MRI Reconstruction Using Local Projection Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing MRI reconstruction methods require accurate estimation of coil sensitivity functions, which is challenging and can lead to image artifacts due to errors in sensitivity estimation, especially when using arbitrary k-space trajectories.
Innovation Solution
The local projection calibration technique removes aliasing artifacts without estimating coil sensitivity functions by synthesizing k-space data using a combination of gradient encoding and receiver coil sensitivities, allowing for reconstruction of images from arbitrary k-space trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If coil sensitivity functions are estimated to enable parallel imaging reconstruction, then image reconstruction can be performed from multiple receiver coils, but errors in sensitivity estimation lead to visible image artifacts
Solution Approach 1:
The system uses the acquired k-space data itself to determine the coil combination weights through autocorrelation analysis, rather than requiring separate calibration scans or external sensitivity measurements. The method extracts sensitivity information directly from the parallel imaging data, making the system self-calibrating and eliminating the source of error introduced by separate sensitivity estimation procedures
Solution Approach 2:
The patent introduces autocorrelation of k-space data as an intermediary mathematical operation that bridges the gap between raw multi-coil data and accurate coil combination weights. This intermediary step transforms the data to reveal the underlying sensitivity profiles without requiring direct measurement, thereby avoiding the errors associated with traditional sensitivity estimation methods
2Measurement precision
If full field-of-view calibration scan is performed to obtain accurate coil sensitivity functions, then reconstruction accuracy is improved, but acquisition time increases
Solution Approach 1:
The method extracts the necessary coil sensitivity information directly from the acquired imaging data by performing autocorrelation analysis on the k-space data. This extraction approach eliminates the need for separate full FOV calibration scans, as the sensitivity profiles are embedded within the parallel imaging data itself and can be recovered through mathematical operations on the existing data
Solution Approach 2:
The patent performs the sensitivity determination action preliminarily and automatically during the data acquisition process itself, rather than requiring a separate subsequent calibration step. The autocorrelation analysis is applied to the acquired data to pre-determine the optimal combination weights before final image reconstruction, thereby eliminating additional calibration time
3Loss of time
If reduced gradient encoding is used with multiple receiver coils, then acquisition time is reduced, but aliasing artifacts appear in the reconstructed image
Solution Approach 1:
The system uses feedback from the acquired multi-coil k-space data itself to determine the optimal coil combination weights through autocorrelation analysis. This feedback mechanism allows the reconstruction process to adapt to the specific encoding patterns and sensitivity profiles present in the data, enabling effective removal of aliasing artifacts while maintaining the benefits of reduced gradient encoding
Solution Approach 2:
The patent changes the parameter being optimized from fixed coil combination weights to data-dependent weights determined through autocorrelation analysis. By allowing the combination parameters to be derived from the actual acquired data rather than predetermined, the system can effectively suppress aliasing artifacts that result from accelerated imaging
Data Source
AI summary
Disclosed is a method of providing magnetic resonance image reconstruction from k-space data obtained from any trajectory of k-space using multiple receiver coils. An image is constructed for data from each coil, and then the multiple coil images are combined such as by using sum of squares of image data, for example.