MRI Phase Reconstruction via Distance Matrix Curve Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating magnetic resonance imaging (MRI) of cyclic movements, such as chewing, face challenges in achieving high temporal and spatial resolution while maintaining natural movement speed and complexity, often requiring complex fixation and additional devices, and are limited by artifacts and ambiguous correlation maxima.
Innovation Solution
A method involving the generation of raw data sets using partial k-space trajectories, reconstruction of intermediate images with high temporal resolution, calculation of a distance matrix, and fitting curves to identify intersections for precise image reconstruction of movement phases, allowing for MRI of cyclic movements without demanding movement execution or speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fixation devices are used to capture individual movement positions, then magnetic resonance tomographic data can be generated for each position, but the device complexity and recording complexity increase significantly
Solution Approach 1:
The patent extracts and removes the complex fixation devices from the system. Instead of using external devices to hold the jaw in fixed positions, the method captures the jaw's natural cyclic movement continuously and reconstructs phase-specific images from the moving sequence, eliminating the need for fixation apparatus while maintaining measurement precision
Solution Approach 2:
The patent creates a temporal copy of the movement information through continuous recording across multiple cycles. By logging jaw positions over several complete movement cycles and using this replicated temporal data, the system reconstructs accurate phase-specific images without requiring physical fixation devices
2Measurement precision
If movement is performed slowly to capture sufficient data during a single cycle, then magnetic resonance tomographic data can be generated, but the method becomes demanding for patients and dynamic effects at normal speeds cannot be mapped
Solution Approach 1:
The patent utilizes the periodic nature of cyclic movements by recording over multiple complete cycles. Instead of requiring a single slow movement, the system captures several natural-speed cycles and assigns data records to corresponding phases, allowing patients to perform movements at normal speed while still achieving sufficient data quality through the periodic repetition
Solution Approach 2:
The patent maintains continuous recording throughout multiple movement cycles without interruption. This continuous action captures all dynamic effects at natural movement speeds while accumulating sufficient data across cycles, eliminating the need to slow down the movement and making the procedure easier for patients
3Reliability
If correlation coefficients are used to identify movement phases, then images can be correlated, but the spatial and temporal resolution are limited due to dense k-space sampling requirements
Solution Approach 1:
The patent applies partial action by using only the necessary portion of k-space data for each phase. Instead of requiring dense sampling of the entire k-space for every image, the method selectively uses data from multiple cycles and applies partial Fourier reconstruction, achieving adequate image quality with reduced sampling requirements and improved resolution
Solution Approach 2:
The patent changes the parameter of data sampling density by transitioning from dense uniform sampling to selective sparse sampling across multiple cycles. By varying the sampling strategy and using phase-assignment algorithms, the system maintains reliability while achieving higher spatial and temporal resolution than traditional correlation methods
4Manufacturing precision
If dense sampling of k-space is performed to obtain artifact-free images, then image quality improves, but the temporal resolution decreases to 50-300 ms
Solution Approach 1:
The patent segments the k-space data into multiple cycles and further segments each cycle into phase-specific data records. By dividing the total data acquisition into discrete phase segments across multiple cycles, the system can reconstruct high-resolution images for each phase without requiring dense sampling of the entire k-space at once, thereby improving temporal resolution while maintaining image quality
Solution Approach 2:
The patent uses partial k-space sampling for each phase reconstruction rather than complete dense sampling. By acquiring data from multiple cycles and using only the relevant portion for each phase, the method reduces the sampling burden per image while maintaining adequate image quality, achieving faster temporal resolution
Data Source
Figure 1~2
AI summary
Method for producing magnetic resonance tomography images (B) of at least one phase of a cyclic movement, comprising the method steps: production of raw data sets (rl... rx) of the cyclic movement during a recording period (T) having radial or almost radial k-space part trajectories (kl... kx); reconstructing of a series of intermediate images (zl... zy), each from at least one raw data set (rl... rx) with high time resolution at least for each region (region of interest, ROI) of the raw data sets (rl... rx); calculation of a distance matrix (D) from the series of intermediate images (zl... zy), wherein each matrix element (D) corresponds to the distance of a first intermediate image (zl... zy) of the series to the first or a further intermediate image (zl... zy) of the series; fitting of functions (vi... vz) to structures forming in the distance matrix (D) by means of an active contour method and reconstruction of at least one image (B) from the raw data sets (rl... rx), said raw data sets corresponding to intersection (S) of the fitted curves (vl... vz) with a line of the distance matrix (D).