# Improved Detection of Capillaries in High‐Resolution Handheld Vital Microscopy by Use of the MicroTools Advanced Computer Vision Algorithm

## TL;DR

A new algorithm improves capillary detection in high-resolution handheld microscopy images, offering better accuracy and broader field of view.

## Contribution

The study introduces a new stabilization algorithm and validates full-frame IDF analysis for improved microcirculatory measurements.

## Key findings

- Full-frame IDF analysis increased FOV by 200% and pixel density by 66%, improving capillary detection.
- MicroTools' new stabilization algorithm performed comparably to existing methods.
- IDF images showed lower red blood cell velocity compared to SDF images.

## Abstract

Handheld vital microscopy (HVM) enables bedside visualization of the microcirculation, with major improvements from sidestream darkfield (SDF) to incident dark field (IDF) imaging. Although IDF offers high‐resolution images, standard analysis methods require down‐sampling to match SDF's lower field of view (FOV) and pixel density. MicroTools, an automated stabilization and analysis algorithm, has previously been validated for use with SDF image sequences. This study aimed to assess the accuracy and precision of microcirculatory analysis using MicroTools on full‐frame, high‐resolution IDF images.

Image sequences from a previous study were re‐analyzed in both IDF and down‐sampled SDF formats. Microcirculatory parameters—including total vessel density (TVD), functional capillary density, and red blood cell velocity (RBCv)—were compared between formats. MicroTools' new stabilization algorithm was also evaluated against existing algorithms.

Full‐frame IDF analysis increased the FOV by 200% and pixel density by 66%, enhancing capillary detection and TVD measurements. RBCv values were lower with IDF images. The updated stabilization algorithm showed performance comparable to prior methods.

MicroTools analysis of full‐frame IDF images improved measurement accuracy and vessel detection, with excellent agreement to standard software. Its new stabilization algorithm proved equally robust, supporting broader adoption in clinical research.