From Medical Images to Digital Patients

1
From Medical Imagesto Digital Patients

• Colloquium in memory of
• Gilles Kahn
• 12 January 2007
• Nicholas Ayache, INRIA
• 2004 Route des Lucioles, 06902 Sophia-Antipolis ,
  France
• http//www-sop.inria.fr/asclepios/

2
Gilles Kahn
A scientific friend, a model and a guide
3
Gilles Kahn

• Vision of the role of IT in Health Sciences and
  Biology
• Computational Medical Imaging
• Epidaure (Medical Imaging and Robotics,
  1989-1992 1992-2005)
• Chir (Interventional Imaging Robotics,
  1999-2004)
• Odyssée (Brain Imaging and Neurosciences, 2002-)
• Visages (Medical Imaging and Interventions,
  2005-)
• Asclepios (Biomedical Image Analysis and
  Simulation, 2005-)
• CVRMed95,IS4TM03,MICCAI04 ARCs etc.
• Chair, High Council for Scientific Cooperation
  between France and Israel (BioMedical Imaging
  research program, 2004-06-08)
• Promoted Collaborations with INSERM, CEA,
  Industry, EC
• Created BioMedical Theme at INRIA

4
BioMedical Theme at INRIAModeling and Simulation
for Biology and Medicine

• Anubis
• Asclepios
• Comore
• Demar
• Digiplante
• Helix
• Magnome
• Mere

• Odyssee
• Reo
• Sequoia
• Sosso2
• Symbiose
• Virtual Plants
• Visages

Alcove, Caiman, Geometrica, Macs, Magrit,
First Successful Evaluation in 2005
5
Medical Imaging Technology
Brain

• Large number of imaging modalities of the human
  body
• Provide complementary anatomical/functional info
  with increasing spatial/temporal resolution
• Emerging new modalities/therapies
• Emerging Large Databases

Heart
Mauna Kea Technologies
200 microns
Cardiac Fibers

• Flow/quantity of information too high to be
  exploited efficiently without the help of
  computer science

Da Vinci Surgical Robot
6
Computational Medical Image Analysis (1980 - 2006)

• Assist Diagnosis
• Objective quantitative measurements
• fusion of multimodal, multidimensional,
  multiparameter images
• Assist Therapy
• Plan, simulate (before)
• Control (during), follow-up (after)

J. Duncan N. A, Medical Image Analysis,
Progress over two decades and the challenges
ahead, IEEE Pami, 2000.
7
Spatio-Temporal Cartography of Multiple
Sclerosis Lesions
Gilles Kahn Computational Microscope
D. Rey, G. Subsol, H. Delingette, N.A Automatic
Detection and Segmentation of Evolving Processes
in 3D Medical Images Application to Multiple
Sclerosis. Medical Image Analysis, 6(2)163-179,
June 2002.
8
Virtual Reality
Picinbono-Delingette-Lombardo-N.A. 2001
N.A.-Cotin-Delingette-1998
Marescaux, Delingette, N.A et al. Annals of
Surgery, 1998 Forest, Delingette, N.A, 2003
INRIA-IRCAD
9
Augmented Reality
Collaboration with IRCAD, J. Marescaux
Gilles Kahn first meeting with Pr. Jacques
Marescaux, Sophia Antipolis, 1994
S. Nicolau, A. Garcia, X. Pennec, L. Soler, and
N.A. Augmented reality guided radiofrequency
tumor ablation. Computer Animation and Virtual
World 16(1) 2005.
10
Computational Medical Image Analysis Current
Trends
Interpretation (diagnosis)
prediction of evolution
Medical Images and Signals
Computational Models of human body
Geometry Statistics Physics Physiology
Therapy planning
Therapy simulation
Identification (personalization)
Computational Models for the Human Body, Handbook
of Numerical Analysis, Elsevier, July 2004.
11
An Illustration

• Cardiac Function

Geometry, Statistics Physics,Physiology
12
New Computational Model of the Heart

• to Simulate and Analyze
• Electrical Mechanical pathologies
• from
• Cardiac images Electrophysiology

INRIA Projects Asclepios, Caiman, Macs, Opale,
Reo, Sosso
13
INRIA Collaborative Actions

• Icema 12, CardioSense 3D strongly encouraged
  and supported by Gilles Kahn
• Unique combination of expertise in computer
  science that INRIA could gather together with
  external partners to advance the state of the art

INRIA Projects Asclepios, Caiman, Macs, Opale,
Reo, Sosso
14
CardioSense3D Main Partners
Coordinated by H. Delingette and M. Fernandez

• Image Analysis/Simulation
• Asclepios, INRIA (M.Sermesant)
• Kings College, London (D. Hil
• Clinical Imaging
• Guys Hospital, London (R. Razavi)
• NIH, Washington (E. McVeigh)
• HEGP, Paris (B. Diebold),
• H. Mondor, Créteil (J. Garot)

• Physiology, Control, Numerical Analysis
• Sosso, INRIA (M. Sorine)
• Macs, INRIA (D. Chapelle)
• Reo, INRIA (JF. Gerbeau)
• Industrial
• Philips Research France

15
1. Statistical Geometrical Model

• Database of Hearts (E. McVeigh, NIH)

• Diffusion Tensor MRI
• Average Geometry of Ventricles
• Average Structure of Fibers

J.M. Peyrat, M. Sermesant, X. Pennec, H.
Delingette, C. Xu, E. McVeigh, N.A., MICCAI, Oct
2006
16
Specific Riemannian Framework

