BE PART OF THE REVOLUTION TRANSFORMING HEALTHCARE WITH AI - - PowerPoint PPT Presentation

1000 ATTENDEES 80 SPEAKERS 10 WORKSHOPS 2 SOCIAL EVENTS 1 AIMed19

www.aimed.events/northamerica-2019/ #AIMed19

BE PART OF THE REVOLUTION

TRANSFORMING HEALTHCARE WITH AI

CALIFORNIA — THE RITZ-CARLTON, LAGUNA NIGUEL 11–14 DECEMBER 2019

Role of Precision Medicine in the Management of Congenital Heart Disease

Sanjeet Hegde, MD, PhD Co-Director of Research, Heart Institute Program Director of 3D Innovation Lab Medical Director of Advanced Cardiac Imaging Rady Children’s Hospital San Diego/ UCSD

AIMed NORTH AMERICA, CALIFORNIA 11–14 DECEMBER 2019

Disclosures: None

Integrative Approaches to Biomedical Science

Reductionism Integration

20th Century Biomedical Science 21st Century Biomedical Science

Molecular biology Genomics Proteomics Structural biology Comp Modeling Simulation Bioengineering

parts catalog

Courtesy: Prof. Andrew McCulloch

Big Data in Congenital Heart Disease ?

Non-invasive imaging

Better diagnosis

“I can see it much more clearly now – but I already knew that”

Registration and characterization Dimension reduction analysis Pattern recognition

Data Acquisition Patient Diagnosis Structural/ Functional modeling Statistical Analysis

R A D I O L O G Y M E D I C I N E C O M P U T E R S C I E N C E B I O E N G I N E E R I N G M A T H E M A T I C S

Software Development

Fonseca et al. Bioinformatics 27(16): 2288–2295; 2011

Computational cardiac atlas integrates huge amounts of otherwise disconnected information to discover the patterns that represent their internal logic or relationships

• Prof. Alistair Young

You can tell by the shape…

Mauger C et al 2019

What we set out to do ….

Population based cardiac modeling Model based cardiac MRI analysis Computer-aided cardiovascular diagnosis Personalized cardiac biomechanics Apply this approach to Congenital Heart Disease

Ca Cardiac c Atlas of

• f Congeni

Congenital He Hear art D t Dis iseas ase

CHD -Cardiac Atlas Project- Collaborative Project RCHSD,UC San Diego & University of Auckland (NIH funded-RO1) >1000 patients

Surgical repair in tetralogy of Fallot (ToF)

Repaired Tetralogy of Fallot

VSD Patch Transannular Patch

Suleiman, T. et al. Frontiers in (2015)

ToF is the fastest growing population among patients with congenital heart disease

Recruiting 1500 patients

https://shaunwhite.com/

Cardiac MRI for Congenital Heart Disease

Courtesy: Dr. Albert Hsiao

Stages of Pre-Surgical Modeling

Implement Optimal Surgical Option Medical Imaging Image Processing Virtual Surgery Computational Analysis

CMR image data

PATIENT DATA SHAPE MODEL STATISTICAL ATLAS DISCOVERY

Guide-point modeling Principal component analysis Regression & clustering

Me Mean n LV end nd-dia diastolic

• lic sha

hape e for

• r

the fi five most ab abnorm rmal al modes re relative to the contro rol atlas

Population based Cardiac Modeling

Atlas-based analysis has the potential to reveal new measures of geometry and function, which may provide novel insights into the remodeling processes of disease

LV CIM, Auckland, New Zealand- Prof. Young

Model based Cardiac MRI analysis

Multicenter Study, UK- Bhuva et al

Biventricular Atlas Generation

Biventricular Cardiac Image Modeler (CIM) Patient-Specific Model (3D+Time) Model Accumulation From Several Patients Shape Modes PCA

The most abnormal modes of systolic wall motion are detrimental to global LV function

Mode of SWM Mean z-score Predicted net effect on LV EF (% pts.) 5 2.85

• 9.61

2 1.68

• 7.93

3

• 2.11
• 6.95

10 1.99

• 1.60

18

• 1.78
• 1.23

4 1.29

• 0.91

7 0.96

• 0.81

13

• 3.01
• 0.80

9 0.35

• 0.26

20

• 0.29
• 0.18

16 0.38

• 0.17

19 1.16

• 0.08

17

• 0.27
• 0.07

8 0.62

• 0.05

11 0.33

• 0.05

12

• 0.06

0.00 14

• 1.23

0.03 15 0.47 0.16 6 0.48 0.36 1 0.28 0.65

*p < 0.00125 for 2-sample t-test

ToF Atlas: Mode 1 (22.1%)

Legend

Wireframe ED Solid Mesh ES LV Endocardium Green RV Endocardium Blue Epicardium Red

Computer-aided Cardiovascular Diagnosis

Patient 1

• Restrictive Valve

Motion

• Mild Hypertrophy
• Mild Dilation

Patient 2

• Mild Hypertrophy

Patient 3

• Ventricular septal

defect

• Elongated and

curved LV shape

End-Diastolic Shape Patient 1 Patient 2 Patient 3 Patient 4 Patient 24 Patient 26 Mode 1

• 0.9

1.6

• 0.7

2.2

• 0.7

1.7 Mode 2 0.6 3.1 1.6

• 0.2

2.9 0.8 Mode 3 1.3 1.4

• 6.6

0.5

• 0.7

2.3 Mode 4

• 1.5
• 3.1
• 1.6

0.1

• 3.5
• 0.6

Patient 24

• Moderately dilated

Patient 26

• Low Function

Computer-aided Cardiovascular Diagnosis

Personalized Cardiac Biomechanics

Where are we going with this….

Machine-learning will enable discovery of imaging biomarkers related to:

• Shape
• Wall motion

For earlier prediction of :

• Heart remodeling
• Clinical outcomes
• Functional response to therapy

“Personalized to Patients”

It of course takes a village….

AHA Precision Medicine Platform Grant

• Prof. Andrew McCulloch
• Prof. Alistair Young
• Prof. Jeff Omens
• Prof. James Perry

Avan Suinesiaputra Kathleen Gilbert Charlene Mauger Pau Medrano-Garcia

• Dr. Hari Narayan

Nick Forsch Sachin Govil Justin Ryan

Thank You