1023B Poster - 16. Techniques and technology
Friday April 08, 2:00 PM - 4:00 PM

Authors:
Heidi Pipkin ¹; Adam Smiley ²; Andrew Arsham ¹

Affiliations:
1) Bemidji State University; 2) University of Minnesota

Keywords:
h. high-throughput phenotyping; f. classroom undergraduate research experience (CURE)

Collection of reproducible high quality quantitative data on whole flies is challenging, especially in training environments where data collection is distributed among many individuals. Common adjustable light sources for dissecting microscopes create glare, shadow, and irreproducible lighting conditions; microscope-mounted cameras are optimized for low noise and high sensitivity but are often low resolution and can only record 1-2 flies at a time; and only a small portion of the fly is in focus. To address these limitations, we assembled a high-resolution imaging method with fixed lighting, staging, and exposure conditions. Whole flies are frozen and imaged without additional fixation or mounting medium. A 3D printed stage standardizes fly placement and reduces preparation time. Batch collection and computer automation of focus stacking eliminates inter-image and inter-operator variability eliminating the need for complex, irreproducible, and ethically murky image manipulation to adjust color, glare, and similar artifacts. A 50MP full frame digital camera, computer-controlled stepper motor, and focus-stacking software generates a single image of dozens of flies with all ommatidia, bodies, wing veins, and bristles in focus. These standardized composite images facilitate a many-fold reduction in imaging time and precise quantitative comparison of phenotypes involving body and organ color, size, shape, and pattern. Semi-automated data extraction and analysis using ImageJ and R further facilitate reproducibility, code-sharing, and flexibility and reduce the confusion and workflow complexity created by intermediary data products (crops, filters, masks, format conversions, etc.). In addition to being accessible to resource-limited labs, these methods are classroom-friendly and support undergraduate learning outcomes in image analysis, data visualization, computational biology, and statistics.