Keywords: Neural Fields, Ultrasound, 3D Reconstruction, Meta-Learning
TL;DR: We show how to effectively represent 3D-US data with Neural Fields, where we first learn the 2D slices of the 3D ultrasound data and expand to 3D.
Abstract: 3D Ultrasound (3D-US) is a powerful imaging modality, but the high storage requirement and low spatial resolution challenge wider adoption. Recent advancements in Neural Fields suggest a potential for efficient storage and construction of 3D-US data. In this work, we show how to effectively represent 3D-US data with Neural Fields, where we first learn the 2D slices of the 3D ultrasound data and expand to 3D. This two-stage representation learning improves the quality of 3D-US in terms of Peak Signal-to-Noise Ratio (PSNR) to 31.84dB from 28.7dB, a significant improvement directly noticeable to the human eye.
Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
Paper Type: novel methodological ideas without extensive validation
Primary Subject Area: Image Acquisition and Reconstruction
Secondary Subject Area: Unsupervised Learning and Representation Learning
Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.