Thermal-3D Skin Segmentation for Repeatable Abnormality Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for measuring skin abnormalities in laboratory animals, such as tumors or wounds, are prone to human error, subjectivity, and lack repeatability, and it is challenging to distinguish between the abnormality and the natural curvature of the skin using three-dimensional models.
Innovation Solution
A method that combines temperature datasets with three-dimensional skin profiles to segment and characterize skin abnormalities, using statistical models to enhance segmentation and improve model accuracy, and incorporates thermal imaging and stereoscopy to differentiate skin regions based on temperature variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual measurement using calipers is used, then measurements can be obtained, but human error and subjectivity increase, reducing measurement precision and reliability
Solution Approach 1:
The patent replaces manual mechanical measurement with calipers by implementing an automated optical imaging system that captures images of skin abnormalities and automatically measures their dimensions. The system uses image processing algorithms to detect abnormality boundaries and calculate measurements, eliminating human error and subjectivity while improving both measurement precision and repeatability.
Solution Approach 2:
The system enables self-service measurement by allowing the imaging system to automatically identify, segment, and measure skin abnormalities without requiring manual intervention. The automated algorithms process images, distinguish abnormalities from surrounding skin, and compute measurements independently, ensuring consistent and repeatable results across different users and time points.
2Shape
If three-dimensional skin profiling is used, then the shape of skin abnormalities can be captured, but it becomes difficult to distinguish between the abnormality and natural skin curvature
Solution Approach 1:
The patent applies local quality analysis by examining specific local features of the skin surface, such as curvature variations, texture patterns, and temperature differences, to distinguish abnormalities from natural skin contours. The system analyzes localized regions rather than treating the entire skin surface uniformly, enabling accurate differentiation between pathological changes and normal anatomical variations.
Solution Approach 2:
The system moves from two-dimensional image analysis to three-dimensional surface profiling by capturing depth information and spatial coordinates. This dimensional enhancement allows the system to differentiate between abnormalities that protrude from or indent into the skin surface versus normal skin curvature, providing more accurate shape characterization and boundary detection.
3Productivity
If automated image processing is implemented, then measurement time is reduced, but complexity of processing the dataset increases
Solution Approach 1:
The patent implements segmentation by dividing the complex image processing task into distinct stages: image acquisition, pre-processing, abnormality detection, boundary segmentation, and measurement calculation. This modular approach manages computational complexity by handling each stage separately with specialized algorithms, enabling automated processing to achieve high productivity without overwhelming system complexity.
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 reduces noise and improves the accuracy of identifying and measuring skin abnormalities by distinguishing between abnormal and normal skin regions, providing reliable and repeatable numerical parameters like volume, thereby enhancing the precision of tumor or wound assessments.
Implementation Method 1
at least a temperature dataset obtained by measuring the temperature of each of a plurality of points of a region of the skin
Implementation Method 2
The model of the skin may, for example, be obtained by combining images of the region of the skin captured using a plurality of cameras. For example, the images may be combined using stereoscopy to create the model.
Data Source
AI summary
A method is proposed for identifying (“segmenting”) at least one portion of the skin of an animal which is a region of interest (e.g. a portion which is subject to an abnormality such as a tumor). The method uses at least a temperature dataset obtained by measuring the temperature of each of a plurality of points of a region of the skin. An initial segmentation may be performed using the temperature data based on a statistical model, in which each point is segmented based on its temperature and optionally that of its neighbors. The initial segmentation based on the temperature data may be improved using a three-dimensional model of the profile of the skin, and the enhanced segmentation may be used to improve the three-dimensional model.