Figure 1.
Our imaging system consists of imaging hardware and the QuantWorm software package. The hardware is composed of a microscope equipped with a digital camera and a motorized stage. The image acquisition software is used to control the stage and take images and videos. Four image analysis software programs are used to analyze body size, lifespan, egg laying, and locomotion from the images or videos. Once the software finishes the fully automated image analysis, a user can correct errors in the manual inspection window. Both native image processing algorithms and ImageJ API/library are used to process images.
Figure 2.
QuantWorm image processing algorithms.
(A) WormLifespan. Two time-lapse images are obtained, and a differential image is created by subtracting the second time-lapse image from the first time-lapse image. Independently, individual worms are detected in the second image and are defined as region of interest (ROI). Worm movement is determined by counting the number of white pixels. (B) WormLocomotion. Image frames from videos are binarized and region-extracted to detect objects. Worms are indentified by analyzing the morphology of detected objects. An individual worm track is constructed by connecting all centroid points of a moving worm. (C) WormLength. Source image is binarized, and worms are detected by region extraction and shape analysis. Once a worm object is identified, a skeleton curve is created through the middle of the worm. The length of the worm is calculated by measuring the length of the skeleton curve (D) WormEgg. Single eggs are detected by applying edge detection, gap filling, flood filling, and morphology analysis. Egg detection parameters are determined by analyzing the detected single eggs. Multi-thresholding binarization is applied to create multiple binary images from which eggs are detected. Results are compiled to conduct clustering to identify highly probable eggs and remove duplicate findings.
Figure 3.
(A) WormLifespan. Moving worms were manually counted by aspirating individual worms from a well under a basic light microscope after images were captured (n = 26 wells). (B) WormLocomotion. Worm simulation videos were created and then analyzed by the WormLocomotion software (n = 11 videos). (C) WormLength. In the manual method, worm length was manually measured from images using an Adobe Photoshop length measurement tool (n = 46 worms). (D) WormEgg. In the manual method, eggs were manually counted by aspirating eggs from a well after images were taken (n = 42 wells). The diagonal line represents the ideal case where the computer measurements equal the manual measurements.
Figure 4.
A lifespan assay was conducted using the manual method (A) or using the QuantWorm system (B). Worms were dosed every other day with 7 µmol/L agar celastrol or the solvent DMSO as a control. n represents the number of worms. **p<0.05; ***p<0.001; p-value by log rank test.
Figure 5.
Phenotypes of Gαq pathway mutants.
(A) Survival curves. The results are from two independent experiments. At least triplicate wells were used (n >180 worms for each strain). (B) Mean lifespan. (C) Worm speed measured as the sum of average worm velocities in individual videos divided by the number of videos. With food: For each strain, videos (n ≥ 10 videos) collected from four independent experiments were analyzed. Without food: For each strain, videos (n ≥ 5 videos) collected from three independent experiments were analyzed. (D) Distribution of individual average speeds of detected tracks (n ≥ 487 tracks for each strain) with food (Day 1∼3 of adulthood). (E) Body length. Worms at 1 day of adulthood were used (n ≥ 113 for each strain). (F). Egg laying rate. Worms at 28 hr of adulthood were used. Shown is a combined result from two independent experiments with ∼10 hermaphrodites per well (n ≥ 12 wells for each strain). *p<0.01; **p<0.05; ***p<0.001; p-value by log rank test (Figure B) and t-test (Figure C, E, and F).