Development of Python-Based Software and ESP32-Based Remote Control for Handheld In Vivo Microscopes

Abstract

Handheld in vivo microscopy is a technique that enables imaging of cellular details of living tissues using a portable handheld device without the need for large, bench-based imaging systems. It offers a compact and portable solution for real-time imaging in a wide range of healthcare settings. Due to its high resolution, however, handheld in vivo microscopy is susceptible to motion artifacts. For example, users must hold the device steady while simultaneously controlling computer software to capture data or adjust the position of the objective lens. This divided attention often leads to small movements of the hand holding the device, which introduce motion artifacts that blur fine structural details and compromise the diagnostic quality of the images. In this thesis, an open-source, Python-based graphical user interface (GUI) and an ESP32-based remote control system are developed to enhance the stability and accessibility of handheld in vivo microscopy systems. The GUI has been developed to replace expensive and inflexible proprietary software like LabVIEW and provide intuitive controls for camera settings, live view configuration, and precise motor control while supporting offline operation. Additionally, the ESP32-based remote has been developed to enable single-user control of key functions, including motor movement, image capture, and axial scans, directly on the device, minimizing the need to interact with a computer during imaging. The effect of the remote on image stability was evaluated by imaging human skin both with and without the remote. A stack of multiple in vivo microscopy images was acquired while the objective lens was axially scanned through the tissue depth. The image stack was spatially registered for two consecutive image frames. The spatial registration showed that microscopy images captured with the remote required smaller shift corrections, averaging 0.96 µm per frame, compared to 1.33 µm per frame without the remote. This reduction in frame-to-frame displacement suggests that the remote-assisted scans exhibited improved stability, with significantly fewer motion artifacts in the registered image stacks. Additionally, the Python-based software was evaluated against the existing LabVIEW implementation in terms of usability and performance. The Python code successfully replicated all core functionalities of the LabVIEW interface while introducing additional features for improved control and flexibility. To validate performance, the framerate was tested by recording a 1 Hz signal. The software consistently captured data at the expected 30 frames per second (fps), confirming that it meets real-time acquisition requirements. The results demonstrate that integrating a remote has the potential to improve the usability and image quality of handheld microscopes. By addressing stability and accessibility, this work provides a practical solution that can accelerate the deployment of handheld in vivo microscopy tools in diverse healthcare settings.