Model-Guided Therapy and the role of DICOM in Surgery

1
Model-Guided Therapy and the role of DICOM in
Surgery
Heinz U. Lemke, PhD
Chair of Working Group 24 DICOM in Surgery
2
Content

• Introduction (problems and solutions)
• Model guided therapy with TIMMS
• Classification and model classes
• Virtual human model examples
• Conclusion

3
Computer Assisted Digital OR Suite for Endoscopic
MISSProblems Multiple Data Sources
Courtesy of Dr. John Chiu
4
Model Guided Therapy and the Patient Specific
Model

• Model Guided Therapy (MGT) is a methodology
  complementing Image Guided Therapy (IGT) with
  additional vital patient-specific data.
• It brings patient treatment closer to achieving a
  more precise diagnosis, a more accurate
  assessment of prognosis, as well as a more
  individualized planning, execution and validation
  of a specific therapy.
• By definition, Model Guided Therapy is based on a
  Patient Specific Model (PSM) and allows for a
  patient specific intervention via an adapted
  therapeutic workflow.

5
Model Guided Therapy and data structures

• Model Guided Therapy based on patient specific
  modelling requires appropriate IT architectures
  and data structures for its realisation.
• For PSMs, archetypes and templates allow
  different levels of generalisation and
  specialisation, respectively.

6
Model Based Patient Care
Modalities (X-ray,CT, US, MR,SPECT, PET,OI)
Model Creation and Diagnosis (Data fusion, CAD,
)
Model Maintenanceand Intervention (Simulation, de
cision support, validation, )
Biosensors (physiology, metabolism, serum,
tissue, )
IT Communication Infrastructure
7
Content

• Introduction (problems and solutions)
• Model guided therapy with TIMMS
• Classification and model classes
• Virtual human model examples
• PM data structures (SDTM and OpenEHR)
• Conclusion

8
Interventional Cockpit/SAS modules
IT Model-Centric World View
Repo- sitory
Engine
Data Exch.
Control
Therapy Imaging and Model Management System
(TIMMS)
Modelling
Simulation
Kernel for WF and KD Management
Visualisation Rep. Manager
Intervention
Validation
IO Imaging and Biosensors
Therapy Imaging and Model Management System
(TIMMS) ICT infrastructure (based on DICOM-X) for
data, image, model and tool communication for
patient model-guided therapy
Models and intervention records
Data and information
9
Model Guided Therapy with TIMMS

• For a therapeutic intervention it is assumed that
  human, mechatronic, radiation or pharmaceutical
  agents interact with the model.
• MGT provides the scientific basis for an
  accurate, transparent and reproducible
  intervention with the potential for validation
  and other services.
• TIMMS is an IT meta architecture allowing for
  interoperability of the agents to facilitate a
  MGT intervention.

10
Model Guided Therapy 

• The basic TIMMS patient model must have the
  following features
• The TIMMS patient model must have components
  which represent the patient as an n-dimensional
  and multiscale (in space and time) data set.
• The TIMMS patient model must facilitate
  interfacing to the surgeon and other operative
  personnel, the TIMMS engines, TIMMS repositories,
  and the IT infrastructure.
• The TIMMS patient model must be capable of
  linking these components, which may be static or
  dynamic, in a meaningful and accurate way.
• For dynamic components, the TIMMS patient model
  must be able to process morphological and
  physiological data and perform the necessary
  mathematical functions to maintain the model in
  an up-to-date state.

11
Model Guided Therapy 

• The TIMMS patient model must be capable of being
  incorporated by the TIMMS executing workflow and
  responding to its changes.
• The TIMMS patient model must be amenable to be
  developed using readily available, standardized
  informatics methodology. Tools may include UML,
  XML, Visio, block diagrams, workflow diagrams,
  MATLAB, Simulink, DICOM (including surgical
  DICOM), Physiome, CDISC SDTM, openEHR and similar
  products and tools.
• The TIMMS patient model must comply to software
  engineering criteria, for example, to open
  standards and service-oriented architectures to
  allow for multi-disciplinary information
  exchange.
• The TIMMS patient model must allow for further
  extensions to incorporate advances in molecular
  medical imaging, genomics, proteomics and
  epigenetics.
• The TIMMS patient model must be amenable to be
  used for clinical trials, predictive modeling,
  personal health records and in the long term
  contribute to a Model Based Medical Evidence
  (EBME) methodology.