• For computations on Tensors (3x3 symmetric
  definite matrices)
• Affine invariant metric
• Similitude invariant metric

Fletcher-Joshi04, Lenglet-Deriche-Faugeras04-05
-06, Batchelor05, etc. Pennec-Arsigny-Fillard03-
05-06
17
2. Electrophysiology
18
Electrical Simulation
Color action potential u
19
3. Electro-mechanical Model
ATP
nano
Inspired by rheological model Of Hill-Maxwell
Model of Bestel-Clément-Sorine
sarcomeres

• active non-linear viscoelastic anisotropic
  incompressible material.

micro
fibers
méso
Kc stiffness u action potential ?c
strain ?c stress
organ
macro
J. Bestel, F. Clément, and M. Sorine. A
Biomechanical Model of Muscle Contraction MICCAI
2001.
20
4. Physiology

• 4 physiological stages (Valves)
• Filling (open)
• Isovolumetric contraction (closed)
• Ejection (open)
• Isovolumetric relaxation (closed)
• Boundary Conditions
• Pre- and after- load ()
• Isovolumetric constraints

() blood pressure in atria (filling) and
arteries (ejection)
21
Electro-Mechanical Simulation

• Action potential u controls contractile element
• u gt 0 Contraction
• u ? 0 Relaxation
• u also modifies stiffness k of the material.

Action Potential u
N.A.-Chapelle-Clément-Coudière-Delingette-
Sermesant-Sorine (FIMH01)
M. Sermesant, H. Delingette, N.A. An
Electromechanical Model of the Heart for Image
Analysis and Simulation. IEEE Transactions on
Medical Imaging. 2006 May25(5)612-25.
22
action potential u
23
action potential u
24
Model Personalization

• Adjust model to imaging and electrical signals

SSFP Steady State Free Precession
Reza Razavi, Kings College London
25
Learning Model Parameters
Compare simulation measurements to adjust
model parameters
Feedback

• Moreau-Villeger, Delingette, Sermesant, Mc
  Veigh, N.A. et al., IEEE Trans. on BioEng. 2006
• Sermesant, Moireau, Sainte-Marie, Hill,
  Chapelle, Razavi et al., Medical Image Analysis
  2006

26
Interventional XMR Imagery in vivo clinical
measurements
Derek Hill, Reza Razavi Kings College, division
of Imaging SciencesThe Guy's, King's and St
Thomas' School of Medicine
Razavi R, Hill DL, Keevil SF, Miquel ME,
Muthurangu V, Hegde S, Rhode K, Barnett M, van
Vaals J, Hawkes DJ, Baker E. Cardiac
catheterisation guided by MRI in children and
adults with congenital heart disease. Lancet.
2003 Dec 6 362(9399) 1877-82.
27
Personalized Healthy Heart
Sermesant M, Rhode K, Sanchez-Ortiz GI, Camara O,
Andriantsimiavona R, Hegde S, Rueckert D,
Lambiase P, Bucknall C, Rosenthal E, Delingette
H, Hill DL, N.A., Razavi R. Simulation of
cardiac pathologies using an electromechanical
biventricular model and XMR interventional
imaging. Med Image Anal. 2005 Oct9(5)467-80.
28
Personalized Pathological Heart
Man, 68, Left Bundle Branch Block
Sermesant M, Rhode K, Sanchez-Ortiz GI, Camara O,
Andriantsimiavona R, Hegde S, Rueckert D,
Lambiase P, Bucknall C, Rosenthal E, Delingette
H, Hill DL, N.A., Razavi R. Simulation of
cardiac pathologies using an electromechanical
biventricular model and XMR interventional
imaging. Med Image Anal. 2005 Oct9(5)467-80.
29
Personalized Therapy Simulation
antero septal Infarct with left bundle branch
block. FE 41
Resynchronization Therapy (biventricular
pace-maker) FE 47
Normal Heart FE 57
Quantitative Assessment of Disease and Therapy
Simulation slowed down 3 times
30 of BV CRT No benefit
30
Medical Perspectives

• More Accurate Diagnosis
• From electrical and mechanical parameters of 3-D
  model

• Realistic Simulation of Therapy
• Biventricular cardiac resynchronization
• RF ablation,
• Stem cells, etc.

CardioSense 3D www-sop.inria.fr/CardioSense3D
31
Future Scientific Challenges

• Invert Model identify electrical and mechanical
  parameters from standard clinical images and ECG
• Couple with Physiological Flow modeling and
  refine energetic modeling (coronary arteries)

CardioSense3D www-sop.inria.fr/CardioSense3D
32
Additional Imaging Modalities
Dynamic C-Arm
In vivo cellular imaging
Coronary tree
Microscopic blood flow
Mauna Kea Technologies
GEMSE
33
Additional Imaging Modalities
Dynamic C-Arm
In vivo cellular imaging
55 microns
Coronary tree
Microscopic blood flow
INRIA-GEMSE
INRIA -Mauna Kea Technologies
34
Heart Fibers
300 µm
Beta-Actin-GFP Mouse
Courtesy of Contag Lab., Stanford University and
Mauna Kea Technologies
35
Computational Medical Imaging

• A new discipline
• To provide powerful tools for a more personalized
  and predictive medicine
• Gilles Kahn had an early vision of the
  potentialities of this discipline and strongly
  supported its development

36
Gilles Kahn

• Scientist
• Vision, IT for Health Sciences and Biology
• Open mindedness and deepness
• Courage to support innovative directions
• Human being
• Charisma, enthusiasm
• Consideration, generosity, humor
• Courage

37
Thank You Gilles!