Volumetric Optical Scattering Neural Networks
Abstract
Optical neural networks rely on planar photonic circuits or spatially separated diffractive planes, limiting volumetric integration and requiring precise alignment. Here we propose a volumetric optical scattering neural network (OSNN) comprising densely packed refractive-index neurons with weak contrast Δn = 0.006. Direction-dependent first-order Born scattering provides local coupling, while ordered propagation accumulates these weak updates into global task-specific transformations. Fabrication-aware training and two-photon nanolithography produced a sub-4 × 10−4 mm3 device containing 376,320 neurons at 1.0 × 109/mm3 in a continuously written, intrinsically registered volume. Experimentally, a classifier performed class-selective energy routing with 95.8% blind-test accuracy on MNIST. The calculated passive optical transit time is below 1 ps. A separately optimized imager provided fourfold full-field spatial-sampling compression, resolved 1-μm features and enabled cross-domain transfer. An OSNN co-designed for multiple tasks encoded single-cell fluorescence-derived images into measured output maps, which were pooled into a 16-dimensional representation supporting organelle classification, morphology reconstruction, doublet detection and morphology regression, with all reported metrics exceeding 0.92. Volumetric weak scattering thus establishes an ultracompact, intrinsically registered optical computing architecture, opening a new route to scalable machine vision and biomedical analysis.
Related
- Beyond the Thin-Layer Limit: Differentiable Volumetric Training for Visible-Range Diffractive Neural Networks
- Graph neural network based isomorphic learning of finite element computation for optical scattering problems
- Deep diffractive optical neural networks for detecting skyrmionic topologies of light and their live training
- Compositional neural networks for multimodal X-ray and neutron scattering inversion
- Approaching the Inverse Neutron Scattering Problem with Neural Networks
BibTeX
@misc{luo2026volumetric,
title = {{Volumetric Optical Scattering Neural Networks}},
author = {Xuhao Luo and Qiang Song and Weiwei Cai and Lei Chen and Enbo Yang and Hao Wang and Zhipei Sun and Yueqiang Hu and Joel K. W. Yang and Huigao Duan},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2605.13177},
eprint = {2605.13177},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.13177},
}
RIS
TY - UNPB
TI - Volumetric Optical Scattering Neural Networks
AU - Xuhao Luo
AU - Qiang Song
AU - Weiwei Cai
AU - Lei Chen
AU - Enbo Yang
AU - Hao Wang
AU - Zhipei Sun
AU - Yueqiang Hu
AU - Joel K. W. Yang
AU - Huigao Duan
PY - 2026
DA - 2026/09/13
PB - arXiv
DO - 10.48550/arxiv.2605.13177
UR - https://arxiv.org/abs/2605.13177
AB - Optical neural networks rely on planar photonic circuits or spatially separated diffractive planes, limiting volumetric integration and requiring precise alignment. Here we propose a volumetric optical scattering neural network (OSNN) comprising densely packed refractive-index neurons with weak contrast Δn = 0.006. Direction-dependent first-order Born scattering provides local coupling, while ordered propagation accumulates these weak updates into global task-specific transformations. Fabrication-aware training and two-photon nanolithography produced a sub-4 × 10^−4 mm^3 device containing 376,320 neurons at 1.0 × 10^9/mm^3 in a continuously written, intrinsically registered volume. Experimentally, a classifier performed class-selective energy routing with 95.8% blind-test accuracy on MNIST. The calculated passive optical transit time is below 1 ps. A separately optimized imager provided fourfold full-field spatial-sampling compression, resolved 1-μm features and enabled cross-domain transfer. An OSNN co-designed for multiple tasks encoded single-cell fluorescence-derived images into measured output maps, which were pooled into a 16-dimensional representation supporting organelle classification, morphology reconstruction, doublet detection and morphology regression, with all reported metrics exceeding 0.92. Volumetric weak scattering thus establishes an ultracompact, intrinsically registered optical computing architecture, opening a new route to scalable machine vision and biomedical analysis.
ER -