Multi-dimensional Array Data Structure for CNN Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current imaging data processing techniques, such as filtered back projection and simultaneous algebraic reconstruction techniques, produce images with artifacts and lack detail, leading to undesirable image quality in applications like medicine and security.
Innovation Solution
The method involves transforming raw imaging data into a multi-dimensional array with subarrays containing local information, suitable for processing by convolutional neural networks, thereby improving image reconstruction and reducing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional imaging reconstruction techniques (filtered back projection, simultaneous algebraic reconstruction) are used, then the processing method is simple and well-established, but the image quality is poor with artifacts and lack of detail
Solution Approach 1:
The imaging data is divided into multiple subarrays within a multi-dimensional array structure, where each subarray contains local information suitable for CNN processing. This segmentation allows the complex data to be organized in a way that enables machine learning algorithms to process local patterns effectively while maintaining overall image reconstruction quality.
Solution Approach 2:
The patent transforms the traditional two-dimensional image data into a multi-dimensional array structure with additional dimensions for organizing local information. This dimensional transformation enables convolutional neural networks to process the data more effectively by providing spatial context and local relationships that were not accessible in traditional flat data structures.
2Reliability
If traditional reconstruction techniques are used, then the processing approach is straightforward, but the images contain artifacts and lack detail
Solution Approach 1:
The patent replaces traditional mechanical/mathematical reconstruction algorithms (filtered back projection, simultaneous algebraic reconstruction) with a machine learning-based convolutional neural network approach. This substitution enables the system to learn complex patterns and relationships in the imaging data that lead to more accurate and artifact-free images, despite the increased complexity of the processing system.
3Manufacturing precision
If data is reorganized into multi-dimensional arrays with subarrays for CNN processing, then image quality and detail are significantly improved, but the data structure becomes more complex
Solution Approach 1:
The patent organizes data into subarrays where each subarray contains local information with specific properties optimized for CNN processing. This local quality approach ensures that each subarray contains the necessary contextual information for accurate local reconstruction, while the overall multi-dimensional structure maintains global image coherence, resolving the contradiction between local processing needs and global image quality.
Data Source
AI summary
Constructing a computer image from raw imaging data or encoded imaging data by transforming a first data structure in which the raw imaging data or the encoded imaging data is stored into a second data structure storing reorganized imaging data. The raw imaging data or the encoded imaging data is received, stored in the first data structure. The computer reorganizes the raw imaging data or the encoded imaging data into the reorganized data and stores the reorganized data in the second data structure, which is a multi-dimensional array having subarrays containing local information needed by a convolutional neural network for processing the reorganized data. Other portions of the multi-dimensional array store other portions of the raw imaging data or the encoded imaging data. The computer also processes the reorganized data using the convolutional neural network to construct the image, whereby a constructed image is formed.