Multiple-Acquisition MRI Image Denoising With Informative PCA Selection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for denoising magnetic resonance imaging data using principal component analysis often lose relevant anatomical and pathological information by blindly selecting principal components based on variance measures, which are ineffective for noisy data.

Innovation Solution

A method that selects principal components based on informative indicators such as variance, standard deviation, or median of scores after filtering, ensuring relevant spatial information is preserved and noise is effectively attenuated.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If principal components are selected based on variance measures, then noise attenuation is achieved, but relevant anatomical and pathological information is lost

Engineering Contradiction:
ImprovenoiseVSAvoidanatomical and pathological information
Core Design Contradiction:
Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The patent changes the selection parameter from variance-based (traditional PCA) to a hybrid criterion that incorporates both variance and a measure of anatomical/pathological information content. This allows the selection process to consider not just how much variation each principal component explains, but also how much relevant medical information it preserves, thereby resolving the contradiction between noise attenuation and information retention

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces a feedback mechanism where the selection of principal components is guided by evaluating their contribution to both noise reduction and information preservation. The selection criterion uses variance ratios as feedback to determine the optimal number of components to retain, ensuring that components contributing to noise are excluded while those containing relevant information are preserved

Inventive Principle:
Principle #23Feedback

2Measurement precision

If multiple acquisitions are performed to improve image quality, then signal-to-noise ratio increases, but acquisition time increases

Engineering Contradiction:
Improvesignal-to-noise ratioVSAvoidacquisition time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent introduces PCA as an intermediary processing step that takes multiple noisy acquisitions as input and produces a denoised output. Instead of requiring numerous acquisitions to achieve good signal-to-noise ratio, the system uses a relatively small number of acquisitions combined with PCA-based denoising, where PCA acts as a mediator that separates signal from noise in the acquired data

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the mechanical approach of acquiring many images to improve quality with a computational approach. Instead of physically acquiring more data through repeated scans, the system uses mathematical transformations (PCA) to enhance image quality from fewer acquisitions, substituting computational processing for physical data collection

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4314854B1Method for attenuating noise in images resulting from multiple MRI acquisitions
Publication Date: 2025.07.02 OLEA MEDICAL
  • EP4314854B1 patent drawingFigure 1~2
  • EP4314854B1 patent drawingFigure 3~4
  • EP4314854B1 patent drawingFigure 5~6

AI summary

The invention relates to a method (100) for denoising experimental data (Z1, …, Zi, …, Zn) resulting from multiple acquisitions by a magnetic resonance imaging (MRI) apparatus (1), through analysis (120) of principal components that are selected (130) with a view to obtaining a better compromise between denoising performance and preservation of relevant information on the experimental data in question during the reconstruction (140) thereof to produce denoised experimental data (Z1', …, Zi', …, Zn'). One selection criterion is based on informational indicators quantifying the spatial information contained in images of scores associated with said principal components. The invention makes further provision for a step of applying adaptive filtering (133) that rejects the remaining spatial noise associated with each selected principal component.