Image Registration Scaling for CT Size Modification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The accuracy of determining size modifications of objects, such as lung nodules, in computed tomography image data sets is compromised due to segmentation errors caused by noise and artifacts, leading to ill-posed problems with large variations in output data from small variations in input data.

Innovation Solution

An apparatus and method that register regions of interest from different image data sets using scaling transformations, generating scaling values to determine size modifications without segmenting the objects, thereby improving accuracy by avoiding segmentation errors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If segmentation is performed separately on each computed tomography image data set to determine volume, then volume measurement can be obtained, but segmentation errors due to noise and artifacts cause large variations in output data from small variations in input data, reducing accuracy

Engineering Contradiction:
Improvevolume measurement accuracyVSAvoidconsistency of volume determination
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent combines multiple image data sets into a single merged data set, performing only one segmentation operation rather than separate segmentations for each time point. This merging approach ensures that the same segmentation algorithm and parameters are applied consistently across all images, eliminating variability introduced by repeated segmentation operations and thereby improving both measurement precision and reliability.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent performs the segmentation operation in advance on a merged data set that incorporates all image information, rather than performing segmentation separately on each individual image. This preliminary action on the consolidated data ensures that small variations in individual images do not propagate through multiple segmentation operations, reducing the amplification of errors and improving consistency.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If separate segmentation is performed on each image data set, then volume can be determined at each time point, but small variations in image data values due to noise and artifacts lead to large variations in segmented volumes

Engineering Contradiction:
Improveability to determine volume at multiple time pointsVSAvoidvolume determination accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent merges multiple image data sets into a single comprehensive data set before performing segmentation. This allows volume determination at multiple time points to be derived from a single segmentation operation, eliminating the compounding of segmentation errors that would occur with separate operations and thereby maintaining high measurement precision while preserving productivity.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a merged data set that serves as a master reference, from which volume measurements at different time points can be derived through consistent transformation and comparison. This master copy approach ensures that all measurements trace back to the same segmentation result, preventing divergence due to noise and artifacts in individual images.

Inventive Principle:
Principle #26Copying

3Loss of time

If multiple separate segmentation operations are performed on image data sets, then volume changes can be tracked over time, but segmentation errors accumulate and reduce the accuracy of size modification determination

Engineering Contradiction:
Improvetime-based size change tracking capabilityVSAvoidsize modification accuracy
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The patent merges image data sets from multiple time points into a single integrated data structure, allowing temporal analysis to be performed on a unified segmentation result. This eliminates the accumulation of segmentation errors that would occur with repeated separate operations, while preserving the ability to track size changes over time through consistent geometric transformations and comparisons.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP2319013B1Apparatus for determining a modification of a size of an object
Publication Date: 2018.02.28 KONINKLIJKE PHILIPS NV
  • EP2319013B1 patent drawingFigure 1
  • EP2319013B1 patent drawingFigure 2
  • EP2319013B1 patent drawingFigure 3~4

AI summary

The present invention relates to an apparatus for determining a modification of a size of an object. The apparatus comprises a registration unit (13) for registering with respect to each other a first region of interest in a first image data set showing an object at a first time and a second region of interest in a second image data set showing the object at a second time being different from the first time, wherein the registration unit is adapted to generate a scaling value by performing at least a scaling transformation for registering the first region of interest and the second region of interest with respect to each other. The apparatus further comprises a modification value determination unit (14) for determining a modification value, which indicates the modification of the size of the object, depending on the generated scaling value.