Thermal Joint Imaging With Local ROI Features for Arthritis Scoring

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing arthritis assessment methods using thermal imaging rely on descriptive statistical features, neglecting more complex thermal patterns, which limits the accuracy of joint inflammation detection and quantification.

Innovation Solution

A machine learning-based method that identifies local regions of interest (ROIs) in thermal images, using corner or blob detectors, and combines these with visible image information to determine a joint inflammation score through feature extraction and data fusion, employing techniques like SIFT and convolutional autoencoders.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If descriptive statistical features are used for thermal image analysis, then the assessment is simpler and faster, but the accuracy and sensitivity of inflammation characterization is reduced

Engineering Contradiction:
Improveassessment speedVSAvoidinflammation characterization accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies segmentation by dividing the thermal image into multiple regions of interest (ROIs) corresponding to different joints and anatomical structures. This allows the system to extract localized thermal features from specific areas while maintaining overall processing efficiency. The segmentation enables parallel processing of multiple regions, preserving speed while capturing detailed local thermal patterns that improve inflammation characterization accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements local quality by extracting different types of features from different regions of the thermal image. Specifically, it identifies both global thermal patterns and local thermal features (such as temperature gradients, hot spots, and thermal asymmetry) from specific joints and soft tissues. This localized feature extraction allows the system to maintain high assessment speed while significantly improving the accuracy of inflammation characterization by focusing computational resources on clinically relevant areas.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If complex local thermal features are extracted from thermal images, then the sensitivity and specificity of arthritis assessment is improved, but the computational complexity and processing time increases

Engineering Contradiction:
Improvearthritis assessment accuracyVSAvoidcomputational processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing preprocessing operations on thermal images before detailed feature extraction. This includes noise filtering, contrast enhancement, and preliminary segmentation to identify potential regions of interest. By preparing the images in advance with these preliminary operations, the system reduces the computational burden during the complex feature extraction phase, thereby maintaining high assessment accuracy while managing processing complexity and time more effectively.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts different types of features from different regions of the thermal image. Specifically, it identifies both global thermal patterns and local thermal features (such as temperature gradients, hot spots, and thermal asymmetry) from specific joints and soft tissues. This localized feature extraction allows the system to maintain high assessment speed while significantly improving the accuracy of inflammation characterization by focusing computational resources on clinically relevant areas.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If manual ROI selection is used in thermal imaging, then the anatomical precision is improved, but the time consumption and operator dependency increases

Engineering Contradiction:
Improveanatomical localization accuracyVSAvoidassessment time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies self-service by implementing automated algorithms that perform region of interest (ROI) selection without requiring manual intervention. The system uses computer vision techniques, including edge detection, contour analysis, and pattern recognition, to automatically identify and delineate joints and soft tissue regions in thermal images. This automated ROI selection maintains high anatomical localization accuracy while eliminating the time consumption and operator dependency associated with manual selection, enabling rapid and consistent assessment.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the manual mechanical process of ROI selection with an automated computational system. Instead of requiring operators to manually draw or select regions based on anatomical knowledge, the system uses image processing algorithms and machine learning models to automatically identify and segment relevant anatomical structures. This substitution of manual mechanics with automated computational methods maintains precision while dramatically reducing assessment time and eliminating inter-observer variability.

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

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

Enhances the accuracy of arthritis assessment by providing a more comprehensive characterization of inflammation, allowing for automated, non-invasive, and cost-effective joint inflammation detection and quantification, suitable for remote patient monitoring.

Implementation Method 1

it has been found a correlation between the ankle thermal measurements and the arthritis activity and severity in rat models... the thermal analysis of rat paws in the arthritis model demonstrated a significant increase in the mean temperature difference

Methodology Applied
Scientific EffectInfrared radiation detection: Infrared Radiation

Data Source

PatentEP4089578B1Method to extract local thermal features of joint inflammation in thermal imaging, and device implementing such a method
Publication Date: 2026.04.29 SINGULARITY BIOMED SL
  • EP4089578B1 patent drawingFigure 1
  • EP4089578B1 patent drawingFigure 2A~2B
  • EP4089578B1 patent drawingFigure 2C~3

AI summary

The invention refers to a method for processing of one or more input thermal images (1) of the hands or feet of a subject, to obtain a plurality of local thermal features (7), characteristics of an inflammatory process. Said method comprises acquiring one or more input thermal images (1) with an image acquisition sensor; preprocessing (2) the thermal images (1) to obtain one or more enhanced images (3); and selecting one or more regions of interest (4') of the enhanced images (3). Local thermal features (7) are described from each ROI comprising information about the inflammation of the hands or feet. The invention also comprises a device performing such a method. The device is useful especially in the area of so-called remote patient monitoring.