Oto-Encoder Ağı ile Ultrason Görüntülerinin Yatay (Lateral) Çözünürlüğünün Artırılması

Abstract

Ultrason görüntüleme, tıbbi tanı amaçlı yaygın olarak kullanılan bir yöntemdir; ancak çözünürlüğü, dalga boyu, odak uzaklığı, tarama çizgisi yoğunluğu ve kare hızı gibi etkenlerle sınırlıdır. Yatay (lateral) ve zamansal çözünürlük arasında temel bir ödünleşim bulunmaktadır; tarama çizgisi yoğunluğunun artırılması, mekânsal ayrıntıyı artırırken kare hızını düşürür. Bu çalışma, zamansal çözünürlükten ödün vermeden yatay çözünürlüğü artırmak amacıyla derin öğrenme temelli, özellikle de Oto-Encoder tabanlı bir yaklaşımın potansiyelini araştırmaktadır. Oto-Encoder’ın başarımı; en yakın komşu, lineer ve spline enterpolasyon gibi geleneksel enterpolasyon yöntemlerine karşı yapısal benzerlik (SSIM), tepe sinyal-gürültü oranı (PSNR), çok ölçekli SSIM (MS-SSIM) ve özellik benzerliği (FSIM) metrikleriyle değerlendirilmiştir. Sonuçlar, Oto-Encoder’ın en yüksek SSIM ve FSIM değerlerini elde ederek yapısal bütünlüğü ve özellik korunumunu en iyi şekilde sağladığını göstermektedir. Ayrıca, RF sinyal analizi, Oto-Encoder’ın genel dalga formunu koruduğunu, ancak küçük genlik ve faz sapmalarının mevcut olduğunu ortaya koymaktadır. Bu bulgular, derin öğrenme tabanlı süper çözünürlük yaklaşımlarının, geleneksel çözünürlük ödünleşimlerini en aza indirerek yatay çözünürlüğü etkili bir şekilde artırabileceğini göstermektedir.

Keywords

• Ultrason Görüntüleme
• Çözünürlük Artırımı
• Yatay (Lateral) Çözünürlük
• Derin Öğrenme
• Enterpolasyon

Lateral Resolution Enhancement of Ultrasound Images via Auto-Encoder Network

Abstract

Ultrasound imaging is widely used for medical diagnostics, but its resolution is inherently constrained by factors such as wavelength, focal length, scan line density, and frame rate. A fundamental trade-off exists between lateral and temporal resolution, where increasing scan line density enhances spatial detail at the expense of reduced frame rates. This study explores the potential of deep learning, specifically an AutoEncoder-based approach, to enhance lateral resolution without sacrificing temporal resolution. The performance of the AutoEncoder is evaluated against traditional interpolation methods, including nearest, linear, and spline interpolation, using structural similarity (SSIM), peak signal-to-noise ratio (PSNR), multi-scale SSIM (MS-SSIM), and feature similarity (FSIM) metrics. The results demonstrate that the AutoEncoder outperforms interpolation methods, achieving the highest SSIM and FSIM, indicating superior structural preservation and feature retention. Additionally, the RF signal analysis shows that while the AutoEncoder maintains the overall waveform structure, minor amplitude and phase deviations exist. These findings suggest that deep learning-based super-resolution can effectively enhance lateral resolution while minimizing traditional resolution trade-offs.

Keywords

• Ultrasound Imaging
• Resolution Enhancement
• lateral resolution
• Deep Learning
• Interpolation.

Supporting Institution

This study was supported by the Scientific and Technical Research Council of Turkey (TÜBİTAK) within the scope of the research project under Project Number 122E140.

Ethical Statement

The authors declare that this study complies with Research and Publication Ethics.

