DCE-MRI Analysis Algorithm for Rectal Cancer Response Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing data from Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) in patients with locally advanced rectal cancer after neoadjuvant radio-chemotherapy are subjective and fail to accurately distinguish between complete, partial, and non-responders due to reliance on visual assessment of Time-Intensity curves.
Innovation Solution
A semi-quantitative analysis method using a computer-implementable algorithm that calculates specific form descriptors from DCE-MRI data, including maximum signal difference and wash-out gradient, to create a linear classification index, the Standardized Index of Shape (SIS), which objectively discriminates between responders and non-responders to neoadjuvant radio-chemotherapy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If visual assessment of Time-Intensity curves is used to analyze DCE-MRI data, then the method is simple and easy to operate, but the measurement precision and reliability are reduced due to subjectivity
Solution Approach 1:
The patent replaces the manual visual assessment method (mechanical/human system) with an automated computer-implementable algorithm (computational system). The algorithm automatically calculates form descriptors from DCE-MRI data and generates classification indices, eliminating subjective human interpretation while maintaining operational simplicity through automation.
Solution Approach 2:
The system enables self-service analysis where the computer algorithm independently processes DCE-MRI data without requiring expert radiologist intervention for curve interpretation. The automated calculation of form descriptors and classification indices allows the system to serve itself in generating diagnostic conclusions.
2Ease of operation
If visual assessment of Time-Intensity curves is used to analyze DCE-MRI data, then the method is simple and easy to operate, but the reliability in discriminating between complete, partial, and non-responders is reduced
Solution Approach 1:
The patent transforms the analysis from qualitative visual assessment to quantitative measurement by calculating specific form descriptors (maximum signal difference, wash-out gradient) and combining them into classification indices. This parameter transformation enables reliable discrimination between responder categories through objective numerical thresholds.
Solution Approach 2:
The automated algorithm replaces subjective human visual assessment with consistent computational processing, ensuring reliable and reproducible discrimination between complete, partial, and non-responders across different patients and time points.
3Ease of manufacture
If morphologic Magnetic Resonance Imaging is used to assess tumour reduction, then the method is simple to perform, but the measurement precision is insufficient to distinguish between fibrosis and tumour residue
Solution Approach 1:
The patent shifts from morphological assessment (simple but imprecise) to functional assessment using DCE-MRI parameters (complex but precise). By measuring contrast medium uptake and wash-out characteristics, the method achieves sufficient measurement precision to differentiate fibrosis from viable tumour residue while maintaining clinical feasibility.
Solution Approach 2:
The analysis focuses on specific local characteristics of the tumour tissue through form descriptors of Time-Intensity curves, examining local contrast enhancement patterns that reveal tissue viability. This local quality analysis enables precise differentiation between fibrotic and tumorous tissues within the treatment area.
Data Source
AI summary
Method for analyzing data provided by Magnetic Resonance Imaging with dynamic administration of contrast medium in patients with locally advanced rectal cancer, after neoadjuvant radio-chemotherapy, comprising: making available pre- and post-treatment digital images, having one or more regions of interest identified; splitting the regions of interest into portions; calculating first and second time-intensity curves for each portion of the regions of pre- and post-treatment interest, respectively; calculating the maximum signal difference and gradient of the wash-out section for the curves calculated; calculating the median value of the maximum signal difference and of the gradient calculated; calculating the percentage variation between the median values of the maximum signal difference and of the gradient of the wash-out section calculated for each of the first and second curves; linearly combining percentage variations of the maximum signal difference and of the gradient of the wash-out section, to define a relative linear classification index.