A High-Throughput Analysis Method to Detect Regions of Interest and Quantify Zebrafish Embryo Images

Abstract

Zebrafish is widely used to understand neural development and model various neurodegenerative diseases. Zebrafish embryos are optically transparent, have a short development period, and can be kept alive in microplates for days, making them amenable to high-throughput microscopic imaging. As a result of high-throughput experiments, a large number of images can be generated in a single experiment, posing a challenge to researchers to analyze them efficiently and quantitatively. In this work, we develop an image processing focused on detecting and quantifying pigments in zebrafish embryos. The algorithm automatically detects a region of interest (ROI) enclosing an area around the pigments and then segment the pigments for quantification. In this process, the algorithm identifies the head and torso at first, and then finds the boundaries corresponding to the back and abdomen by taking advantage of a priori information about the anatomy of zebrafish embryos. The method is robust in terms that it can detect and quantify pigments even when the embryos have different orientations and curvatures. We used real data to demonstrate the performance of the method to extract phenotypic information from zebrafish embryo images and compared its results with manual analysis for verification.

Automated Phenotype Recognition for Zebrafish Embryo Based In Vivo High Throughput Toxicity Screening of Engineered Nano-Materials

“…The major advantages of using zebrafish for HTS toxicity studies include: (a) large number of embryos can be obtained at low cost, (b) zebrafish embryos undergo rapid development from eggs to larvae in three days, (c) zebrafish embryos and larvae can be kept alive in micro-plates for days, and (d) zebrafish embryos and larvae are close to being optically transparent [25], [26]. As the application of zebrafish-based toxicity assays expands in HTS studies, researchers will be confronted with the challenge of efficiently resolving/extracting the latent semantics (e.g., phenotypic maldevelopment of zebrafish embryos in exposure to eNMs) embedded in the potential large number of images being generated in a single experiment [25].…”

“…As the application of zebrafish-based toxicity assays expands in HTS studies, researchers will be confronted with the challenge of efficiently resolving/extracting the latent semantics (e.g., phenotypic maldevelopment of zebrafish embryos in exposure to eNMs) embedded in the potential large number of images being generated in a single experiment [25]. In order to isolate and quantify the image based data, the majority of the published studies on zebrafish high throughput screening have resorted primarily to fluorescence-based microscopy using specifically developed transgenic zebrafish lines (e.g., Tg(fli1:EGFP)) [27]–[32].…”

“…On the other hand, for non-fluorescence based HTS, the usual grayscale image analysis is significantly more challenging. Recently, a bright-field (grayscale) zebrafish image analysis algorithm, based on a heuristic approach, was proposed that detects and segments a region enclosing an area surrounding the pigments [25] (a.k.a., the Region of Interest, ROI). The pigmentation in the ROI could reflect the response of the zebrafish embryos to various environmental cues [25].…”

“…Recently, a bright-field (grayscale) zebrafish image analysis algorithm, based on a heuristic approach, was proposed that detects and segments a region enclosing an area surrounding the pigments [25] (a.k.a., the Region of Interest, ROI). The pigmentation in the ROI could reflect the response of the zebrafish embryos to various environmental cues [25]. In the above approach, the ROI was detected from images acquired from 24-well plates with the help of a priori anatomical information of zebrafish embryos.…”

“…The approach was tested using 18 images of zebrafish embryos treated with dimethyl sulfoxide and gamma secretase inhibitor (GSI-18) and resulted in false positive and negative identification rates (compared to a manual analysis) of 28.6% and 37.5%, respectively. The authors indicated that their image analysis approach was difficult to generalize to different size plates since the algorithm used was specific to the image size and resolution [25].…”