DOI QR코드

DOI QR Code

A Divide-Conquer U-Net Based High-Quality Ultrasound Image Reconstruction Using Paired Dataset

짝지어진 데이터셋을 이용한 분할-정복 U-net 기반 고화질 초음파 영상 복원

  • Minha Yoo (National Institute for Mathematical Sciences) ;
  • Chi Young Ahn (National Institute for Mathematical Sciences)
  • 유민하 (국가수리과학연구소 산업수학연구본부) ;
  • 안치영 (국가수리과학연구소 산업수학연구본부)
  • Received : 2024.05.08
  • Accepted : 2024.06.11
  • Published : 2024.06.30

Abstract

Commonly deep learning methods for enhancing the quality of medical images use unpaired dataset due to the impracticality of acquiring paired dataset through commercial imaging system. In this paper, we propose a supervised learning method to enhance the quality of ultrasound images. The U-net model is designed by incorporating a divide-and-conquer approach that divides and processes an image into four parts to overcome data shortage and shorten the learning time. The proposed model is trained using paired dataset consisting of 828 pairs of low-quality and high-quality images with a resolution of 512x512 pixels obtained by varying the number of channels for the same subject. Out of a total of 828 pairs of images, 684 pairs are used as the training dataset, while the remaining 144 pairs served as the test dataset. In the test results, the average Mean Squared Error (MSE) was reduced from 87.6884 in the low-quality images to 45.5108 in the restored images. Additionally, the average Peak Signal-to-Noise Ratio (PSNR) was improved from 28.7550 to 31.8063, and the average Structural Similarity Index (SSIM) was increased from 0.4755 to 0.8511, demonstrating significant enhancements in image quality.

Keywords

Acknowledgement

본 연구는 정부의 재원으로 국가수리과학연구소(NIMS)와 한국연구재단(NRF)의 지원을 받아 수행하였음(NIMS-B24910000, NRF-2021R1A2C1010993). 초음파 영상 데이터 획득을 위해 수고해주신 Sonographer 김정 선생님께 감사의 뜻을 전합니다.

References

  1. Stock KF, Klein B, Steubl D, Lersch C, Heemann U, Wagenpfeil S, Eyer F, Clevert DA. Comparison of a pocket-size ultrasound device with a premium ultrasound machine: diagnostic value and time required in bedside ultrasound examination. Abdominal Imag. 2015;40(7):2861-6. https://doi.org/10.1007/s00261-015-0406-z
  2. Nelson BP, Sanghvi A. Out of hospital point of care ultrasound: Current use models and future directions. Eur J Trauma Emergency Surgery. 2016;42(2):139-50. https://doi.org/10.1007/s00068-015-0494-z
  3. Tasinkiewicz J, 3D System Aperture Imaging Method in Spectrum Domain for Low-Cost Portable Ultrasound Systems. Arch. Acoust. 2023;48(4):559-72.
  4. https://ultrasoundenhance2023.grand-challenge.org/
  5. Zhou Z, Guo Y, Wang Y. Handheld Ultrasound Video High-Quality Reconstruction Using a Low-Rank Representation Multipathway Generative Adversarial Network. IEEE Trans Neural Netw Learn Syst. 2021;32(2):575-88. https://doi.org/10.1109/TNNLS.2020.3025380
  6. Michailovich OV, Tannenbaum A. Despeckling of medical ultrasound images. IEEE Trans Ultrason Ferroelectr Freq Control. 2006;53(1):64-78. https://doi.org/10.1109/TUFFC.2006.1588392
  7. Coupe P, Hellier P, Kervrann C, Barillot C. Nonlocal means-based speckle filtering for ultrasound images. IEEE Trans Image Process. 2009;18(10):2221-9. https://doi.org/10.1109/TIP.2009.2024064
  8. Hyun D, Brickson LL, Looby KT, Dahl JJ. Beamforming and Speckle Reduction Using Neural Networks. IEEE Trans Ultrason Ferroelectr Freq Control. 2019;66(5):898-910. https://doi.org/10.1109/TUFFC.2019.2903795
  9. Lan Y, Zhang X. Real-Time Ultrasound Image Despeckling Using Mixed-Attention Mechanism Based Residual UNet. IEEE Access. 2020;8:195327-40. https://doi.org/10.1109/ACCESS.2020.3034230
  10. Moinuddin M, Khan S, Alsaggaf AU, Abdulaal MJ, Al-Saggaf UM, Ye JC. Medical ultrasound image speckle reduction and resolution enhancement using texture compensated multi-resolution convolution neural network. Front Physiol. 2022;13:961571.
  11. Karaoglu O, Bilge HS, Uluer I. Removal of speckle noises from ultrasound images using five different deep learning networks. Int J Eng Sci Technol. 2022;29:101030.
  12. Skandarani Y, Jodoin P-M, Lalande A. GANs for Medical Image Synthesis: An Empirical Study. J. Imaging. 2023;9(3):69.
  13. Zagzebski JA, Essentials of ultrasound physics. Maryland Heights: Mosby; 1996. pp. 45-65.
  14. Jang J, Ahn CY. Industrial Mathematics in Ultrasound Imaging. J Ksiam. 2016;20(3):175-202. https://doi.org/10.12941/jksiam.2016.20.175
  15. Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation. Medical Image Computing and Computer-Assisted Intervention (MICCAI). Springer, 2015; 9351:234-41.
  16. Hu X, Naiel MA, Wong A, Lamm M, Fieguth P, editors. RU-Net: A robust UNet architecture for image super-resolution. Conf Proc IEEE/CVF Computer Vision and Pattern Recognition Workshops. 2019.