A Low Power and Real-Time Architecture for Hough Transform Processing Integration in a Full HD-Wireless Capsule...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computer Vision

Background:

• Wireless endoscopic capsules (WCEs) are crucial for non-invasive gastrointestinal diagnostics.
• Current WCEs transmit all captured images, leading to high data loads and potential inefficiencies.
• Polyp detection is a key challenge in endoscopic imaging.

Purpose of the Study:

• To develop a smart WCE capable of on-board polyp detection and image selection.
• To design a real-time image processing algorithm for polyp identification within WCE constraints.
• To validate the algorithm's effectiveness and hardware implementation for practical use.

Main Methods:

• An image processing system was designed using the Hough Transform (HT) for shape-based polyp detection.
• A novel algorithm was developed for real-time HT computation on high-definition images (1920x1080).
• The algorithm was implemented and validated on a Xilinx Spartan 7 FPGA, considering area and power constraints.

Main Results:

• The algorithm successfully detected circular shapes, including irregular contours characteristic of polyps, in synthetic and real endoscopic images.
• The FPGA implementation operated at 132 MHz with 76 mW power consumption, enabling ~10 hours of operation.
• An ASIC estimation projected operation at 125 MHz with 17.2 mW power consumption, extending duration to ~50 hours.

Conclusions:

• The proposed smart WCE paradigm with real-time Hough Transform-based polyp detection is feasible.
• The developed algorithm and hardware architecture meet the stringent constraints of WCE integration.
• This technology offers a significant advancement in efficient and intelligent gastrointestinal image analysis.