Bazow Brad, Phan Thuc, Raub Christopher B, Nehmetallah George
Opt Express. 2023 Aug 14;31(17):28382-28399. doi: 10.1364/OE.491707.
Optical diffraction tomography (ODT) solves an inverse scattering problem to obtain label-free, 3D refractive index (RI) estimation of biological specimens. This work demonstrates 3D RI retrieval methods suitable for partially-coherent ODT systems supported by intensity-only measurements consisting of axial and angular illumination scanning. This framework allows for access to 3D quantitative RI contrast using a simplified non-interferometric technique. We consider a traditional iterative tomographic solver based on a multiple in-plane representation of the optical scattering process and gradient descent optimization adapted for focus-scanning systems, as well as an approach that relies solely on 3D convolutional neural networks (CNNs) to invert the scattering process. The approaches are validated using simulations of the 3D scattering potential for weak phase 3D biological samples.
光学衍射层析成像（ODT）通过解决逆散射问题来获取生物样本的无标记三维折射率（RI）估计。这项工作展示了适用于部分相干ODT系统的三维RI检索方法，该系统由轴向和角度照明扫描组成的仅强度测量支持。这个框架允许使用简化的非干涉技术获得三维定量RI对比度。我们考虑一种基于光学散射过程的多平面表示和适用于聚焦扫描系统的梯度下降优化的传统迭代层析求解器，以及一种仅依赖三维卷积神经网络（CNN）来反转散射过程的方法。这些方法通过对弱相位三维生物样本的三维散射势进行模拟来验证。
