NemaDIm: Detection and Image Capture of C. elegans

Abstract

The Gourgou Research Group is seeking a method of detecting nematodes (C. elegans) and capturing an image as they pass from one chamber, through a hallway, and into a second chamber. This behavioral environment is 3D printed and filled with Nematode Growth Medium (NGM), an agar gel poured hot into a mold and then cooled to ambient temperature. The final design implemented consisted of a Raspberry Pi 4 processor, a Raspberry Pi HQ Camera, a 16mm telephoto lens, and arena housing to control external factors. The system runs over the course of an experiment and streams images from the camera to the Raspberry Pi. The Pi then implements an image detection algorithm that utilizes adaptive thresholding as well as image binarization to compute a pixel ratio (with a calibrated offset) for a given image. If this pixel ratio is above a threshold (0.0025) then the image is classified as positive for the presence of C. elegans. The positively classified images are saved to a file directory and the negatives discarded. Samples of cropped and original positive images can be seen above. The system couldn’t be successfully integrated as the lenses tested didn’t have a small enough Minimum Object Distance to allow close up images of the worms while still being focused. The scripts for nematode detection and file organization were implemented in Matlab and can be used to process videos from the Olympus DP23 camera currently used in the lab. With this post-processing system the sensor was able to detect and organize the video frames with zero false negatives and <10% false positives.