A design of an associative memory array processor for ultrasonograph image acquisition and processing

Area of Science:

• Computer Engineering
• Medical Imaging Technology
• Artificial Intelligence

Background:

• Ultrasonography is crucial in medical diagnostics.
• Efficient processing of ultrasonograph images is essential for timely diagnosis.
• Current image processing methods can be time-consuming, delaying clinical decisions.

Purpose of the Study:

• To design and describe a novel associative memory array processor.
• To enhance the speed and efficiency of ultrasonograph image acquisition and processing.
• To support radiology specialists by providing faster and more reliable image analysis tools.

Main Methods:

• Development of a parallel architecture based on single-instruction stream, multiple-data stream (SIMD) principles.
• Utilization of content-addressable associative memory slabs for parallel data access.
• Implementation of one-pixel processing elements with comparison logic for image analysis.
• Parallel execution of image processing operations including component labeling, size filtering, pattern centralization, and pattern recognition.

Main Results:

• The proposed architecture enables concurrent analysis of multiple image parts, significantly reducing task execution time.
• The system effectively performs learning, feature extraction, and stores reference patterns.
• Parallel image processing operations like intelligent region labeling and isolation are achieved.
• The system accurately compares input images with pre-stored templates for pattern recognition.

Conclusions:

• The novel associative memory array processor offers a significant speed-up in ultrasonograph image recognition.
• This architecture facilitates intelligent isolation and analysis of specific image regions.
• The system's ability to learn and compare patterns aids in efficient diagnostic processes.
• The technology has the potential to increase confidence and efficiency for radiology specialists.