Volumetric Segmentation for Integral Microscopy with Fourier Plane Recording

Light Field (LF) microscopy has emerged as a fast growing field of interest in the last two decades for its undoubted capacity of capturing in-vivo samples from multiple perspectives. In this work we present a framework for Volumetric Segmentation of LF images created following the setup of a Fourier Integral Microscope (FIMic). In the proposed framework, we convert the FIMic-captured LF into a three-dimensional Focal Stack (FS) to be used as an input to machine learning models with the aim to get the 3D locations of the specimen of interest. Using a synthetic dataset generated in Blender, we train three neural networks based on the U-Net architecture and merge their outputs to achieve the desired volumetric segmentation. In our main test results we achieve a precision of more than 95%, while in the related tests we still achieve a value higher than 80%.