3D Tissue Reconstruction and Generation for Single-Cell Spatial Transcriptomics using Neural Radiance Fields
Keywords: Single-Cell Spatial Transcriptomics, Neural Radiance Fields, 3D reconstruction
Abstract: Single-cell spatial transcriptomics (scST) is a groundbreaking technique that allows for the exploration of gene expression patterns, cell-cell interactions, and tissue organization at the single-cell level. Traditional approaches in scST reconstruction mainly focus on assigning two-dimensional (2D) coordinates to individual cells within a pre-established region. This often requires a large amount of 2D slice data, such as ssDNAs images, which escalates both costs and the complexity involved in studying and reconstructing the tissue's three-dimensional (3D) organization. Here, we introduce a novel method for scST reconstruction, which is a Neural Radiance Fields (NeRF)-based 3D-aware generative model termed STscan, that aims to reconstruct a 3D scST scene using a minimal amount from 2D images (fewer than 10). Additionally, STscan can identify cell types and their expression levels within this 3D environment. To the best of our knowledge, STscan is the first NeRF-based method specifically designed for single-cell ST reconstruction, and it is the first end-to-end solution capable of directly reconstructing in vitro cell-cell environments from ssDNA images. This approach has the potential to significantly reduce both the complexity and cost associated with scST studies.
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 5715
Loading