Prior Image Constrained Compressed Sensing for Cardiac CT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tomographic imaging systems face challenges in achieving high temporal resolution due to motion artifacts caused by cardiac or respiratory movements, leading to blurring and image artifacts, especially in applications like cardiac imaging, where existing methods either require costly upgrades or compromise on system reliability.
Innovation Solution
The implementation of the Prior Image Constrained Compressed Sensing (PICCS) method, which uses a prior image and a temporally reduced subset of short-scan projection data to reconstruct images, effectively mitigating limited-view-angle shading artifacts and improving temporal resolution without hardware modifications, by minimizing an objective function that includes terms for image sparsity and motion streaks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional short scan reconstruction is used, then image reconstruction is achievable with standard hardware, but temporal resolution is insufficient leading to motion artifacts
Solution Approach 1:
A prior image is reconstructed from a previous cardiac phase and used as a constraint in the current reconstruction. This preliminary image provides structural information that guides the reconstruction algorithm to achieve higher temporal resolution without requiring additional hardware or longer scan times.
Solution Approach 2:
The reconstruction algorithm incorporates an objective function that includes sparsity constraints and motion streak suppression terms. By changing the mathematical parameters of the reconstruction process rather than the physical acquisition parameters, the system achieves improved temporal resolution from the same short scan data.
2Loss of time
If dual-source CT system is implemented, then temporal resolution is improved, but system cost and complexity increase significantly
Solution Approach 1:
Instead of using a second physical x-ray source and detector array, the system creates a virtual second dataset by reconstructing a prior image from previous cardiac phases. This copied structural information is then used to enhance the current image reconstruction, achieving dual-source-like temporal resolution with single-source hardware.
Solution Approach 2:
The prior image acts as an intermediary between the limited short scan data and the final high-temporal-resolution image. It provides the missing structural information that allows the reconstruction algorithm to resolve motion artifacts and achieve improved temporal resolution without additional hardware.
3Loss of time
If faster gantry rotation is used, then temporal resolution improves, but mechanical stress and reliability concerns increase
Solution Approach 1:
Rather than changing the physical acquisition parameters like gantry speed, the system changes the reconstruction parameters by incorporating prior image constraints and sparsity regularization. This allows the same mechanical system to produce images with effectively higher temporal resolution through computational methods.
Solution Approach 2:
The system replaces mechanical solutions (faster gantry rotation, dual-source hardware) with computational methods. The reconstruction algorithm uses mathematical optimization with prior image constraints to achieve the temporal resolution that would otherwise require mechanical modifications or more complex hardware configurations.
Data Source
AI summary
A computerized tomographic system configured to acquire short scan data of an object within a single revolution of a detector array about the object over a first angular range of rotation of the detector array about the object, define a temporal subset of the acquired short scan data over a second angular range of rotation of the detector that is less than the first angular range of rotation, generate a mathematical function that is based on the acquired short scan data and the defined temporal subset of data, minimize the mathematical function, and generate an image of the object using the minimized mathematical function and the data acquired over the second angular range of rotation of the detector.