An Enhanced Mountain Climbing Search Algorithm to Enable Fast and Accurate Autofocusing in High Resolution Fluorescence Microscopy

Abstract

Accurate and efficient autofocusing is essential for the automation of fluorescence microscopy, but background noise and shallow depth of field at high magnifications make autofocusing particularly challenging. Here, we present a fast and accurate autofocus algorithm to address these challenges. It is highly effective for high-magnification imaging, while performing equally well for low-magnification imaging tasks. The method is based on the mountain climbing search algorithm and yields improvements on autofocusing precision of up to 200-fold over current methods, whilst offering competitive speed and greatly extended search ranges. Our approach is broadly applicable: it demonstrated good stability and reproducibility across magnifications ranging from 20X to 100X, excels in both live cell imaging and high-resolution fixed sample imaging, and it is compatible with various microscopy techniques without the need for fiducial markers or hardware modifications on existing microscopes. To maximise its accessibility, we constructed a user-friendly interface compatible with the widely used Micromanager software. It generalises well across various imaging modalities and hardware platforms, making it particularly suitable for use in high-resolution screening of candidate drugs.