Zachary Little

Zachary Little

Years participated in RESESS:

Download PDF »

An Overview

Major: Environmental Science, Geography, GIS Science
Academic Affiliation: University of Maryland Baltimore County
Research Mentor: Sarah Spaulding
Communications Mentor: Magali Barba


Growing up in Baltimore, Maryland, Zachary became fascinated in geoscience during his childhood explorations along the shorelines of Ocean City, Maryland. Zachary’s curiosity in coastal processes and the natural environment has inspired him to pursue a career as a professor and researcher of coastal and fluvial geomorphology. This summer as a RESESS intern, he focused on developing a new method to automate particle analysis processing of diatoms, a type of single-celled algae. His work is the first to use VisualSpreadsheet software, independent of its complementary FlowCam hardware, to conduct automated particle analysis on microscope images of diatoms and will contribute to techniques of diatom enumeration employed by academic institutions and federal agencies.


Automated Diatom Analysis Applied to Traditional Light Microscopy: A Proof-of-Concept Study

Diatom identification and enumeration by high resolution light microscopy is required for many areas of research and water quality assessment. Such analyses, however, are both expertise and labor-intensive. These challenges motivate the need for an automated process to efficiently and accurately identify and enumerate diatoms. Improvements in particle analysis software have increased the likelihood that diatom enumeration can be automated. VisualSpreadsheet software provides a possible solution for automated particle analysis of high-resolution light microscope diatom images. We applied the software, independent of its complementary FlowCam hardware, to automated analysis of light microscope images containing diatoms. Through numerous trials, we arrived at threshold settings to correctly segment 67% of the total possible diatom valves and fragments from broad fields of view. (183 light microscope images were examined containing 255 diatom particles. Of the 255 diatom particles present, 216 diatoms valves and fragments of valves were processed, with 170 properly analyzed and focused upon by the software). Manual analysis of the images yielded 255 particles in 400 seconds, whereas the software yielded a total of 216 particles in 68 seconds, thus highlighting that the software has an approximate five-fold efficiency advantage in particle analysis time. As in past efforts, incomplete or incorrect recognition was found for images with multiple valves in contact or valves with little contrast. The software has potential to be an effective tool in assisting taxonomists with diatom enumeration by completing a large portion of analyses. Benefits and limitations of the approach are presented to allow for development of future work in `image analysis and automated enumeration of traditional light microscope images containing diatoms.