+1 Recommend
0 collections
      • Record: found
      • Abstract: found
      • Article: not found

      Thyroid Nodule Classification in Ultrasound Images by Fine-Tuning Deep Convolutional Neural Network


      Read this article at

          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.


          With many thyroid nodules being incidentally detected, it is important to identify as many malignant nodules as possible while excluding those that are highly likely to be benign from fine needle aspiration (FNA) biopsies or surgeries. This paper presents a computer-aided diagnosis (CAD) system for classifying thyroid nodules in ultrasound images. We use deep learning approach to extract features from thyroid ultrasound images. Ultrasound images are pre-processed to calibrate their scale and remove the artifacts. A pre-trained GoogLeNet model is then fine-tuned using the pre-processed image samples which leads to superior feature extraction. The extracted features of the thyroid ultrasound images are sent to a Cost-sensitive Random Forest classifier to classify the images into “malignant” and “benign” cases. The experimental results show the proposed fine-tuned GoogLeNet model achieves excellent classification performance, attaining 98.29% classification accuracy, 99.10% sensitivity and 93.90% specificity for the images in an open access database (Pedraza et al. 16), while 96.34% classification accuracy, 86% sensitivity and 99% specificity for the images in our local health region database.

          Related collections

          Most cited references15

          • Record: found
          • Abstract: found
          • Article: not found

          Image restoration by the method of convex projections: part 1 theory.

          A projection operator onto a closed convex set in Hilbert space is one of the few examples of a nonlinear map that can be defined in simple abstract terms. Moreover, it minimizes distance and is nonexpansive, and therefore shares two of the more important properties of ordinary linear orthogonal projections onto closed linear manifolds. In this paper, we exploit the properties of these operators to develop several iterative algorithms for image restoration from partial data which permit any number of nonlinear constraints of a certain type to be subsumed automatically. Their common conceptual basis is as follows. Every known property of an original image f is envisaged as restricting it to lie in a well-defined closed convex set. Thus, m such properties place f in the intersection E(0) = E(i) of the corresponding closed convex sets E(1),E(2),...EE(m). Given only the projection operators PE(i) onto the individual E(i)'s, i = 1 --> m, we restore f by recursive means. Clearly, in this approach, the realization of the P(i)'s in a Hilbert space setting is one of the major synthesis problems. Section I describes the geometrical significance of the three main theorems in considerable detail, and most of the underlying ideas are illustrated with the aid of simple diagrams. Section II presents rules for the numerical implementation of 11 specific projection operators which are found to occur frequently in many signal-processing applications, and the Appendix contains proofs of all the major results.
            • Record: found
            • Abstract: not found
            • Article: not found

            A Review on Ultrasound-Based Thyroid Cancer Tissue Characterization and Automated Classification

              • Record: found
              • Abstract: found
              • Article: not found

              Computer-aided diagnosis for the differentiation of malignant from benign thyroid nodules on ultrasonography.

              We sought to evaluate the diagnostic performance of an artificial neural network (ANN) and binary logistic regression (BLR) in differentiating malignant from benign thyroid nodules on ultrasonography.

                Author and article information

                J Digit Imaging
                J Digit Imaging
                Journal of Digital Imaging
                Springer International Publishing (Cham )
                10 July 2017
                10 July 2017
                August 2017
                : 30
                : 4
                : 477-486
                [1 ]ISNI 0000 0001 2154 235X, GRID grid.25152.31, Department of Computer Science, , University of Saskatchewan, ; 176 Thorvaldson Bldg, 110 Science Place, Saskatoon, SK S7N 5C9 Canada
                [2 ]ISNI 0000 0001 2154 235X, GRID grid.25152.31, Department of Medical Imaging, , University of Saskatchewan, ; 103 Hospital Dr, Saskatoon, SK S7N 0W8 Canada
                [3 ]ISNI 0000 0004 0462 8356, GRID grid.412271.3, Department of Surgery, , Royal University Hospital, ; 103 Hospital Drive, Suite 2646, Saskatoon, SK S7N 0W8 Canada
                © The Author(s) 2017

                Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

                Funded by: FundRef http://dx.doi.org/10.13039/501100000106, Saskatchewan Health Research Foundation;
                Award ID: 3702
                Award Recipient :
                Custom metadata
                © Society for Imaging Informatics in Medicine 2017

                Radiology & Imaging
                ultrasonography,thyroid nodules,machine learning,computer vision,deep learning,convolutional neural network,fine-tuning


                Comment on this article