Caltech Team Develops Inexpensive Microscope Autofocus Technique

Adjusting a microscope to achieve sharp focus demands time and precision. Moving a slide blurs the image, requiring users to pause and turn a knob for clarity. For researchers handling dozens or hundreds of samples, this process—even when semi-automated—requires considerable time and effort.

Now, scientists at Caltech have introduced a simple, durable solution relying on two LED lights and physics-guided computation. Detailed in a paper published in Nature Communications, the technique addresses longstanding focusing challenges in microscopy. Graduate students Haowen Zhou and Shi "Josh" Zhao led the effort in the lab of Changhuei Yang.

The method works by illuminating samples with LEDs from slightly different angles. Merging the resulting two images uncovers a pattern of stripes, which varies predictably with distance from the focal point—the optimal spot for clear imaging. Software analyzes these stripes to direct focus corrections automatically. This insight arose unexpectedly during troubleshooting. "We were debugging for another project," Zhou says. "When we summed up the contributions from the photos taken from two different locations, we found this fringe pattern. If we defocused more, the fringe would be much denser. If we defocused less, it would be more spread out. So, it seemed there was a strong correlation between the defocus value and the fringe density."

Extensive testing spanned six microscope types, from basic compound light models to sophisticated ones for live cells, tissues, and thick 3D specimens, delivering consistent success. With thin, flat samples, it sustained focus over a range more than 400 times the natural depth of a standard lens. "Our technique, which is enabled by a physics-based observation, is reliable, high performance, and also very simple," Zhou says. "This makes it useful and powerful for automated, high-throughput microscopy." For 3D samples up to 150 micrometers thick, the range reached nearly 300 times beyond typical limits. "This offers truly robust autofocusing of 3D samples, which has never been possible with other techniques," Zhao adds.