ABSTRACT
Thyroid nodule is a common clinical disease, and its incidence is increasing year by year. With the continuous development of medical imaging technology, most thyroid nodules can be detected by ultrasound images, but it is difficult for even the most experienced radiologists to classify the nature of thyroid nodules accurately every time. The reason for this is the uneven appearance of benign and malignant nodules. Therefore, this paper proposes an algorithm for the classification of benign and malignant thyroid nodules based on ConvNeXt. The algorithm contains the following main contents. The first is to remove artificial markers from the ultrasound image. Then, data expansion was performed on the dataset of thyroid ultrasound images. Finally, an attention mechanism is introduced into the ConvNeXt convolutional neural network, and the enhanced ultrasound image set is used as a training set to perform transfer learning on the pre-trained ConvNeXt convolutional neural network to classify benign and malignant nodules. The method proposed in this paper achieved a correct rate of 99.32%, a sensitivity of 99.82%, and a specificity of 96.13% on the public dataset of thyroid nodules.The final experiment shows that artificial markers removal and data expansion of the original image can improve the training effect. The image region concerned by the network model is drawn by Grad-CAM. It can be concluded that the introduction of attention mechanism into ConvNeXt network model can make the model pay more attention to the texture information of nodules in ultrasound images and describe the difference between the properties of focal and non-focal areas.
- Haugen,Bryan R . 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: What is new and what has changed?[J]. Cancer: A Journal of the American Cancer Society, 2017.Lee, Won etc. Doppler ultrasound-guided thread lifting[J]. JOURNAL OF COSMETIC DERMATOLOGY,2020,19(8):1921-1927.Google Scholar
- Pellegriti G, Frasca F, Regalbuto C, Worldwide increasing incidence of thyroid cancer: update on epidemiology and risk factors[J]. Journal of cancer epidemiology, 2013, 2013.A. P. Kadi and T. Loupas, "On the performance of regression and step-initialized IIR clutter filters for color Doppler systems in diagnostic medical ultrasound," in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 42, no. 5, pp. 927-937, Sept. 1995, doi: 10.1109/58.464825.Google Scholar
- Iakovidis D K, Keramidas E G, Maroulis D. Fuzzy local binary patterns for ultrasound texture characterization[C]//International conference image analysis and recognition. Springer, Berlin, Heidelberg, 2008: 750-759.Yu A C H, Cobbold༲S C. Single ensemble based eigen-processing methods for color flow imaging part 1 the Hankel-SVD filter [J]. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 2008, 55( 3) : 559-573.Google Scholar
- Katsigiannis S, Keramidas E G, Maroulis D. A contourlet transform feature extraction scheme for ultrasound thyroid texture classification[J]. International Journ Engineering Intelligent Systems Electrical Engineering Communications, 2010, 18(3): 171.Google Scholar
- Savelonas M A, Iakovidis D K, Dimitropoulos N, Computational characterization of thyroid tissue in the radon domain[C]//Twentieth IEEE International Symposium on Computer-Based Medical Systems (CBMS'07). IEEE, 2007: 189-192.Kargel C, Höbenreich G, Trummer B, Adaptive clutter rejection filtering in ultrasonic strain-flow imaging[J]. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2003, 50(7): 824-835.Google Scholar
- Chi J, Yu X, Zhang Y. Ultrasound image diagnosis of thyroid nodule canceration by fusion of deep network and superficial texture features [J]. Chinese Journal of Image and Graphics, 2018, 23(10):12.(In Chinese)Han, Tian, etc. Combination bidirectional long short-term memory and capsule network for rotating machinery fault Diagnosis [J]. Diagnosis: JOURNAL OF THE INTERNATIONAL MEASUREMENT CONFEDERATION,2021,176.Google Scholar
