Rapid and reliable image analysis pipeline for semi-automated quantification of CNS cell types in MATLAB

Studying CNS cell responses is essential for understanding disease, injury, and developing effective therapies. While immunofluorescence and transgenic reporter models allow for specific labeling, automated quantification remains difficult due to tissue heterogeneity. Consequently, most analyses are conducted manually, introducing user bias and limiting reproducibility.

We developed a MATLAB-based semi-automated workflow for quantifying immunofluorescence-stained CNS cells, focusing on nuclear signal detection. The pipeline uses DAPI masking and the imfindcircles function to detect round nuclei, requiring minimal user input.

The pipeline enabled robust quantification of CNS-resident cells. Automated analyses of brain and spinal cord tissue sections closely resembled manual quantification, with minimal error. In a mouse model of contusion spinal cord injury, it revealed a rostro-caudal decline in myelinating oligodendrocytes from the lesion epicenter, confirming the method's accuracy and sensitivity in detecting injury-induced cellular changes.

Unlike many commonly used quantification-based software, this novel pipeline does not perform full image segmentation. Instead, it uses nuclear morphology to detect round shapes. Moreover, the pipeline has been specifically designed and optimized for the quantification of CNS cells, whose heterogeneity and cytoarchitecture pose specific challenges to existing methods that are more generalized.

This study presents an alternative to classical segmentation models by offering a reproducible quantification of CNS-resident cells using nuclear morphology. Its simplicity, minimal input requirements, reduced time for semi-automated quantification, and specificity for CNS tissues make it a valuable tool for studying cellular responses in both healthy and pathological contexts.