GPU X-ray Imaging Data Aggregation for High Frame Rates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current x-ray imaging systems face challenges in processing large volumes of data at high frame rates, especially in real-time applications like cardiac imaging, leading to performance losses and increased costs due to the need for additional GPUs, which can compromise image quality and frame rates.
Innovation Solution
The system employs a scanning beam x-ray source with multiple focal spots, a multi-detector array, and a graphics processing unit (GPU) architecture that uses direct memory access and PCI-E buses for efficient data transfer and image reconstruction, allowing for high-speed processing and aggregation of detector images to achieve high frame rates and improved image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple GPUs are used to process imaging data in parallel, then processing speed and frame rate are improved, but system cost and complexity increase
Solution Approach 1:
The patent segments the imaging data processing workload across multiple GPUs, with each GPU handling a portion of the parallel processing tasks. This segmentation allows the system to achieve high processing speeds by distributing the computational burden, while each individual GPU operates within manageable complexity parameters. The data is divided into multiple streams that can be processed simultaneously by different GPU units.
Solution Approach 2:
The patent introduces a temporal dimension to the processing architecture by implementing a buffer system that stores complete frame data sets. This buffer allows data to be accumulated and then transferred to GPUs in optimized batches, adding a time-based layer to the spatial parallelism already provided by multiple GPUs. This dimensional addition helps manage data flow and reduces overhead complexity.
2Speed
If data is streamed continuously to multiple GPUs, then real-time processing is achieved, but severe performance losses occur
Solution Approach 1:
The patent implements periodic action by buffering complete frame data sets before transferring them to GPUs for processing. Instead of continuous streaming, the system accumulates data in periodic batches, allowing GPUs to process complete frames at optimized intervals. This periodic approach maintains real-time capabilities while avoiding the performance degradation associated with continuous data streaming.
Solution Approach 2:
The patent applies preliminary action by pre-buffering complete frame data sets in memory before GPU processing begins. This preliminary buffering allows the system to prepare data in an optimized format and location, reducing the overhead and latency that would otherwise occur during real-time streaming operations. The data is readied in advance, enabling GPUs to operate at peak performance without continuous data arrival interruptions.
3Productivity
If additional GPUs are used to compensate for performance losses, then processing capacity is maintained, but cost and system complexity increase
Solution Approach 1:
The patent ensures continuity of useful action by implementing a buffer system that maintains a ready supply of complete frame data sets for GPU processing. This continuous availability of prepared data allows GPUs to operate continuously at optimal performance levels without idle time or waiting for data arrivals, maximizing the useful action of each GPU without requiring additional units to compensate for performance losses.
4Manufacturing precision
If high frame rates are required for smooth motion imaging, then image quality is improved, but data processing volume and complexity increase
Solution Approach 1:
The patent segments the high-volume data processing required for high frame rate imaging across multiple GPUs, with each GPU handling a specific portion of the frame data. This segmentation allows the system to maintain high image quality by processing sufficient data at high frame rates, while distributing the computational complexity across multiple parallel units rather than overwhelming a single processor.
Solution Approach 2:
The patent adds a temporal buffering dimension to manage the complexity of high frame rate processing. By buffering complete frame data sets and transferring them in optimized batches, the system handles the large data volumes required for high image quality without creating proportional complexity in the processing pipeline. The temporal layer allows data to be organized and transferred efficiently, reducing overhead.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the production of high-quality images at high frame rates, reducing x-ray exposure and radiation dose while maintaining image quality, and optimizing GPU performance by aggregating data in short bursts, thus overcoming the limitations of single GPU processing.
Implementation Method 1
A plurality of x-ray illumination source positions are utilized to produce x-ray radiation at each of the x-ray illumination source positions and to project x-ray radiation towards an object
Implementation Method 2
A detector detects x-ray radiation from the object and transmits detector images for each of the illumination source positions
Data Source
AI summary
An x-ray imaging system utilizes enhanced computing arrays. A plurality of x-ray illumination source positions are utilized to produce x-ray radiation at each of the x-ray illumination source positions and to project x-ray radiation towards an object. A detector detects x-ray radiation from the object and transmits detector images for each of the illumination source positions. A memory buffer stores the detector images from the detector. A graphics processing unit formats the detector images and constructs a complete frame data set with the detector images for each of the illumination source positions. Another graphics processing unit receives the complete frame data set and performs image reconstruction on the complete frame data set.