- Song W, Li S, Liu J, Multitask cascade convolution neural networks for automatic thyroid nodule detection and recognition[J]. IEEE journal of biomedical and health informatics, 2018, 23(3): 1215-1224.(In Chinese)Chu Ya-li, Zheng Hong, HOU Xiu-ping. Chinese semantic similarity calculation based on dynamic semantic encoding bidirectional LSTM [J]. Computer applications and software, 2020,37(6):224-229.Google Scholar
- Li Y, Zhao Y, Xie R, cRes-GAN algorithm for classification of benign and malignant thyroid nodules [J]. Machinery & Electronics, 2020, 38(4):5.(In Chinese)Zhu Jiayin, Wang Rongbo, Huang Xiaoxi Multi-level metaphor recognition method based on bi-lstm [J]. Journal of dalian university of technology,2020,60(2):209-215.Google Scholar
- Zhang F, Weng Y, Su J, Thyroid nodule image classification based on TV model and GoogLeNet [J]. Application Research of Computers, 2020(S01):3.(In Chinese)Google Scholar
- Shao X, Liu Z, Song B. An adaptive image restoration method based on TV model[J].Journal of Circuits and Systems, 2004, 9(2):5.(In Chinese)Google Scholar
- Liu Z, Mao H, Wu C Y, A convnet for the 2020s[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 11976-11986.Google Scholar
- Woo S, Park J, Lee J Y, Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.Google Scholar
- Pedraza L, Vargas C, Narváez F, An open access thyroid ultrasound image database[C]//10th International Symposium on Medical Information Processing and Analysis. International Society for Optics and Photonics, 2015, 9287: 92870W.Google Scholar
- Selvaraju R R, Cogswell M, Das A, Grad-cam: Visual explanations from deep networks via gradient-based localization[C]//Proceedings of the IEEE international conference on computer vision. 2017: 618-626.Google Scholar
- Luo S, Kim E H, Dighe M, Thyroid nodule classification using ultrasound elastography via linear discriminant analysis[J]. Ultrasonics, 2011, 51(4): 425-431.Google ScholarCross Ref
- Acharya U R, Sree S V, Swapna G, Effect of complex wavelet transform filter on thyroid tumor classification in three-dimensional ultrasound[J]. Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine, 2013, 227(3): 284-292.Google ScholarCross Ref
- Lim K J, Choi C S, Yoon D Y, Computer-aided diagnosis for the differentiation of malignant from benign thyroid nodules on ultrasonography[J]. Academic radiology, 2008, 15(7): 853-858.Google ScholarCross Ref
- Savelonas M, Maroulis D, Sangriotis M. A computer-aided system for malignancy risk assessment of nodules in thyroid US images based on boundary features[J]. Computer methods and programs in biomedicine, 2009, 96(1): 25-32.Google Scholar
- Benign and Malignant Classification of Thyroid Nodules Based on ConvNeXt
Recommendations
Cascade UNet and CH-UNet for Thyroid Nodule Segmentation and Benign and Malignant Classification
Segmentation, Classification, and Registration of Multi-modality Medical Imaging DataAbstractThe thyroid gland secretes indispensable hormones that are necessary for all the cells in your body to work normally. In order to diagnose and treat thyroid cancer at the earliest stage, it is desired to characterize the nodule accurately. We ...
Ensemble Classifier for Benign-Malignant Mass Classification
Mammography is currently the most effective imaging modality for early detection of breast cancer. In a CAD system for masses based on mammography, a mammogram is segmented to detect the masses. The segmentation gives rise to mass regions of interested ...
A Support Vector Machine based Prediction Model for Discrimination of Malignant Pulmonary Nodules from Benign Nodules
BIOSTEC 2016: Proceedings of the International Joint Conference on Biomedical Engineering Systems and TechnologiesLung cancer is the leading cause of cancer death in the United States and worldwide. Most patients are
diagnosed at an advanced stage, usually stage III or IV. Identification of lung cancer patients at an early
stage might enable oncologists to ...
Comments