Journal of the Korean Society of Radiology (한국방사선학회논문지)

The Korean Society of Radiology (한국방사선학회)
권호 표지
DOI QR코드

DOI QR Code

Image Classification of Thyroid Ultrasound Nodules using Machine Learning and GLCM

머신러닝과 GLCM을 이용하여 갑상샘 초음파영상의 결절분류에 관한 연구

  • Ye-Na Jung (Department of Radiological Science, College of Health Sciences, Catholic University of Pusan) ;
  • Soo-Young Ye (Department of Radiological Science, College of Health Sciences, Catholic University of Pusan)
  • 정예나 (부산가톨릭대학교 보건과학대학 방사선학과) ;
  • 예수영 (부산가톨릭대학교 보건과학대학 방사선학과)
  • Received : 2024.07.01
  • Accepted : 2024.08.31
  • Published : 2024.08.31
Download PDF
⟨ Previous Next ⟩

Abstract

This study aimed to classify normal and nodule images in thyroid ultrasound images using GLCM and machine learning. The research was conducted on 600 patients who visited S Hospital in Busan and were diagnosed with thyroid nodules using thyroid ultrasound. In the thyroid ultrasound images, the ROI was set to a size of 50x50 pixels, and 21 parameters and 4 angles were used with GLCM to analyze the normal thyroid patterns and thyroid nodule patterns. The analyzed data was used to distinguish between normal and nodule diagnostic results using the SVM model and KNN model in MATLAB. As a result, the accuracy of the thyroid nodule classification rate was 94% for SVM model and 91% for the KNN model. Both models showed an accuracy of over 90%, indicating that the classification rate is excellent when using machine learning for the classification of normal thyroid and thyroid nodules. In the ROC curve, the ROC curve for the SVM model was generally higher compared to the KNN model, indicating that the SVM model has higher within-sample performance than the KNN model. Based on these results, the SVM model showed high accuracy in diagnosing thyroid nodules. This result can be used as basic data for future research as an auxiliary tool for medical diagnosis and is expected to contribute to the qualitative improvement of medical services through machine learning technology.

본 연구는 갑상샘 초음파 영상에서 정상영상과 결절영상을 GLCM과 머신러닝을 이용하여 분류하고자 하였다. 부산 소재 S병원에 내원하여 갑상샘 초음파를 이용하여 갑상선 결절 진단받은 600명을 대상으로 연구를 진행하였다. 갑상샘 초음파 영상에서 ROI 50 X 50 픽셀 크기로 설정 하고 GLCM을 이용하여 21개의 파라메터와 4가지 각도를 사용하여 갑상샘 정상 패턴과 갑상샘 결절 패턴을 분석하였다. 분석된 자료는 MATLAB 모델 중 SVM모델과 KNN모델을 이용하여 진단 결과가 정상과 결절을 구별할 수 있도록 하였다. 그 결과 갑상샘 결절 분류율의 정확도는 SVM모델은 94%, KNN모델은 91%으로 나타났다. 두 모델 모두 90% 이상의 정확도를 나타내었는데 이는 갑상샘 정상과 갑상샘 결절의 분류를 위해 머신러닝을 이용할 경우 분류율이 우수하다는 것을 알 수 있다. ROC곡선에서도 SVM 모델에 대한 ROC 곡선은 전반적으로 KNN모델과 비교해 ROC곡선이 높으며, 이는 KNN모델보다 표본 내 성능이 높다는 결과가 나타났다. 이러한 결과를 바탕으로 SVM 모델이 갑상샘 결절 진단에서 높은 정확도를 보여주었다. 이 결과는 향후 의료 진단보조 도구로서의 연구에 기초자료로 활용될 수 있으며, 머신러닝 기술이 의료 서비스의 질적 개선에 기여할 것으로 기대된다.

