OpenLM: Open-Source Pixel Super-Resolution Lens-Free Microscopy for Early Bacterial Colony Detection

Rapid and on-field visualization of early-stage bacterial growth is limited by the bulky optics, skilled operation, and high costs of conventional microscopes. We present a compact, Open-source Lens-free Microscopy (OpenLM) platform that achieves high-resolution, large field of view (10.16 mm2) imaging using inexpensive hardware and computational imaging. The system comprises a Raspberry Pi CMOS image sensor (1.12 µm pixels, 3280×2464), an 8x8 LED array, and a 4 nm-bandwidth 532 nm filter housed in a 3D-printed enclosure. Custom-developed Python GUIs for Raspberry Pi and Windows provide platform control, automated image capture, and pixel super-resolution (PSR) reconstruction. Sequential illumination with the central 25 LEDs introduces controlled sub-pixel shifts that are registered and fused to form a high-resolution hologram, followed by angular-spectrum back-propagation and digital refocusing. With 25 frames in ~50 s per time point, OpenLM improves measured spatial resolution from 1.95 µm to 0.87 µm across its wide field, enabling time-lapse imaging without mechanical focusing. Prototyping materials cost ~$284, and the design (hardware, control code, and image processing) is fully reproducible. We demonstrated two use cases: (1) Colony growth monitoring: E. coli colonies on custom shallow dishes become visible within 30 min at room temperature, the large FOV supports long-term observation of spatial dynamics and colony–colony interactions. (2) Early detection with AI, a YOLOv11s detector trained on >100k annotated instances, identifies colonies as early as ~35 min, achieving precision 0.937, recall 0.908, mAP@0.5 0.965 with ~3 s inference per image on a desktop GPU. OpenLM combines accessible hardware, PSR-enhanced holography, and user-friendly software to deliver reproducible, large-FOV, sub-micron-scale imaging and rapid colony detection suitable for a variety of microbiological applications.