Tumor characterization and outcome prediction through quantitative measurements of tumor-associated vasculature
Inventors
Madabhushi, Anant • Braman, Nathaniel
Assignees
Case Western Reserve University • US Department of Veterans Affairs
Publication Number
US-11896349-B2
Publication Date
2024-02-13
Expiration Date
2040-12-09
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
Embodiments discussed herein facilitate determination of a response to treatment and/or a prognosis for a tumor based at least in part on features of tumor-associated vasculature (TAV). One example embodiment is a method, comprising: accessing a medical imaging scan of a tumor, wherein the tumor is segmented on the medical imaging scan; segmenting tumor-associated vasculature (TAV) associated with the tumor based on the medical imaging scan; extracting one or more features from the TAV; providing the one or more features extracted from the TAV to a trained machine learning model; and receiving, from the machine learning model, one of a predicted response to a treatment for the tumor or a prognosis for the tumor.
Core Innovation
The invention facilitates determination of a response to treatment and/or a prognosis for a tumor based at least in part on features of tumor-associated vasculature (TAV). It includes methods comprising accessing medical imaging scans of a tumor that is segmented on the imaging scan, segmenting tumor-associated vasculature based on the medical imaging scan, extracting one or more features from the TAV, providing those features to a trained machine learning model, and receiving from the model a predicted response to a treatment or a prognosis for the tumor.
The problem addressed relates to the importance of angiogenesis in tumor growth and therapeutic outcome. Existing quantitative analyses on dynamic contrast enhanced (DCE) MRI provide indirect characterization of tumor vascularization, but direct computational analysis of the tumor-associated vessel network remains under-explored as a potential marker of therapy response. Complex 3-dimensional shapes and poor blood flow in surrounding vasculature may indicate more aggressive tumors and impede therapeutic agent delivery.
Claims Coverage
The patent includes multiple independent claims focused on methods and apparatuses for determining tumor therapy response or prognosis by analyzing features of tumor-associated vasculature using machine learning models.
Feature extraction from three-dimensional segmentation of tumor-associated vasculature
Extracting features from the three-dimensional segmentation of tumor-associated vasculature and its projections that quantify spatial orientation and morphology to represent TAV characteristics in three dimensions.
Use of trained machine learning model for predicting treatment response or prognosis
Providing extracted TAV features to a trained machine learning model that predicts either a response to treatment or prognosis for the tumor based on these features.
Definition of TAV morphology and spatial organization features
Quantifying TAV morphology using features such as torsion per branch, curvature statistics per branch, vessel volume and length, and fraction of vessels entering the tumor, as well as spatial orientation features such as vessel orientation along various projection images (XY, XZ, YZ, rotation-elevation, distance-rotation, distance-elevation).
Incorporation of tumor-associated vasculature function features
Including functional features that measure temporal dynamics of contrast agent in the tumor or TAV, exemplified by signal enhancement ratio, time to peak enhancement, rate of uptake, or rate of washout.
Training of machine learning models using features extracted from TAV for response and prognosis prediction
Accessing a training set of medical imaging scans with associated tumor and response/prognosis data, segmenting TAV, extracting feature values, selecting the best predictive features, and training various machine learning models (e.g., logistic regression, Cox regression, random forest, support vector machine, neural networks) to generate predictive algorithms.
The claims cover innovative methods and apparatuses that extract quantitatively defined morphological, spatial, and functional features from segmented tumor-associated vasculature in clinical medical imaging, and apply trained machine learning models to predict therapeutic response or prognosis for tumors, with detailed feature sets capturing vessel shape, organization, and dynamics.
Stated Advantages
Provides value in identifying patients who will respond to neoadjuvant chemotherapy before administration of treatment.
Enables prediction of response and prognosis using features of tumor-associated vessel network morphology and function extracted from routine medical imaging.
Improves predictive performance over existing clinical variables and functional tumor volume measures.
Provides a pan-cancer, cross-modality and multi-treatment predictive and prognostic biomarker applicable across breast and lung cancers and imaging modalities like MRI and CT.
Demonstrates robustness to errors in vessel segmentation and imaging heterogeneity, facilitating practical clinical application.
Documented Applications
Predicting neoadjuvant chemotherapy response and prognosis in breast cancer patients using pre-treatment dynamic contrast-enhanced MRI.
Predicting treatment response and progression free survival for non-small cell lung cancer patients receiving platinum-based chemotherapy using pre-treatment CT imaging.
Assessing pathological complete response (pCR), major pathological response (MPR), and RECIST-defined responses in various cancer treatment cohorts.
Generating quantitative tumor-associated vasculature (QuanTAV) morphological and spatial organization features and incorporating them into machine learning classifiers to predict therapy outcomes and survival across different cancers and treatment regimens.
Potential use in patient stratification for anti-angiogenic therapies and personalized treatment planning based on predicted tumor response and prognosis.
Interested in licensing this patent?