Using DeepLabCut as a Real-Time and Markerless Tool for Cardiac Physiology Assessment in Zebrafish

Abstract

DeepLabCut (DLC) is a deep learning-based tool initially invented for markerless pose estimation in mammals. In this study, we explored the possibility of adopting this tool for conducting markerless cardiac physiology assessment in an important aquatic toxicology model of zebrafish (Danio rerio). Initially, high-definition videography was applied to capture heartbeat information at a frame rate of 30 frames per second (fps). Next, 20 videos from different individuals were used to perform convolutional neural network training by labeling the heart chamber (ventricle) with eight landmarks. Using Residual Network (ResNet) 152, a neural network with 152 convolutional neural network layers with 500,000 iterations, we successfully obtained a trained model that can track the heart chamber in a real-time manner. Later, we validated DLC performance with the previously published ImageJ Time Series Analysis (TSA) and Kymograph (KYM) methods. We also evaluated DLC performance by challenging experimental animals with ethanol and ponatinib to induce cardiac abnormality and heartbeat irregularity. The results showed that DLC is more accurate than the TSA method in several parameters tested. The DLC-trained model also detected the ventricle of zebrafish embryos even in the occurrence of heart abnormalities, such as pericardial edema. We believe that this tool is beneficial for research studies, especially for cardiac physiology assessment in zebrafish embryos.

Uncovering developmental time and tempo using deep learning

“…As a consequence of structural and temporal variation, characterization of embryonic development and the transitions between morphological states remains subjective. Computer-driven methods have been proposed to tackle this problem and to enable standardization by addressing structural or temporal variability [27][28][29][30][31][32][33][34][35] . However, approaches based on supervised machine-learning techniques require large databases, training resources and human-assisted annotation.…”

Accurate prediction of neurologic changes in critically ill infants using pose AI

“…Most are trained on adults and have poor performance on infants, likely because infant body proportions are different from adults. 21 We chose DeepLabCut for its generalizability to many animals, 16,22 including humans, 23 and robots. 24 DeepLabCut uses transfer learning to leverage ResNet, 25 which is trained on >1 million images, for pose tracking.…”

Bringing Artificial Intelligence (AI) into Environmental Toxicology Studies: A Perspective of AI-Enabled Zebrafish High-Throughput Screening

“…While the integration of algorithms like Inception and YOLO with automated mechanical devices for real-time positioning and posture adjustment in zebrafish larvae is innovative, one must consider the potential for biases introduced by these automated systems. Through advanced machine learning algorithms used in analyzing microscopic video, real-time segmentation of the zebrafish atria and ventricles as well as dynamic reconstruction of heart and blood flow can be performed by analyzing the microscopic video, thus enabling the analysis of cardiac function indicators like heart rate, end-diastolic volume, end-systolic volume, ejection fraction, and shortening fraction. ,, Yet, the reliance on these advanced algorithms necessitates a thorough evaluation of their accuracy and the potential for overfitting or misinterpretation of complex biological phenomena. The development of a CNN-based multichannel feature learning model for neutrophil tracking and the use of the random forest algorithm for clustering analysis of drug-induced changes in neuronal activity further illustrate the growing dependence on computational methods in biological research. − …”