3D Morphology-Based Disease Severity Prediction From Medical Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for quantifying morphological characteristics in medical images fail to accurately predict disease severity due to reliance on coarse geometric quantifiers that do not fully utilize 3-dimensional morphological information, leading to low prediction accuracy.
Innovation Solution
Reconstruct meshes from 3-dimensional segmentation masks, simplify and smooth them to remove inconsistencies, compute the distribution of curvatures, and use these features along with baseline characteristics in a supervised machine learning model for disease severity prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If coarse geometric quantifiers are used for morphological characterization, then the quantification process is simple, but the prediction accuracy of disease severity is low
Solution Approach 1:
The patent transitions from 2D image analysis to 3D volume analysis by reconstructing 3D meshes from segmented medical images. This dimensional change enables extraction of volumetric morphological features and 3D curvature distributions, which provide significantly more information for disease severity prediction while maintaining computational feasibility through efficient mesh processing algorithms
Solution Approach 2:
The patent changes the type of morphological parameters extracted from coarse geometric quantifiers (area, volume) to fine-grained local features including curvature distributions, mesh energies, and surface irregularities. These parameter changes enable the system to capture subtle pathological deformations that correlate with disease severity, directly improving prediction accuracy
2Measurement precision
If 3-dimensional morphological information is fully utilized, then the prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by first segmenting the medical images to isolate anatomical structures, then reconstructing simplified 3D meshes before extracting morphological features. This preliminary processing reduces the complexity of subsequent curvature calculations and feature extraction by working with cleaned, structured mesh data rather than raw image volumes
Solution Approach 2:
The patent extracts only the essential morphological features (curvature distributions, mesh energies, surface irregularities) from the 3D meshes that are most relevant to disease severity prediction. This selective extraction avoids computing all possible geometric parameters, reducing computational complexity while maintaining high prediction accuracy
Data Source
Figure 1
Figure 2A~2B
Figure 3A
AI summary
It is provided a method and system for generating a machine learning model for prediction of the severity risk of disease from a plurality of 3-dimensional medical images. The method comprises the steps of: providing a plurality of 3-dimensional medical images, each image of the plurality of 3-dimensional medical images comprising at least one anatomical structure of a patient; obtaining at least one 3-dimensional segmentation mask from each image of the plurality of 3-dimensional medical images, each mask of said at least one 3-dimensional segmentation mask comprising at least one anatomical structure; reconstructing mesh from each mask of said at least one 3-dimensional segmentation mask; extracting morphological features from each reconstructed mesh; and training a machine learning model to perform classification of the severity risk of disease, the training is performed on the extracted morphological features. It is further provided a method and system for prediction of the severity risk of disease from a plurality of 3-dimensional medical images by applying a machine learning model generated for the same purpose.