Language :
According to statistics, cardiovascular disease is the leading cause of death globally and is ranked second among the top ten causes of death in Taiwan. Professor Liao Yi-Han of the Department of Applied Chemistry and the Institute of Molecular Sciences at National Yang Ming Chiao Tung University, along with his team, has combined biomedical imaging and AI technology to reconstruct dynamic 3D images of the living zebrafish heart chamber and heart calcium ions. They have then simultaneously obtained important physiological information such as parameters of ventricular contraction function, calcium ion transient changes in the central muscle cells of the beating heart, and cardiac electrophysiological signal conduction, creating a faster platform for heart physiological and pathological research.
The zebrafish's genes and heart physiology are similar to humans, making it a commonly used model organism in heart-related research. However, due to the small size (about 200 micrometers) and fast heartbeat (about 200 times per minute, much faster than the human heart rate of 60-100 beats per minute) of larval zebrafish heart, accurate measurement of ventricular systolic function and 3D dynamic imaging of cardiac calcium ions are a great challenge. In addition, the processing and analysis of 3D dynamic images requires a lot of time and manpower. Dr. Liao's team used their own multi-layered fluorescence imaging system, first segmenting the dynamic 2D images of the zebrafish heart, then identifying the boundary of the heart chamber and reproducing the dynamic 3D animation of the heart chamber based on the periodic heartbeat characteristics. Leadtek NVIDIA RTX professional graphics cards not only improve the efficiency of image recognition and data processing, but also reduce the uncertainty of manual processing, greatly accelerating the development of the zebrafish cardiac physiological function analysis platform based on image technology.
Imagine 100 x 100 2D images of 1024 X 512 pixels each are stacked into a 3D heart model. Its file size is at least 10 GB. For the development team, marking the boundaries of the heart chambers and individual cells on tens of thousands of images is a tedious and difficult task. According to Dr. Liao Yi-Han, in the past, it took at least ten hours for researchers to manually analyze a single object. Even after they switched to program-assisted image analysis in the later stage, it would take more than one hour. However, it was still difficult to mark the dynamic trajectory of cells. Now NVIDIA RTX A6000 is used for model training and verification under the TensorFlow framework. Using AI technology, it can accurately identify the inner boundary of the cavity and mark individual cells in complex images that are not easily recognizable. What's even more impressive is that the analysis of a single object can be completed within a few minutes, and the realistic heart chamber contraction and the dynamic trajectory of hundreds of cardiomyocytes can be outlined, resulting in a significant increase in efficiency.
(The picture above shows Professor Liao Yi-Han from the Department of Applied Chemistry of Yang Ming Chiao Tung University and Xu Yi-Jie, a first-year graduate student responsible for developing this AI application)
The zebrafish heart physiological function analysis tool developed by Professor Liao Yi-Han's team at National Yang Ming Chiao Tung University can help improve the efficiency of heart-related physiological and pathological research, and accelerate the screening of drug toxicity and efficacy. If it is developed into a primary screening platform for cardiotoxicity of preclinical drugs in the future, it is expected to reduce the time and cost of new drug development and has extremely high application value and promotion prospects. It is expected that with the support of AI, the research and development process will be greatly shortened and the vision will be realized much sooner.
The NVIDIA RTX A6000 features the latest NVIDIA Ampere GPU and 10,752 NVIDIA CUDA cores. Its 48GB GDDR6 memory and up to 768GB/s memory bandwidth are specially designed for computing-intensive tasks. Connecting two NVIDIA RTX A6000 cards with NVLink can effectively double the memory space for larger projects. With a maximum power consumption of only 300W, it's efficient for extended operation.
The entire lineup of NVIDIA RTX GPU comes with optimized drivers, over 100 professional software certifications, and IT management tools, designed to accelerate professional workflows. In addition, Leadtek offers exclusive 3-year warranty and has established a global GPU technology consulting center, providing users with the most professional and secure service and graphics environment.
Photo: Leadtek NVIDIA Quadro RTX A6000
Future
