Improving Agar Plate Image Quality Using a Low‐Cost 3D‐Printed Lightbox: A Proof‐of‐Concept Evaluation for Machine Learning–Assisted Analysis

Agar plate photography is widely used to document microbial growth, but images captured under routine laboratory lighting are often affected by glare, reflections, shadows, and uneven illumination, which may compromise both visual interpretation and computational analysis. This proof-of-concept study developed and evaluated a low-cost, three-dimensional (3D)-printed lightbox for standardized smartphone-based agar plate imaging. Image characteristics were assessed across five uncultured agar media under traditional and lightbox conditions using Fiji/ImageJ. Machine learning-assisted analysis was then performed on cultured horse blood agar plates using QuPath. Separate Random Trees pixel classifiers were developed for traditional and lightbox images to distinguish colonies from background agar and noncolony artifacts. A semiautomated QuPath-assisted workflow was also compared with manual colony counts across 14 growth-positive horse blood agar plates. Lightbox imaging reduced reflective artifact, improved illumination consistency, and enhanced visualization of colony margins and hemolysis-associated changes. Traditional-lighting images produced more artifact-related classifier detections, including glare, reflections, and uncolonized agar being detected as colony-like regions. In the exploratory colony-count comparison, semiautomated QuPath-assisted counts showed close agreement with manual reference counts (Pearson r = 0.999; mean absolute error = 2.4 colonies per plate). These findings demonstrate that a low-cost, 3D-printed lightbox can improve the quality and standardization of smartphone-based agar plate images and support more interpretable machine learning-assisted analysis. Further validation using larger independent data sets, additional organisms, media types, and imaging devices is required.