Radiomic Mesenteric Fat Analysis for Crohn's Disease Therapy Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current biomarkers for Crohn's disease, such as eosinophil sedimentation rate, C-reactive protein, and stool Calprotectin, have low sensitivity and specificity for detecting the presence and severity of Crohn's disease, and existing imaging methods rely on manual quantification of visceral adipose tissue (VAT) which is time-consuming and not very accurate, limiting their usefulness in clinical settings.
Innovation Solution
The use of radiomics to extract quantitative features from medical images, specifically from mesenteric fat regions on magnetic resonance enterography (MRE) images, to quantify responses and classify disease severity, facilitating the prediction of therapy response in Crohn's disease patients through automated segmentation and machine learning classifiers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual quantification of visceral adipose tissue (VAT) is used, then measurement can be performed, but it requires significant expert interaction and is time-consuming
Solution Approach 1:
The system performs automated segmentation of visceral adipose tissue using computer algorithms that independently identify and quantify VAT regions on CT images without requiring expert interaction. The algorithm automatically thresholds images, segments anatomical structures, and calculates VAT volume and density metrics.
Solution Approach 2:
Manual expert measurement and visual assessment of VAT is replaced with automated image processing algorithms and computational methods. The system uses computer-based segmentation techniques, thresholding algorithms, and automated volumetric calculations to substitute the mechanical process of manual measurement.
2Reliability
If volumetric measures of VAT are used, then disease activity can be detected with 80% specificity, but sensitivity varies between 40% and 80%
Solution Approach 1:
Instead of using only global volumetric measures of VAT, the system analyzes local properties including VAT density, texture characteristics, and spatial distribution patterns. This local quality analysis provides additional discriminatory information that improves sensitivity for detecting active Crohn's disease while maintaining high specificity.
Solution Approach 2:
The system transitions from one-dimensional volumetric measurements to multi-dimensional analysis by incorporating density metrics, texture features, and spatial distribution patterns. This dimensional expansion allows the system to detect subtle changes in VAT properties that correlate with disease activity, thereby improving sensitivity.
3Reliability
If existing biomarkers such as CRP or stool Calprotectin are used, then disease presence can be detected, but sensitivity and specificity vary widely and they do not provide localized information
Solution Approach 1:
The system segments the abdominal cavity into distinct anatomical regions and specifically identifies visceral adipose tissue surrounding the small bowel. This segmentation allows the system to isolate and analyze VAT in the specific anatomical location where Crohn's disease pathology occurs, providing localized information that systemic biomarkers cannot provide.
Solution Approach 2:
Visceral adipose tissue serves as an intermediary marker that reflects local inflammatory processes in the bowel. The VAT density and texture changes act as a mediator between the hidden bowel inflammation and the observable imaging findings, providing localized information about disease activity in the gastrointestinal tract.
Data Source
AI summary
Embodiments discussed herein facilitate predicting response to therapy in Crohn's disease. A first set of embodiments discussed herein relates to accessing a radiological image of a region of tissue demonstrating Crohn's disease associated with a patient; defining a mesenteric fat region by segmenting mesenteric fat represented in the radiological image; extracting a set of radiomic features from the mesenteric fat region; providing the set of radiomic features to a machine learning classifier configured to compute a probability of response to therapy in Crohn's disease based, at least in part, on the set of radiomic features; receiving, from the machine learning classifier, a probability that the region of tissue will respond to therapy; generating a classification of the patient as a responder or non-responder based, at least in part, on the probability; and displaying the classification.