RADISTAT Radiomic Spatial Textural Descriptor for Tissue Heterogeneity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radiomics-based approaches inadequately capture the diversity of radiomic expression in tumors, leading to incomplete characterization of tissue heterogeneity and suboptimal prediction of disease outcomes and treatment responses in cancers like Glioblastoma multiforme (GBM) and rectal cancer (RCa).
Innovation Solution
The implementation of a Radiomic Spatial Textural Descriptor (RADISTAT) that characterizes the spatial arrangement and textural appearance of radiomic features within a target region of interest, using superpixel clustering and re-partitioning to identify hot and cold spots, and quantify their adjacency, providing a more comprehensive description of tissue heterogeneity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional statistical descriptors (mean, skewness) are used to characterize radiomic features, then the approach is simple and computationally efficient, but the diversity of radiomic expression and tissue heterogeneity is inadequately captured
Solution Approach 1:
The patent segments the radiomic feature distribution into multiple discrete bins representing different expression levels (e.g., low, medium, high). This segmentation allows the method to capture the diversity of radiomic expression across different tissue sub-compartments, moving beyond single statistical descriptors to a multi-level characterization that better reflects tissue heterogeneity.
Solution Approach 2:
The patent introduces a new dimension to radiomic analysis by creating a histogram-based representation that combines spatial frequency information with expression level distribution. This transforms the analysis from simple statistical moments to a multi-dimensional characterization that includes both the magnitude and spatial organization of radiomic features, enabling more precise tissue heterogeneity assessment.
2Reliability
If conventional statistical methods are used for treatment response prediction, then the computational approach is straightforward, but prediction accuracy for treatment response and survival outcomes is suboptimal
Solution Approach 1:
The patent changes the parameters used for prediction from conventional single-value statistical descriptors to a comprehensive set of histogram-based features that capture distribution shape, spread, and spatial frequency. This parameter transformation enables the predictive model to utilize more informative characteristics of radiomic expression patterns, thereby improving prediction accuracy for treatment response and survival outcomes.
Solution Approach 2:
The patent creates a composite radiomic descriptor that integrates multiple types of information: histogram bin frequencies, spatial frequency content, and texture metrics. This composite approach combines different aspects of radiomic feature organization into a unified predictive framework, enhancing the reliability of treatment response and survival predictions compared to single-metric methods.
Data Source
AI summary
Embodiments access an image of a region of interest (ROI) demonstrating cancerous pathology; extract radiomic features from the ROI; define a radiomic feature expression scene based on the ROI and radiomic features; generate a cluster map by superpixel clustering the expression scene; generate an expression map by repartitioning the cluster map into expression levels; compute a textural and spatial phenotypes for the expression map based on the expression levels; construct a radiomic spatial textural (RADISTAT) descriptor by concatenating the textural and spatial phenotypes; provide the RADISTAT descriptor to a machine learning classifier; receive, from the machine learning classifier, a first probability that the ROI is a responder or non-responder, or a second probability that the ROI will experience long-term survival or short-term survival, based, at least in part, on the RADISTAT descriptor; and generate a classification of the ROI as a responder or non-responder, or long-term survivor or short-term survivor.