SPIN MRI Circuit Optimizes Pulse Sequence and Reconstruction Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Fourier-space-based MRI techniques face challenges such as weak detected signals, T2 defocusing, and assumptions of stationary protons, leading to errors in image reconstruction, especially when data acquisition time is shortened in applications like cardiovascular or brain studies.
Innovation Solution
A synergized pulsing-imaging network (SPIN) circuitry optimizes the pulse sequence and reconstruction network using a loss function associated with a reconstruction network, iteratively adjusting parameters like flip angle, magnetic field gradients, and proton density, based on intermediate raw MRI data and ground truth images, without relying on k-space data or Fourier transforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Fourier-space-based pulse sequences are used for MRI data acquisition, then the imaging process follows established theoretical frameworks, but the detected signal becomes weak and reconstruction errors increase
Solution Approach 1:
The patent changes the fundamental parameters of MRI data acquisition by abandoning Fourier-space sampling in favor of direct spatial domain sampling. This parameter change allows the system to capture signals that would otherwise be lost in Fourier transformation, thereby improving both signal detection strength and reconstruction accuracy simultaneously.
Solution Approach 2:
The patent substitutes the traditional Fourier transform mathematical mechanism with a direct neural network-based reconstruction mechanism. This substitution eliminates the need for Fourier-space sampling and enables the system to work directly with spatial domain data, resolving the contradiction between signal strength and reconstruction accuracy.
2Productivity
If data acquisition time is shortened for cardiovascular or brain studies, then the imaging speed increases, but reconstruction errors and biases increase
Solution Approach 1:
The patent performs preliminary action by training the neural network reconstruction model in advance with comprehensive data. This pre-training enables the model to compensate for incomplete or accelerated data acquisition, allowing short scanning times without sacrificing reconstruction accuracy. The model learns to reconstruct high-quality images even from limited data acquired in reduced time.
Solution Approach 2:
The patent introduces a neural network as an intermediary between raw MRI data and final image reconstruction. This intermediary model acts as a bridge that can process accelerated or incomplete data and produce accurate reconstructions, thereby decoupling the relationship between acquisition speed and reconstruction quality.
3Device complexity
If T2 defocusing effects are ignored in the Fourier formulation, then the processing complexity is reduced, but image reconstruction errors increase
Solution Approach 1:
The patent substitutes the Fourier transform mathematical model with a neural network-based model that inherently accounts for T2 defocusing effects. This substitution eliminates the need to explicitly model and correct T2 effects separately, as the neural network learns these effects during training and automatically compensates for them during reconstruction, maintaining both simplicity and accuracy.
4Device complexity
If spinning protons are modeled as stationary or low order moments, then the computational complexity is reduced, but reconstruction errors increase for moving protons
Solution Approach 1:
The patent introduces a neural network as an intermediary that learns the complex relationship between MRI signals and proton positions without requiring explicit motion models. The network automatically adapts to handle moving protons by learning from training data that includes various motion patterns, eliminating the need for simplified stationary or low-order moment models while maintaining high positioning accuracy.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach maximizes the information content of raw MRI data, resulting in optimal image reconstruction and improved data acquisition, reducing errors and artifacts, as demonstrated by increased SSIM and PSNR values over iterations.
Implementation Method 1
These in-plane vectors generate alternating electromagnetic fields in nearby coils to produce so-called free induction decay (FID) signals
Implementation Method 2
during data acquisition, the signal may experience a T2 defocusing
Implementation Method 3
optimizing, by a synergized pulsing-imaging network (SPIN) circuitry, a pulse sequence based, at least in part, on a loss function associated with a reconstruction network
Data Source
AI summary
A synergized pulsing-imaging network is described. A method of optimizing a magnetic resonance imaging (MRI) system includes optimizing, by a synergized pulsing-imaging network (SPIN) circuitry a pulse sequence based, at least in part, on a loss function associated with a reconstruction network. The method further includes optimizing, by the SPIN circuitry, the reconstruction network based, at least in part, on intermediate raw MRI data and based, at least in part, on a ground truth MRI image data. The intermediate raw MRI data is determined based, at least in part on the pulse sequence.