Parallelizable Processing Unit for Real-Time QUS Tissue Characterization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional CPU-based approaches to computing QUS parameters from ultrasound radiofrequency data are inefficient, leading to serial processing that limits real-time data processing capabilities and restricts ultrasound tissue characterization to post-processing paradigms due to the lack of parallelization in computational architectures.
Innovation Solution
The use of parallelizable processing units, specifically graphical processing units (GPUs) with SIMD architectures, to concurrently compute QUS values across multiple locations, enabling real-time or near real-time tissue characterization by processing ultrasound data in parallel across axial scan lines and windows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If traditional CPU-based serial processing is used to compute QUS parameters, then computational accuracy is maintained, but processing speed is limited and real-time tissue characterization cannot be achieved
Solution Approach 1:
The patent segments the computation of QUS parameters into independent parallel tasks that can be executed simultaneously across multiple processing units. Each processing unit computes spectral parameters for a specific region or frequency range independently, allowing the overall computation to be divided and executed in parallel, thereby dramatically increasing processing speed while maintaining accuracy
Solution Approach 2:
The patent replaces traditional CPU-based sequential processing with GPU-based parallel processing architecture. This substitution of computational mechanics enables simultaneous execution of multiple QUS parameter calculations across thousands of cores, transforming the processing paradigm from serial to parallel and achieving real-time tissue characterization capabilities
2Productivity
If parallel processing is implemented to accelerate QUS computation, then real-time processing capability is achieved, but computational resource requirements increase
Solution Approach 1:
The patent implements partial parallelization by selecting specific computational stages of QUS parameter calculation that are most suitable for parallel execution. Rather than parallelizing the entire processing pipeline, the invention identifies and parallelizes only the computationally intensive spectral analysis portions, achieving high throughput while minimizing unnecessary resource consumption
Solution Approach 2:
The patent designs a parallel processing architecture where processing units can handle multiple QUS parameter calculations simultaneously. Each processing unit is configured to perform various spectral analysis functions (power spectrum calculation, spectral moment computation, frequency estimation) on different data segments, maximizing resource utilization and reducing overall computational resource requirements
3Loss of time
If sequential computation of aggregated values is performed for each location, then memory usage is optimized, but processing time increases significantly
Solution Approach 1:
The patent transitions from sequential one-dimensional processing to parallel multi-dimensional processing by organizing computational tasks across spatial dimensions (multiple processing units working simultaneously on different locations). This dimensional transformation allows the system to process computational data volume in parallel, reducing processing time from sequential O(n) to parallel O(1) complexity for each location
Data Source
AI summary
A quantitative ultrasound (QUS) system for characterizing a specimen, the system comprising an ultrasound transducer operable to transmit ultrasound signals into the specimen along multiple adjacent scan lines extending axially within the specimen, and collect returned ultrasound signals therefrom and generate RF signals based on said returned ultrasound signals, wherein said RF signals are associated with respective ones of said scan lines to represent a characteristic of the specimen at each of multiple locations within the specimen along each of said scan lines; and a parallelizable processing unit communicatively coupled to said ultrasound transducer and operable to concurrently compute from said RF signals respective QUS values representative of said characteristic for each of a plurality of said multiple locations in parallel, wherein successive parallel outputs of said respective QUS values are characteristic of the specimen along each of said multiple scan lines.