Related Experiment Video
Updated: May 8, 2026
07:19
Microfluidic Imaging Flow Cytometry by Asymmetric-detection Time-stretch Optical Microscopy (ATOM)
Published on: June 28, 2017
Graphics Processing Unit (GPU) implementation of image processing algorithms to improve system performance of the
S N Swetadri Vasan1, Ciprian N Ionita, A H Titus
1Department of Electrical Engineering, University at Buffalo ; Toshiba Stroke Research Center, University at Buffalo.
Proceedings of Spie--The International Society for Optical Engineering
|September 13, 2013
Summary
Upgrades to the Control, Acquisition, Processing, and Image Display System (CAPIDS) for a Micro-Angiographic Fluoroscope (MAF) detector using Graphics Processing Units (GPUs) significantly improved image processing speeds. The new system achieves stable 30 frames per second operation for various imaging procedures.
Area of Science:
- Medical Imaging
- Computer Science
- Biomedical Engineering
Background:
- Current image processing in the Control, Acquisition, Processing, and Image Display System (CAPIDS) for Micro-Angiographic Fluoroscope (MAF) detectors is largely pixel-independent, enabling parallel processing.
- Graphics Processing Units (GPUs) are designed for massive parallel processing, offering significant speed advantages over Central Processing Units (CPUs) for highly parallelizable algorithms.
Purpose of the Study:
- To implement and evaluate image processing algorithm upgrades on a GPU within the CAPIDS for a custom MAF detector.
- To enhance the performance and capabilities of the MAF imaging system through parallel processing.
Main Methods:
- Image processing algorithms including flat field correction, temporal filtering, image subtraction, roadmap mask generation, and display window/leveling were upgraded for GPU implementation.
- The upgraded CAPIDS system was compared to its previous version to quantify performance improvements.
Main Results:
- The GPU implementation of image processing algorithms resulted in substantial improvements in processing speed and frame rates.
- The upgraded system achieved stable operation at 30 frames per second (fps) during fluoroscopy, digital subtraction angiography (DSA), and roadmap procedures.
- Automatic image windowing and leveling were successfully implemented on a per-frame basis.
Conclusions:
- Implementing image processing on a GPU significantly accelerates the CAPIDS for MAF detectors.
- The GPU-accelerated system enables real-time, high-frame-rate imaging crucial for advanced fluoroscopic procedures.
- These upgrades enhance the overall efficiency and diagnostic utility of the MAF system.
