MRI Intensity Standardization via Tissue Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing medical image intensity standardization techniques fail to accurately match spatially corresponding tissue intensities across different scanners, leading to inefficient intensity adjustments and loss of biological meaning, particularly in multi-centric settings like the Alzheimer's Disease Neuroimaging Initiative.

Innovation Solution

The development of a novel automated technique, STI, which incorporates tissue spatial intensity information by using spatial correspondences and available tissue masks to adjust intensities, adding minimal and maximal data pairs for precise interpolation and employing pre-processing steps like scaling, filtering, and intensity adjustment formulas for robust standardization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If histogram matching techniques are used for intensity standardization, then the image histogram can be matched to a reference histogram, but the spatial correspondence between tissue intensities is lost and biological meaning is distorted

Engineering Contradiction:
Improvehistogram matching precisionVSAvoidbiological meaning
Core Design Contradiction:
Manufacturing precisionVSLoss of information

Solution Approach 1:

The patent applies local quality by performing intensity standardization separately for different tissue types (grey matter, white matter, CSF) rather than applying a global histogram matching transformation. Each tissue type receives a tailored intensity mapping function that preserves its specific biological characteristics while achieving standardization, thus maintaining local biological meaning while improving overall standardization precision.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the image into different tissue types using tissue masks before applying intensity standardization. This segmentation allows the method to treat each tissue type independently, preserving the spatial correspondence and biological meaning of each tissue while achieving histogram matching at the tissue level rather than globally, thereby resolving the contradiction between matching precision and biological meaning preservation.

Inventive Principle:
Principle #1Segmentation

2Reliability

If global intensity transformation is applied to standardize MRI images from different scanners, then scanner effects can be reduced, but tissue-specific intensity variations are not adequately addressed

Engineering Contradiction:
Improvescanner effect reductionVSAvoidtissue-specific intensity accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent implements local quality by computing separate intensity mapping functions for each tissue type (grey matter, white matter, CSF) based on their specific intensity distributions in the reference image. This allows the method to reduce scanner effects globally while simultaneously preserving tissue-specific intensity characteristics, as each tissue receives a customized transformation that maintains its measurement precision.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the approach from a single global intensity transformation parameter to multiple tissue-specific transformation parameters. By deriving separate mapping functions for each tissue type based on their respective intensity histograms and spatial distributions, the method achieves both scanner effect reduction and tissue-specific intensity accuracy through parameter differentiation.

Inventive Principle:
Principle #35Parameter changes

3Ease of manufacture

If simple histogram matching is used, then the method remains simple and robust, but it cannot distinguish between different tissue types with similar intensity profiles

Engineering Contradiction:
Improvemethod simplicityVSAvoidtissue type discrimination
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent maintains method simplicity by using basic histogram matching operations while improving tissue type discrimination through segmentation with tissue masks. The masks separate different tissue types before histogram matching, allowing the simple histogram operation to be applied tissue-specifically. This combines the robustness of simple histogram matching with the precision of tissue-type-aware processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces tissue masks as an intermediary element between the input image and the histogram matching process. These masks enable the simple histogram matching algorithm to operate on segmented tissue regions rather than the entire image, thereby maintaining algorithmic simplicity while achieving accurate tissue type discrimination through the mediating role of the masks.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP2591456B1Image intensity standardization
Publication Date: 2018.11.28 UNIVERSITE LAVAL
  • EP2591456B1 patent drawingFigure 1
  • EP2591456B1 patent drawingFigure 2
  • EP2591456B1 patent drawingFigure 3

AI summary

Intensity standardization of MRI data sets aims at correcting scanner- dependent intensity variations. An automatic technique, called STI, which shares the simplicity and robustness of histogram-matching techniques, but also incorporates tissue spatial intensity information, has been discovered. The method comprises registering a medical image to a standard image; applying one or more masks to the medical and standard images for isolating certain specific image components; determining the most common intensity data pair between the medical and standard images for each isolated image component; calculating a formula that joins the most common intensity data pair of each image component; and interpolating an intensity data adjustment using the formula and applying it to the medical image data to generate a standardized version of the medical image.