P1.15 - Data-driven frequency-domain full waveform inversion for ultrasonic breast imaging
- Event
- 2025 ICU PADERBORN - 9th International Congress on Ultrasonics
2025-09-21 - 2025-09-25
Paderborn - Band
- Posters
- Chapter
- Posters
- Author(s)
- M. Qin, Y. Zhao, H. Yin - Harbin Institute of Technology, Harbin (China)
- Pages
- 332 - 335
- DOI
- 10.5162/Ultrasonic2025/P1.15
- ISBN
- 978-3-910600-08-9
- Price
- free
Abstract
Frequency-domain full waveform inversion is a high-resolution ultrasound method. It shows potential for breast imaging. However, it requires heavy computation and storage. These limitations restrict clinical use. We propose an end-to-end deep learning framework. This framework is constructed upon the encoder-decoder architecture of convolutional neural networks (CNNs) and is capable of directly mapping ultrasonic field data to the corresponding sound velocity distribution. Numerical simulations verify its effectiveness and efficiency.