A Non-Invasive Follicular Thyroid Cancer Risk Prediction System Based on Deep Hybrid Multi-feature Fusion Network

doi:10.37015/AUDT.2023.230023

Abstract

Abstract:

Objective A non-invasive assessment of the risk of benign and malignant follicular thyroid cancer is invaluable in the choice of treatment options. The extraction and fusion of multidimensional features from ultrasound images of follicular thyroid cancer is decisive in improving the accuracy of identifying benign and malignant thyroid cancer. This paper presents a non-invasive preoperative benign and malignant risk assessment system for follicular thyroid cancer, based on the proposed deep feature extraction and fusion of ultrasound images of follicular thyroid cancer.

Methods First, this study uses a convolution neural network (CNN) to obtain a global feature map of the image, and the fusion of global features cropped to local features to identify tumor images. Secondly, this tumour image is also extracted by googleNet and ResNet respectively to extract features and recognize the image. Finally, we employ an averaging algorithm to obtain the final recognition results.

Results The experimental results show that the method proposed in this study achieved 89.95% accuracy, 88.46% sensitivity, 91.30% specificity and an AUC value of 96.69% in the local dataset obtained from Peking University Shenzhen Hospital, all of which are far superior to other models.

Conclusion In this study, a non-invasive risk prediction system is proposed for ultrasound images of thyroid follicular tumours. We solve the problem of unbalanced sample distribution by means of an image enhancement algorithm. In order to obtain enough features to differentiate ultrasound images, a three-branched feature extraction network was designed in this study, and a balance of sensitivity and specificity is ensured by an averaging algorithm.

Key words: Follicular thyroid cancer; Ultrasound image; Risk prediction system; Hybrid multi-feature fusion; Convolutional neural network

Yalin Wu, PhD, Qiaoli Ge, MM, Linyang Yan, PhD, Desheng Sun, MD. A Non-Invasive Follicular Thyroid Cancer Risk Prediction System Based on Deep Hybrid Multi-feature Fusion Network. Advanced Ultrasound in Diagnosis and Therapy, 2023, 7(4): 373-380.

