Deep Learning Accelerator RAM for Real-Time Medical Image Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently processing medical images in real-time with reduced energy consumption and computation time using Artificial Neural Networks (ANNs), particularly in medical imaging devices like ultrasound probes and CT scanners.
Innovation Solution
An integrated circuit with a Deep Learning Accelerator (DLA) and random access memory is used to perform real-time medical image processing, where the DLA performs parallel vector and matrix calculations optimized for ANN operations, and the random access memory stores trained ANN parameters and instructions, reducing reliance on a Central Processing Unit (CPU).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a Central Processing Unit (CPU) is used for medical image processing, then the system has high flexibility and programmability, but the computation time and energy consumption increase significantly
Solution Approach 1:
The system segments the processing tasks by separating the Deep Learning Accelerator (specialized for ANN computations) from the CPU (general-purpose control). The DLA handles computationally intensive matrix and vector operations while the CPU manages data flow and control logic, resolving the contradiction by assigning specific functions to specialized components.
Solution Approach 2:
The patent introduces an intermediary memory structure that bridges the CPU and DLA, enabling efficient data transfer and coordination. This intermediary layer allows the CPU to offload computations to the DLA without direct intervention, reducing CPU workload and computation time while maintaining system flexibility.
2Use of energy by moving object
If traditional image processing methods are used, then the system has simpler hardware requirements, but the energy consumption and computation time increase
Solution Approach 1:
The patent replaces traditional general-purpose CPU-based processing with a specialized Deep Learning Accelerator that uses parallel processing architectures optimized for neural network computations. This substitution dramatically reduces energy consumption per computation while increasing processing speed through parallel matrix and vector operations.
Solution Approach 2:
The system changes the computational parameters by using fixed-point arithmetic instead of floating-point operations, and by optimizing memory access patterns to match the parallel processing capabilities of the DLA. These parameter changes enable efficient real-time processing with reduced energy consumption.
3Productivity
If real-time processing is implemented, then the diagnostic efficiency improves, but the computation time and processing complexity increase
Solution Approach 1:
The system performs preliminary actions by pre-training the neural network model offline and storing the trained weights and biases in the DLA's internal memory. During real-time operation, the DLA directly applies these pre-computed parameters to incoming images, eliminating the need for complex real-time training computations and simplifying the real-time processing architecture.
Solution Approach 2:
The DLA is designed to autonomously execute neural network computations using its internal memory and dedicated processing units. The accelerator independently handles matrix multiplications, convolutions, and activation functions without requiring complex coordination with external processors, enabling real-time processing while reducing overall system complexity.
Data Source
AI summary
Systems, devices, and methods related to a Deep Learning Accelerator and memory are described. For example, an integrated circuit device may be configured to execute instructions with matrix operands and configured with random access memory. The random access memory is configured to store an image generated in an imaging apparatus configured to image a portion of a person, parameters of an artificial neural network, and instructions executable by the Deep Learning Accelerator to perform matrix computation to generate an output of the artificial neural network. The output can include a feature identified by the artificial neural network and a diagnosis determined by the artificial neural network to assist or guide the imaging of the portion of the person.