Proactive Health Monitoring Using Individualized Analysis of Tissue Elasticity

PI: Ming C. Lin

Co-PI: Ronald Chen and Vladimir Jojic

Senior Investigators:: Jun Lian and Hongtu Zhu

Research Assisants: Shan (Alex) Yang, Junbang (William) Liang, and Tanya Amert

University of North Carolina at Chapel Hill

 

System Overview: (a) elasticity parameter estimation using medical image analysis and biomedical modeling; (b) longitudinal analysis of patient data against clinical criteria to establish predictive models; and (c) clinical classifier using machine learning and individual tissue elasticity.

 

Motivation

To develop a novel computational framework for proactively monitoring the wellbeing of at-risk groups through individualized analysis of tissue elasticity, taking into account of other explanatory variables, including ages, family history, genetic disposition, chronic conditions, and other factors.

Approaches

Efficient algorithms for non-invasive, image-based techniques for automatic extraction of tissue elasticity;

New predictive models for cancer staging and grading based on patient-specific parameters;

Novel regression models and inference procedures for survival analysis;

A health monitoring system for at-risk groups based on individual tissue elasticity along with other variables.

Results

Synthetic Experiment for Multi-Region Elasticity Parameter Reconstruction

Tumor-to-Region ratio

0.022

0.14

0.30

0.49

0.65

0.76

0.85

Region with tumor (kPa)

30.63

31.54

39.18

43.93

51.23

61.16

71.01

Region with normal tissue (kPa)

29.15

28.89

31.22

30.17

31.49

29.31

30.46

As the volume ratio of tumor to the embedded region increases, so does the average stiffness value for the tumor embedded region.


Tumor Elasticity Parameter (kPa)

70

140

210

280

350

420

490

Region 1 with tumor (kPa)

51.23

112.92

157.44

186.78

202.22

254.20

272.58

Region 2 with normal tissue (kPa)

31.49

28.28

30.04

28.56

27.62

29.61

25.18

As the tumor becomes more stiff, the average elasticity value in the tumor region increase as well.



Real Patient Cancer Stage Correlation Experiment


 

Publications with Videos

  • Classification of Prostate Cancer Grades and T-Stages based on Tissue Elasticity Using Medical Image Analysis (MICCAI 2016)
  • MaterialCloning: Acquiring Elasticity Parameters from Images for Medical Applications (IEEE TVCG2015)
  • Bayesian Estimation of Non-Rigid Mechanical Parameters Using Temporal Sequences of Deformation Samples (IEEE ICRA 2015)
  • Simultaneous Estimation of Elasticity for Multiple Deformable Bodies (CASA/CAVW 2015)

    • Other Related Videos

  • Generalization of the Method to Cloth

    • Technical Report

    1.  "Simulation-Based Estimation of Elasticity Parameters for Multi-Body Deformation", Janunary 2014.

    2.  "MaterialCloning: Acquiring Elasticity Parameters from Images", November 2014.

    Acknowledgement

    This project is supported in part by the joint NSF-NIH Smart and Connected Health Program:

    NIH Grant Number: R01 EB020426-01