PaFSe: A Parameter-Free Segmentation Approach for 3D Fluorescent Images

Abstract Confocal fluorescent microscopy is a major tool to investigate the molecular orchestration of biomedical samples. The quality image acquisition depends critically on tissue and thickness, type, concentration antibodies used, as well microscope parameters. Due these factors, intra-sample inter-sample variability inevitably arises. Segmentation quantification targeted proteins can thus become challenging process. Image processing techniques need therefore address acquisitions minimize risk biases originating from changes in signal intensity, background noise, parameterization. Here, we introduce PaFSe, parameter-free segmentation algorithm for 3D images. based our established PRAQA approach, which evaluates dispersion several pixel intensity neighborhoods allowing statistical assessment whether individual subfields an be considered positive or background. PaFSe extends by fully automatic estimate parameters, thereby provides completely robust algorithm. By comparing with Ilastik synthetic examples, show that method achieves similar performances supervised approach low-to-moderate noise environments without tedious training. Furthermore, validate efficiency segmenting quantifying abundance hyperphosphorylated Tau protein post-mortem human brain samples Alzheimer’s disease patients age-matched controls, where obtain values highly correlated manual neuropathological segmentation. parameter-free, fast, adaptive complex images freely available at https://doi.org/10.17881/j20h-pa27 .