Keywords

References

  1. D. S. Ross, "Nonpalpable thyroid nodules--managing an epidemic", Journal of Clinical Endocrinology & Metabolism, Vol. 87, No. 5, pp. 1938-1940, 2002. http://doi.org/10.1210/jcem.87.5.8552
  2. G. H. Tan, H. Gharib, "Thyroid incidentalomas: management approaches to nonpalpable nodules discovered incidentally on thyroid imaging", Annals of Internal Medicine, Vol. 126, No. 3, pp. 226-231, 1997. https://doi.org/10.7326/0003-4819-126-3-199702010-00009
  3. J. N. Patel, J. Klopper, E. E. Cottrill, "Molecular diagnostics in the evaluation of thyroid nodules: Current use and prospective opportunities", Frontiers in Endocrinology, Vol. 14, pp. 1101410, 2023. http://doi.org/10.3389/fendo.2023.1101410
  4. Y. G. Park, E. k. Lee, Y. S. song, S. H. Kang, B. S. Koo, S. W. Kim, D. G. Na, S. K. Baek, S. W. Oh, M. K. Lee, S. W. Lee, Y. A. Lee, Y. S. Lee, J. Y. Lee, D. J. Lee, L. H. Joo, Y. S. Jung, C. K. Jung, Y. Y. Cho, Y. J. Chung, W. B. Kim, K. H. Yi, H. C. Kang, D. J. Park, "2023 Korean Thyroid Association Management Guidelines for Patients with Thyroid Nodules", International Journal of Thyroidology, Vol. 16, No. 1, pp. 1-31, 2023. http://doi.org/10.11106/ijt.2023.16.1.1
  5. S. Y. Kim, Y. J. Cho, H. S. Shing, J. E. Oh, D. H. Hong, B. W. Yoo, H. Joe, K. S. Shing, D. Y. Son, "Relationship between the Thyroid Nodules and Metabolic Syndrome in Healthy Adults: Using Health Examination Data at One Medical Institution's Health Examination Center (2020)", Korean Journal of Family Practice, Vol. 12, No. 4, pp. 268-273, 2020. http://doi.org/10.21215/kjfp.2022.12.4.268
  6. S. J. Ko, J. S. Lee, S. Y Ye, C. S. Kim, "Application of Texture Features algorithm using Computer Aided Diagnosis of Papillary Thyroid Cancer in the Ultrasonography", The Journal of the Korea Contents Association, Vol. 13, No. 5, pp. 303-310, 2013. https://doi.org/10.5392/JKCA.2013.13.05.303
  7. J. H. Lee, "Research on Ultrasound B-mode Image Aided Diagnosis System Using Bio Big Data and Machine Learning", Journal of the Korean Institute of Elctrical and Electronic Material Engineers, Vol. 28, No. 1, pp.13-25, 2015. https://koreascience.kr/article/JAKO201507037196246.pdf
  8. Vadhiraj, Vijay Vyas, et al., "Ultrasound Image Classification of Thyroid Nodules using Machine Learning Techniques", Journal of Medicina, Vol. 57, No. 6, pp. 527, 2021. http://doi.org/10.3390/medicina57060527
  9. Y. H. Cho, "A Performance Improvement of GLCM Based on Nonuniform Quantization Method", Journal of Korean Institute of Intelligent Systems, Vol. 25, No. 2, pp. 133-138, 2015. https://doi.org/10.5391/JKIIS.2015.25.2.133
  10. J. Y. Cho, S. Y. Ye, "Analysis of Malignant Tumor Using Texture Characteristics in Breast Ultrasonograph", Journal of the Institute of Convergence Signal Processing, Vol. 20, No. 2, pp. 70-77, 2019. http://dx.doi.org/10.23087/jkicsp.2019.20.2.002
  11. J. Y. Cho, C. S. Kim, S. S. Kang, S. J. Ko, S. Y. Ye, "Image Analysis of Computer Aided Diagnosis using Gray Level Co-occurrence Matrix in the Ultrasonography for Benign Prostate Hyperplasia", The Journal of the Korea Contents Association, Vol. 15, No. 3, pp. 184-191, 2015. https://doi.org/10.5392/JKCA.2015.15.03.184
  12. Z. Nitish, P. Vrushsen, "GLCM Textural Features for Brain Tumor Classification", International Journal of Computer Science, Vol. 9, No. 3, pp. 354-359, 2012. www.IJCSI.org
  13. S. Poonguzhali, G. Ravindran, "Automatic Classification of Focal Lesions in Ultrasound Liver Images Using Combined Texture Features", Information Technology Journal, Vol. 7, No. 1, pp. 205-209, 2008. http://dx.doi.org/10.3923/itj.2008.205.209
  14. Y. J. Yang, "A Statistical Approach of the Texture Analysis for the Symptom Classification on the Carotid Plaque Ultrasound Images", Ewha Womans University, Master of Dept. of Computer Science and Engineering, 2007. https://www.riss.kr/link?id=T11039145
  15. M. Nixon, A. Aguado, Feature extraction and image processing. Newnes, Reed Educational and Professional Publiching Ltd. Vol. 30, pp. 747-757, 2002.
  16. J. H. Shin, "Classification of Diagnostic Factors for Atherosclerosis Using Non-Uniform Rational B-Spline Surface and Support Vector Machine Learning", Sungkyunkwan University, Master of Dept. of Industrial Engineering, 2008. https://www.riss.kr/link?id=T11576462
  17. H. Drucker, B. Shahrary, D. C. Gibbon, "Support Vector Machines : Relevance Feedback and Information retrieva", Journal of Information Processing & Management, Vol. 38, No. 3, pp. 305-323, 2002. http://doi.org/10.1016/S0306-4573(01)00037-1