References

1. [1] Bing, X., Zhang, W., Zheng, L., & Zhang, Y. (2019). Medical image super-resolution using improved generative adversarial networks. IEEE Access, 7, 145030-145038.
2. [2] Housden, R. J., Gee, A. H., Prager, R. W., & Treece, G. M. (2008). Rotational motion in sensorless freehand three-dimensional ultrasound. Ultrasonics, 48(5), 412-422.
3. [3] Housden, R. J., Treece, G. M., Gee, A. H., & Prager, R. W. (2008). Calibration of an orientation sensor for freehand 3D ultrasound and its use in a hybrid acquisition system. Biomedical engineering online, 7, 1-13.
4. [4] Nguon, L. S., Seo, J., Seo, K., Han, Y., & Park, S. (2022). Reconstruction for plane-wave ultrasound imaging using a modified U-Net-based beamformer. Computerized Medical Imaging and Graphics, 98, 102073.
5. [5] Temiz, H., & Bilge, H. S. (2020). Super-resolution of B-mode ultrasound images with deep learning. IEEE Access, 8, 78808-78820.
6. [6] Mikaeili, M., & Bilge, H. Ş. (2023, November). Evaluating Deep Neural Network Models on Ultrasound Single Image Super Resolution. In 2023 Medical Technologies Congress (TIPTEKNO) (pp. 1-4). IEEE.
7. [7] van Sloun, R. J., Solomon, O., Bruce, M., Khaing, Z. Z., Wijkstra, H., Eldar, Y. C., & Mischi, M. (2020). Super-resolution ultrasound localization microscopy through deep learning. IEEE transactions on medical imaging, 40(3), 829-839.
8. [8] Liu, X., Zhou, T., Lu, M., Yang, Y., He, Q., & Luo, J. (2020). Deep learning for ultrasound localization microscopy. IEEE transactions on medical imaging, 39(10), 3064-3078.

9. [9] Zhang, J., He, Q., Xiao, Y., Zheng, H., Wang, C., & Luo, J. (2021). Ultrasound image reconstruction from plane wave radio-frequency data by a self-supervised deep neural network. Medical Image Analysis, 70, 102018.
10. [10] Rothlübbers, S., Strohm, H., Eickel, K., Jenne, J., Kuhlen, V., Sinden, D., & Günther, M. (2020, September). Improving image quality of single-plane wave ultrasound via deep learning-based channel compounding. In 2020 IEEE International Ultrasonics Symposium (IUS) (pp. 1-4). IEEE.
11. [11] Goudarzi, S., & Rivaz, H. (2022). Deep reconstruction of high-quality ultrasound images from raw plane-wave data: A simulation and in vivo study. Ultrasonics, 125, 106778.
12. [12] Strohm, H., Rothlübbers, S., Eickel, K., & Günther, M. (2020). Deep learning-based reconstruction of ultrasound images from raw channel data. International Journal of Computer Assisted Radiology and Surgery, 15(9), 1487-1490.
13. [13] Li, Z., Wiacek, A., & Bell, M. A. L. (2020, September). Beamforming with deep learning from single plane wave RF data. In 2020 IEEE International Ultrasonics Symposium (IUS) (pp. 1-4). IEEE.
14. [14] Hyun, D., Brickson, L. L., Looby, K. T., & Dahl, J. J. (2019). Beamforming and speckle reduction using neural networks. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 66(5), 898-910.
15. [15] Wang, Y., Kempski, K., Kang, J. U., & Bell, M. A. L. (2020, September). A conditional adversarial network for single plane wave beamforming. In 2020 IEEE International Ultrasonics Symposium (IUS) (pp. 1-4). IEEE.
16. [16] Nair, A. A., Tran, T. D., Reiter, A., & Bell, M. A. L. (2018, April). A deep learning-based alternative to beamforming ultrasound images. In 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) (pp. 3359-3363). IEEE.
17. [17] Simson, W., Göbl, R., Paschali, M., Krönke, M., Scheidhauer, K., Weber, W., & Navab, N. (2019). End-to-end learning-based ultrasound reconstruction. arXiv preprint arXiv:1904.04696.
18. [18] Hyun, D., Brickson, L. L., Looby, K. T., & Dahl, J. J. (2019). Beamforming and speckle reduction using neural networks. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 66(5), 898-910.
19. [19] Wang, Y., Kempski, K., Kang, J. U., & Bell, M. A. L. (2020, September). A conditional adversarial network for single plane wave beamforming. In 2020 IEEE International Ultrasonics Symposium (IUS) (pp. 1-4). IEEE.
20. [20] Goudarzi, S., & Rivaz, H. (2022). Deep reconstruction of high-quality ultrasound images from raw plane-wave data: A simulation and in vivo study. Ultrasonics, 125, 106778.
21. [21] Wasih, M., Ahmad, S., & Almekkawy, M. (2023). A robust cascaded deep neural network for image reconstruction of single-plane wave ultrasound RF data. Ultrasonics, 132, 106981.
22. [22] https://field-ii.dk/