by the School of Engineering and Applied Science at Columbia University
The application of BeatProfiler in motion to cardiovascular cells, viewed from a computer perspective, offers a unique insight into heart function analysis.
Traditionally, understanding brain function, assessing heart conditions, and testing new cardiac drugs have been arduous tasks requiring significant time and effort. While utilizing mobile and engineered cell models for disease examination and drug testing shows promise, the current methodologies for studying heart cell contraction and calcium handling are labor-intensive, error-prone, and reliant on costly specialized equipment.
There is a pressing need in the medical field for a more efficient, precise, and accessible approach to studying heart function, leveraging artificial intelligence (AI) and machine learning techniques.
Introducing BeatProfiler: A Cutting-Edge Tool for Heart Cell Analysis
Addressing these challenges directly, experts at Columbia Engineering have developed BeatProfiler. This innovative tool integrates the analysis of key heart performance metrics such as contraction, calcium handling, and force output into a single platform. By automating the analysis of heart cell function from video data, BeatProfiler streamlines the process, reduces errors, and enhances efficiency.
BeatProfiler empowers researchers to swiftly and accurately evaluate the impact of medications on heart function, as well as differentiate between various diseases and their severity levels.
The groundbreaking study introducing BeatProfiler has been published in the IEEE Open Journal of Engineering in Medicine.
“This tool represents a significant advancement,” remarked project lead Gordana Vunjak-Novakovic, University Professor and the Mikati Foundation Professor of Biomedical Engineering, Medical Sciences, and Dental Medicine at Columbia. “Its speed, comprehensiveness, automation, and compatibility with various computer systems make it easily accessible to experts and practitioners alike.”
Open-Source Availability of the Software
The team, which includes Barry Fine, associate professor of medicine (in Cardiology) at Columbia University Irving Medical Center, has opted not to pursue a patent for the AI software. Instead, they have chosen to release it as open-source, enabling free access and utilization by researchers worldwide.
This decision reflects their commitment to disseminating research outcomes and soliciting input from academic, clinical, and industrial laboratories to further enhance the software’s capabilities.
Addressing the Urgency of Accurate Heart Disease Diagnosis
Driven by the imperative to diagnose heart conditions swiftly and accurately, this project aligns with Vunjak-Novakovic’s broader research goals. Over several years, the team iteratively enhanced the tool to accelerate the assessment of cardiac models in real-time. Their primary objective was to enhance the capability of cardiac models to study heart diseases and assess potential therapeutics efficiently, particularly in the context of evolving innovations such as milliPillar and multiorgan tissue models.
Lead author Youngbin Kim and co-authors developed a user-friendly graphical interface (GUI) atop the code in the final phase, enabling biomedical researchers without programming expertise to leverage the tool effectively.
This collaborative effort brought together experts in software development, machine learning, signal processing, engineering, and user experience to create a versatile and intuitive tool for cardiac analysis.
Achievements and Future Directions
The study demonstrated that BeatProfiler excelled in analyzing cardiomyocyte function, surpassing existing tools in speed—up to 50 times faster in certain scenarios—and reliability. It detected subtle variations in engineered heart tissue force responses that might elude other analytical methods.
Kim, a Ph.D. candidate in Vunjak-Novakovic’s lab at Columbia Engineering, emphasized, “The unprecedented speed and versatility in cardiac research offered by BeatProfiler’s functional measurements, analyzed through machine learning, enabled us to differentiate between healthy and diseased heart cells with high accuracy and classify various cardiac drugs based on their effects.”
Looking ahead, the team aims to expand BeatProfiler’s capabilities for diverse applications in heart research, encompassing a spectrum of heart diseases affecting cardiac function and drug development. By validating its performance against different in-vitro cardiac models, they seek to ensure its versatility in addressing a wide array of research inquiries.
Furthermore, ongoing efforts focus on enhancing the machine learning algorithm to accommodate various heart conditions and drug classifications, with the ultimate goal of simplifying drug testing processes in pharmaceutical settings.
More information: Youngbin Kim et al, BeatProfiler: Multimodal In Vitro Analysis of Cardiac Function Enables Machine Learning Classification of Diseases and Drugs, IEEE Open Journal of Engineering in Medicine and Biology (2024). DOI: 10.1109/OJEMB.2024.3377461
Provided by the School of Engineering and Applied Science at Columbia University
Citation: Biomedical engineers use AI to build new tool for studying and diagnosing heart function (2024, April 8) retrieved 8 April 2024 from https://medicalxpress.com/news/2024-04-biomedical-ai-tool-heart-function.html
This document is subject to copyright. No part may be reproduced without the prior written consent, aside from any fair dealing made for the purpose of private study or research. The information is only provided for information purposes.
