Asynchronous Medical Image Processing via Real-Time Controller and GPU
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems fail to simultaneously perform real-time image display and processing on a commercially available operating system like Windows, which is not inherently real-time, leading to potential delays and glitches during medical procedures that can harm patients.
Innovation Solution
An asynchronous medical image processing system with a real-time controller connected to a GPU and CPU, allowing direct asynchronous communication and operation, enabling real-time image data processing and display without CPU intervention, using DMA for direct memory access and asynchronous communication to prevent delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If images are directly injected into graphics card memory using DMA without OS involvement, then image update speed is improved, but the system cannot perform both display and real-time processing simultaneously
Solution Approach 1:
The system divides the GPU functionality into separate processing streams: one dedicated to display output and another to real-time processing. This segmentation allows independent operation of display and processing functions, resolving the contradiction between fast image updates and simultaneous multi-function capability
Solution Approach 2:
The patent introduces an intermediary mechanism (asynchronous command queues and synchronization primitives) that mediates between the DMA image injection path and the CPU-based control path, enabling both high-speed display updates and real-time processing to coexist without direct OS intervention for each frame
2Ease of operation
If the system uses a commercially available operating system like Windows, then ease of operation is improved, but delays and glitches occur during real-time processing
Solution Approach 1:
The patent extracts the real-time processing path from the general-purpose OS scheduling domain by implementing a dedicated asynchronous command stream that bypasses OS intervention for critical real-time operations, while still allowing the OS to manage user interface and non-critical functions
Solution Approach 2:
The system performs preliminary setup of asynchronous command queues and synchronization mechanisms during initialization, establishing a dedicated real-time processing channel that operates independently of OS scheduling decisions, thereby ensuring reliable real-time performance while maintaining OS compatibility
3Measurement precision
If the CPU is involved in every image processing operation, then control precision is improved, but processing speed and continuity are reduced
Solution Approach 1:
The system segments control functions into two levels: high-level control remains with the CPU for precision, while low-level real-time execution is delegated to the GPU's asynchronous command stream, allowing CPU precision control without bottlenecking the processing speed
Solution Approach 2:
Asynchronous command queues serve as intermediaries between CPU control commands and GPU execution, allowing the CPU to issue precise control instructions at its own pace while the GPU processes them continuously without waiting for CPU intervention, thus maintaining both control precision and processing speed
Data Source
AI summary
An asynchronous medical image processing system is described that includes a real-time controller connectable to a medical imaging device, a graphics processing unit (GPU) connectable to a display device, and a central processing unit (CPU) that executes an operating system and related application(s). The real-time controller is directly connected to a memory of the GPU and performs respective operations asynchronously with respect to the CPU. The real-time controller additionally obtains medical imaging data, generates instructions for the medical imaging-data and transmits the medical imaging data and the instructions to the memory of the GPU. The GPU additionally receives and processes the medical imaging data based on the instructions from the real-time controller and instructions sent independently from the CPU.