83 gross ETE nodules)] and the external test set [563 images (95 no gross ETE nodules and 48 gross ETE nodules)], the diagnostic performance of DL model was verified based on the pathological results. And then, compared the results with the diagnosis by 2 senior and 2 junior radiologists.

Findings: In the internal test set, DL model demonstrated the highest AUC (0.91; 95% CI: 0.87, 0.96), which was significantly higher than that of two senior radiologists [(AUC, 0.78; 95% CI: 0.71, 0.85; P < 0.001) and (AUC, 0.76; 95% CI: 0.70, 0.83; P < 0.001)] and two juniors radiologists [(AUC, 0.65; 95% CI: 0.58, 0.73; P < 0.001) and (AUC, 0.69; 95% CI: 0.62, 0.77; P < 0.001)]. DL model was significantly higher than clinical model [(AUC, 0.84; 95% CI: 0.79, 0.89; P = 0.019)], but there was no significant difference between DL model and clinical and DL combined model [(AUC, 0.94; 95% CI: 0.91, 0.97; P = 0.143)]. In the external test set, DL model also demonstrated the highest AUC (0.88, 95% CI: 0.81, 0.94), which was significantly higher than that of one of senior radiologists [(AUC, 0.75; 95% CI: 0.66, 0.84; P = 0.008) and (AUC, 0.81; 95% CI: 0.72, 0.89; P = 0.152)] and two junior radiologists [(AUC, 0.72; 95% CI: 0.62, 0.81; P = 0.002) and (AUC, 0.67; 95 CI: 0.57, 0.77; P < 0.001]. There was no significant difference between DL model and clinical model [(AUC, 0.85; 95% CI: 0.79, 0.91; P = 0.516)] and clinical + DL model [(AUC, 0.92; 95% CI: 0.87, 0.96; P = 0.093)]. Using DL model, the diagnostic ability of two junior radiologists was significantly improved.

Interpretation: The DL model based on ultrasound imaging is a simple and helpful tool for preoperative diagnosis of gross ETE thyroid cancer, and its diagnostic performance is equivalent to or even better than that of senior radiologists.

Funding: Jiangxi Provincial Natural Science Foundation (20224BAB216079), the Key Research and Development Program of Jiangxi Province (20181BBG70031), and the Interdisciplinary Innovation Fund of Natural Science, Nanchang University (9167-28220007-YB2110).

Keywords: Deep learning; Gross extrathyroidal extension thyroid cancer; Ultrasound.

Fig. 1

Inclusion criteria flowchart for the initial population and exclusion. ETE, extrathyroidal extension.

Fig. 2

Image processing and distribution. ROI, region of interest.

Fig. 3

Illustration shows the deep learning (DL) neural network architecture. The data are fed into the backbone network consisting of ResNet and FPN for feature extraction, and the extracted feature maps are fed into the regional proposal network (RPN) to generate regions of interest (ROI). Then all the ROIs are reset by RoIAlign network, and finally the ROI are classified and regressed by the classification branch, and the mask of the object is generated by the mask branch.

Fig. 4

Diagnostic performance comparison among DL models and radiologists. SR, senior radiologist; JR, junior radiologist; SR + DL, DL model assistant senior radiologist; JR + DL, DL model assistant junior radiologist.

Fig. 5

Clinical model nomogram and clinical deep learning combined nomogram. Lymph, lymph node status; Angle, angle between tumor and trachea; Capsule echo, capsule echo state; Ratio, capsule contact ratio; Age, patient's age; DL, deep learning.

Fig. 6

Visualization and focus of the DL model and the guidance to radiologists. ETE, extrathyroidal extension; DL, deep learning model; SR, senior radiologist; JR, junior radiologist. Inside, no gross ETE nodule; in advance, gross ETE nodule. In four sets of representative cases, the first picture of each set is the US picture of thyroid nodule (TN), the second is the diagnostic effect picture of DL model, and the third is the heat map of the nodule and peri-tumor. A: the heat map highlights the area where the TN is in contact with the surrounding capsule; B: the heat map highlights the area where the TN is in contact with the anterior capsule; C: the heat map highlights the area where the TN is in contact with the trachea and nerve; D: the heat map highlights the area where the TN is in contact with the trachea and tracheoesophageal groove.