DOI QR코드

DOI QR Code

Enhancement of Tongue Segmentation by Using Data Augmentation

데이터 증강을 이용한 혀 영역 분할 성능 개선

  • Chen, Hong (College of Information and Computer Engineering, Pingxiang University) ;
  • Jung, Sung-Tae (Department of Computer Engineering, Wonkwang University)
  • Received : 2020.08.27
  • Accepted : 2020.09.14
  • Published : 2020.10.30

Abstract

A large volume of data will improve the robustness of deep learning models and avoid overfitting problems. In automatic tongue segmentation, the availability of annotated tongue images is often limited because of the difficulty of collecting and labeling the tongue image datasets in reality. Data augmentation can expand the training dataset and increase the diversity of training data by using label-preserving transformations without collecting new data. In this paper, augmented tongue image datasets were developed using seven augmentation techniques such as image cropping, rotation, flipping, color transformations. Performance of the data augmentation techniques were studied using state-of-the-art transfer learning models, for instance, InceptionV3, EfficientNet, ResNet, DenseNet and etc. Our results show that geometric transformations can lead to more performance gains than color transformations and the segmentation accuracy can be increased by 5% to 20% compared with no augmentation. Furthermore, a random linear combination of geometric and color transformations augmentation dataset gives the superior segmentation performance than all other datasets and results in a better accuracy of 94.98% with InceptionV3 models.

많은 양의 데이터는 딥 러닝 모델의 견고성을 향상시키고 과적합 문제를 방지할 수 있게 해준다. 자동 혀 분할에서, 혀 영상 데이터 세트를 실제로 수집하고 라벨링하는 데에는 많은 어려움이 수반되므로 많은 양의 혀 영상 데이터를 사용하기 쉽지 않다. 데이터 증강은 새로운 데이터를 수집하지 않고 레이블 보존 변환을 사용하여 학습 데이터 세트를 확장하고 학습 데이터의 다양성을 증가시킬 수 있다. 이 논문에서는 이미지 자르기, 회전, 뒤집기, 색상 변환과 같은 7 가지 데이터 증강 방법을 사용하여 확장된 혀 영상 학습 데이터 세트를 생성하였다. 데이터 증강 방법의 성능을 확인하기 위하여 InceptionV3, EfficientNet, ResNet, DenseNet 등과 같은 전이 학습 모델을 사용하였다. 실험 결과 데이터 증강 방법을 적용함으로써 혀 분할의 정확도를 5~20% 향상시켰으며 기하학적 변환이 색상 변환보다 더 많은 성능 향상을 가져올 수 있음을 보여주었다. 또한 기하학적 변환 및 색상 변환을 임의로 선형 조합한 방법이 다른 데이터 증강 방법보다 우수한 분할 성능을 제공하여 InveptionV3 모델을 사용한 경우에 94.98 %의 정확도를 보였다.

Keywords

References

  1. L. Wang, X. He, Y. Tang, P. Chen, and G. Yuan, "Tongue Semantic Segmentation Based on Fully Convolutional Neural Network," International Conference on Intelligent Computing, Automation and Systems, pp. 298-301, 2019.
  2. B. Lin, J. Xie, C. Li, and Y. Qu, "Deeptongue: Tongue Segmentation Via Resnet," IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1035-1039, 2018.
  3. C. Bowles, L. Chen, R. Guerrero, P. Bentley, R.N. Gunn, A. Hammers, D.A. Dickie, M.D. Hernandez, K.M. Wardlaw, and D. Rueckert, "GAN Augmentation: Augmenting Training Data using Generative Adversarial Networks," Available:https://arxiv.org/abs/1810.10863.
  4. L. Taylor and G. Nitschke, "Improving Deep Learning with Generic Data Augmentation," IEEE Symposium Series on Computational Intelligence, pp. 1542-1547, 2018
  5. F. Perez, C.N. Vasconcelos, S. Avila, and E. Valle. "Data Augmentation for Skin Lesion Analysis." ISIC Skin Image Analysis Workshop and Challenge, pp. 303-311, 2018.
  6. J. Rama, C. Nalini, and A. Kumaravel, "Image Pre-Processing: Enhance the Performance of Medical Image Classification Using Various Data Augmentation Technique," ACCENTS Transactions on Image Processing and Computer Vision, vol. 5, no. 14, pp. 7-14, Feb. 2019.
  7. I. Sirazitdinov, M. Kholiavchenko, R. Kuleev, and B. Ibragimov, "Data Augmentation for Chest Pathologies Classification," IEEE 16th International Symposium on Biomedical Imaging, pp. 1216-1219, 2019.
  8. S. Kayal, F. Dubost, H. Tiddens, and M. de Bruijne, "Spectral Data Augmentation Techniques to Quantify Lung Pathology from CT-Images," IEEE 17th International Symposium on Biomedical Imaging, pp. 586-590, 2020.
  9. J. Pandian, G. Geetharamani, and B. Annette, "Data Augmentation on Plant Leaf Disease Image Dataset Using Image Manipulation and Deep Learning Techniques," IEEE 9th International Conference on Advanced Computing , pp. 199-204, 2019.
  10. P. Dimitrakopoulos, G. Sfikas, and C. Nikou, "ISING-GAN: Annotated Data Augmentation with a Spatially Constrained Generative Adversarial Network," IEEE 17th International Symposium on Biomedical Imaging, pp. 1600-1603, 2020.
  11. C. Shorten and T.M. Khoshgoftaar, "A survey on Image Data Augmentation for Deep Learning," Journal of Big Data, vol. 6, article. 60, pp. 1-48, 2019. https://doi.org/10.1186/s40537-018-0162-3
  12. L. Huang, W. Pan, Y. Zhang, L. Qian, N. Gao, and Y. Wu, "Data Augmentation for Deep Learning-Based Radio Modulation Classification," IEEE Access, vol. 8, pp. 1498-1506, Dec. 2019. https://doi.org/10.1109/access.2019.2960775
  13. M. Frid-Adar, E. Klang, M. Amitai, J. Goldberger, and H. Greenspan, "Synthetic data augmentation using GAN for improved liver lesion classification," IEEE 15th International Symposium on Biomedical Imaging, pp. 289-293, 2018.
  14. T. Yang , Y. Yoshimura, A. Morita, T. Namiki ,and T. Nakaguchi, "Fully Automatic Segmentation of Sublingual Veins from Retrained U-Net Model for Few Near Infrared Images," The Ninth International Workshop on Image Media Quality and its Applications , Available: https://arxiv.org/abs/1812.09477, 2018.
  15. S. Minaee, Y. Boykov, F. Porikli, A. Plaza, N. Kehtarnavaz, and D. Terzopoulos, "Image Segmentation Using Deep Learning: A Survey," Available: https://arxiv.org/abs/2001.05566, 2020.
  16. O. Ronneberger, P. Fischer, and T. Brox. "U-Net: Convolutional Networks for Biomedical Image Segmentation," International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 234-241, 2015.