Saved in:
Bibliographic Details
Main Authors: He, Feng, Tan, Guodong, Li, Qiankun, Yu, Jun, Wen, Quan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.22577
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909870118141952
author He, Feng
Tan, Guodong
Li, Qiankun
Yu, Jun
Wen, Quan
author_facet He, Feng
Tan, Guodong
Li, Qiankun
Yu, Jun
Wen, Quan
contents Light field microscopy (LFM) has become an emerging tool in neuroscience for large-scale neural imaging in vivo, notable for its single-exposure volumetric imaging, broad field of view, and high temporal resolution. However, learning-based 3D reconstruction in XLFM remains underdeveloped due to two core challenges: the absence of standardized datasets and the lack of methods that can efficiently model its angular-spatial structure while remaining physically grounded. We address these challenges by introducing three key contributions. First, we construct the XLFM-Zebrafish benchmark, a large-scale dataset and evaluation suite for XLFM reconstruction. Second, we propose Masked View Modeling for Light Fields (MVN-LF), a self-supervised task that learns angular priors by predicting occluded views, improving data efficiency. Third, we formulate the Optical Rendering Consistency Loss (ORC Loss), a differentiable rendering constraint that enforces alignment between predicted volumes and their PSF-based forward projections. On the XLFM-Zebrafish benchmark, our method improves PSNR by 7.7% over state-of-the-art baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2510_22577
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy
He, Feng
Tan, Guodong
Li, Qiankun
Yu, Jun
Wen, Quan
Computer Vision and Pattern Recognition
Light field microscopy (LFM) has become an emerging tool in neuroscience for large-scale neural imaging in vivo, notable for its single-exposure volumetric imaging, broad field of view, and high temporal resolution. However, learning-based 3D reconstruction in XLFM remains underdeveloped due to two core challenges: the absence of standardized datasets and the lack of methods that can efficiently model its angular-spatial structure while remaining physically grounded. We address these challenges by introducing three key contributions. First, we construct the XLFM-Zebrafish benchmark, a large-scale dataset and evaluation suite for XLFM reconstruction. Second, we propose Masked View Modeling for Light Fields (MVN-LF), a self-supervised task that learns angular priors by predicting occluded views, improving data efficiency. Third, we formulate the Optical Rendering Consistency Loss (ORC Loss), a differentiable rendering constraint that enforces alignment between predicted volumes and their PSF-based forward projections. On the XLFM-Zebrafish benchmark, our method improves PSNR by 7.7% over state-of-the-art baselines.
title From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.22577