Figures/Tables 12

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Table 1

Figure 11

References 23

[1]	Grani G, Lamartina L, Durante C, Filetti S, Cooper DS. Follicular thyroid cancer and Hürthle cell carcinoma: Challenges in diagnosis, treatment, and clinical management. Lancet Diabetes Endocrinol 2018; 6:500-514. doi: 10.1016/S2213-8587(17)30325-X
[2]	Tallini G, Tuttle RM, Ghossein RA. The history of the follicular variant of papillary thyroid carcinoma. J Clin Endocrinol Metab 2017; 102:15-22. doi: 10.1210/jc.2016-2976 pmid: 27732333
[3]	Pstrag N, Ziemnicka K, Bluyssen H, Wesoly J. Thyroid cancers of follicular origin in a genomic light: In-depth overview of common and unique molecular marker candidates. Mol Cancer 2018; 17:116. doi: 10.1186/s12943-018-0866-1 pmid: 30089490
[4]	Baldini E, Sorrenti S, Tartaglia F, Catania A, Palmieri A, Pironi D, et al. New perspectives in the diagnosis of thyroid follicular lesions. Int J Surg 2017; 41 Suppl 1:S7-S12.
[5]	Xu B, Tallini G, Scognamiglio T, Roman BR, Tuttle RM, Ghossein RA. Outcome of large noninvasive follicular thyroid neoplasm with papillary-like nuclear features. Thyroid 2017; 27:512-517. doi: 10.1089/thy.2016.0649 pmid: 28136139
[6]	de Jong MC, McNamara J, McGlashan N, Winter L, and Mihai R. Risk of malignancy in thyroid nodules selected for fine needle aspiration biopsy based on ultrasound risk stratification. British Journal of Surgery 2022 ;109(Supplement_2),znac056.
[7]	Tian T, Chen Y, Xiang Y, Liu L, Liu B. Remarkable response of pulmonary metastases rather than remnant thyroid in 131I therapy of follicular thyroid cancer. Clin Nucl Med 2019; 44:327-329. doi: 10.1097/RLU.0000000000002477 pmid: 30688748
[8]	García-Burillo A, Monturiol-Duran JA, Iglesias-Felip C, Villasboas-Rosciolesi D, Castell-Conesa J. Follicular thyroid carcinoma metastases on round ligament of liver. Clin Nucl Med 2021; 46:326-328. doi: 10.1097/RLU.0000000000003526 pmid: 33512955
[9]	Ohori NP, Nishino M. Follicular neoplasm of thyroid revisited: current differential diagnosis and the impact of molecular testing. Adv Anat Pathol 2023; 30:11-23. doi: 10.1097/PAP.0000000000000368
[10]	Xu B, Ghossein RA. Advances in thyroid pathology: high grade follicular cell-derived thyroid carcinoma and anaplastic thyroid carcinoma. Adv Anat Pathol 2023; 30:3-10. doi: 10.1097/PAP.0000000000000380
[11]	Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer 2012; 48:441-446. doi: 10.1016/j.ejca.2011.11.036 pmid: 22257792
[12]	Erickson BJ, Korfiatis P, Akkus Z, Kline TL. Machine learning for medical imaging. Radiographics 2017; 37:505-515. doi: 10.1148/rg.2017160130 pmid: 28212054
[13]	Yu J, Deng Y, Liu T, Zhou J, Jia X, Xiao T, et al. Lymph node metastasis prediction of papillary thyroid carcinoma based on transfer learning radiomics. Nat Commun 2020; 11:4807. doi: 10.1038/s41467-020-18497-3 pmid: 32968067
[14]	Shin I, Kim YJ, Han K, Lee E, Kim HJ, Shin JH, et al. Application of machine learning to ultrasound images to differentiate follicular neoplasms of the thyroid gland. Ultrasonography 2020; 39:257-265. doi: 10.14366/usg.19069 pmid: 32299197
[15]	Lim KJ, Choi CS, Yoon DY, Chang SK, Kim KK, Han H, et al. Computer-aided diagnosis for the differentiation of malignant from benign thyroid nodules on ultrasonography. Acad Radiol 2008; 15:853-858. doi: 10.1016/j.acra.2007.12.022 pmid: 18572120
[16]	Wu H, Deng Z, Zhang B, Liu Q, Chen J. Classifier model based on machine learning algorithms: application to differential diagnosis of suspicious thyroid nodules via sonography. AJR Am J Roentgenol 2016; 207:859-864. doi: 10.2214/AJR.15.15813
[17]	Matiz S, Barner KE. Inductive conformal predictor for convolutional neural networks: Applications to active learning for image classification. Pattern Recognition 2019; 90:172-182. doi: 10.1016/j.patcog.2019.01.035
[18]	Wang Y, Chen Y, Yang N, Zheng L, Dey N, Ashour AS, et al. Classification of mice hepatic granuloma microscopic images based on a deep convolutional neural network. Applied Soft Computing 2019; 74:40-50. doi: 10.1016/j.asoc.2018.10.006
[19]	Meng N, Lam EY, Tsia KK, So HK. Large-scale multi-class image-based cell classification with deep learning. IEEE J Biomed Health Inform 2019; 23:2091-2098. doi: 10.1109/JBHI.6221020
[20]	Wang C, Chen D, Hao L, Liu X, Zeng Y, Chen J, et al. Pulmonary image classification based on inception-v3 transfer learning model. IEEE Access 2019; 7:146533-146541. doi: 10.1109/Access.6287639
[21]	Yu L, Chen H, Dou Q, Qin J, Heng PA. Automated melanoma recognition in dermoscopy images via very deep residual networks. IEEE Trans Med Imaging 2017; 36:994-1004. doi: 10.1109/TMI.2016.2642839
[22]	Yan L, Shi Y, Wei M, Wu Y. Multi-feature fusing local directional ternary pattern for facial expressions signal recognition based on video communication system. Alexandria Engineering Journal 2023; 63:307-320. doi: 10.1016/j.aej.2022.08.003
[23]	Wu Y, Zhang Q, Hu Y, Sun-Woo K, Zhang X, Zhu H, et al. Novel binary logistic regression model based on feature transformation of XGBoost for type 2 diabetes mellitus prediction in healthcare systems. Future Generation Computer Systems 2022; 129:1-2. doi: 10.1016/j.future.2021.11.003

Method	Accuracy(%)	Sensitivity(%)	Specificity(%)	AUC(%)
LeNet	72.60	70.19	74.78	80.99
AlexNet	76.71	75.96	77.39	86.36
ZFNet	53.88	2.88	100.00	70.61
VGGNet	72.60	60.58	83.48	80.14
Inception v1	90.41	86.54	93.91	94.38
ResNet	84.93	83.65	86.09	91.30
SENet	56.62	98.08	19.13	72.03
Cropping	84.93	88.46	81.74	92.07
Proposed method	89.95	88.46	91.30	96.69