Machine learning algorithms for FPGA Implementation in biomedical engineering applications: A review.

Area of Science:

• Biomedical Engineering
• Computer Engineering
• Machine Learning

Background:

• Field Programmable Gate Arrays (FPGAs) offer reconfigurable hardware solutions.
• Machine Learning (ML) algorithms are increasingly vital in healthcare technology.
• Traditional processors lack the customization capabilities of FPGAs for specialized tasks.

Purpose of the Study:

• To review ML algorithms implemented on FPGAs for healthcare applications from 2001-2023.
• To focus on real-time ML algorithms and hybrid System-on-a-chip (SoC) FPGA architectures for biomedical uses.
• To synthesize research on ML classifiers and regression algorithms optimized for FPGA deployment.

Main Methods:

• Comprehensive literature review of ML algorithms on FPGAs in biomedical applications.
• Analysis of optimization strategies for ML algorithms and FPGA designs.
• Focus on real-time processing and hybrid SoC FPGA architectures.
• Synthesis of studies implementing classifier and regression algorithms.

Main Results:

• FPGAs enable high-performance ML in healthcare, overcoming resource limitations.
• Optimized ML algorithms and FPGA designs address challenges like memory and power constraints.
• Hybrid SoC FPGA architectures show promise for complex biomedical tasks.
• Classifiers and regression algorithms are widely adopted for diverse biomedical applications.

Conclusions:

• FPGA-based ML accelerators significantly enhance biomedical applications.
• Customization and optimization are key to deploying ML effectively on FPGAs in healthcare.
• This review informs researchers on advancing FPGA-enabled ML for future biomedical innovations.