12
Interventional Cockpit/SAS modules
IT Model-Centric World View
Repo- sitory
Engine
Data Exch.
Control
Therapy Imaging and Model Management System
(TIMMS)
Modelling
Simulation
Kernel for WF and KD Management
Visualisation Rep. Manager
Intervention
Validation
IO Imaging and Biosensors
Therapy Imaging and Model Management System
(TIMMS) ICT infrastructure (based on DICOM-X) for
data, image, model and tool communication for
patient model-guided therapy
Models and intervention records
Data and information
13
Generic and patient specific n-D modelling tools
Modelling tools

• Geometric modelling
• Prosthesis modelling
• Properties of cells and tissue
• Segmentation and reconstruction
• Biomechanics and damage
• Tissue growth
• Tissue shift
• Properties of biomaterials
• ...

14
Model Guided Therapy

• MGT in its simpliest instantiation is an
  intervention with a subset, a single or a set of
  voxels representing locations within the patient
  body. With this view, it is an extension from
  Image (pixel) Guided Therapy (IGT) to model
  (voxel) guided therapy. Examples of model guided
  therapy are
• a) interventions within a subset of a voxel,
  e.g. cells, organelles, molecules,
  etc.
• b) interventions with a voxel, e.g. small
  tissue parts of an organ or lesion,
  etc.
• c) interventions with a set of voxels, e.g.
  part of functional structures of organs,
  organ components, soft tissue, lesions,
  etc.

15
Model Guided Therapy 
In a simple PSM, voxels may be associated with
several dimensions of data

1. 1-D signals (e.g. EEG)
2. 2-D projection and tomographic images
3. 3-D reconstructions
4. Temporal change
5. Tissue/cell type
6. Ownership to organ, lesion, system, prothesis,
   chronic condition, etc.
7. Spatial occupancy/extension
8. Permeability (blood brain barrier)
9. Flow (e.g. electric, heat, liquid, perfusion,
   diffusion, etc.)

16
Model Guided Therapy 
In a simple PSM, voxels may be associated with
several dimensions of data

• Level of oxygenation (e.g. level of hypoxia)
• Pharmacokinetics (e.g. effect of tissue on
  pharmaceutical agent, flow parameters, time to
  peak, etc.)
• Pharmacodynamics (effect of pharmaceutical agent
  on tissue, ablation parameters)
• Biological marker types (in vitro and/or in vivo
  molecular spectrum)
• Reference coordinate system (e.g.
  Schaltenbrand/Warren, Talaraich/Tourneaux)
• Value (life critical to life threatening)
• Neighbourhood (e.g. 3³, 5³, 7³, etc.)
• ...

17
Example ENT model elements
Source G. Strauss
18
Example ENT model elements
Source G. Strauss
19
Content

• Introduction (problems and solutions)
• Model guided therapy with TIMMS
• Classification and model classes
• Virtual human model examples
• Conclusion

20
Strategies for multiscale modelling

• Modelling is essential for understanding the
  knowledge of human characteristics such as,
  anatomy, physiology, metabolism, genomics,
  proteomics, pharmacokinetics, etc.
• Because of the complexity of integrating the
  knowledge about the different characteristics the
  model of a human has to be realised on different
  levels (multiscale in space and time) and with
  different ontologies, depending on the questions
  posed and answered delivered.
• The problems associated with using reduced-form
  components within large systems models stem
  primarily from their limited range of validity.

21
Source J. Bassingthwaighte
22
Patient specific and associated modelling
functions
In the Model-Centric World View a wide variety of
information, relating to the patient, can be
integrated with the images and their derivatives,
providing a more comprehensive and robust view of
the patient. By default, the broader the
spectrum of different types of interventional/surg
ical workflows which have to be considered, the
more effort has to be given for designing
appropriate multiscale PSMs and associated
services.
23
Patient specific and associated modelling
functions
Management of n-D and multi resolutional
knowledge (model of the biologic continuum in
space and time) is still a research and
development challenge. If solved successfully,
it will transform surgery into a more
scientifically based activity.
24
Content

• Introduction (problems and solutions)
• Model guided therapy with TIMMS
• Classification and model classes
• Virtual human model examples
• Conclusion

25
Patient Specific CMB
Human Laser Scan (CAESAR DB)
Multimodal Imaging(MRI, CT, Angio,..DT-MRI)
Visible Human Anatomical Templateorgan surface
meshes
Roberts JHU
Spitzer 2006 Virtual Anatomy
PKPD
FEM Mesh (Roberts JHU)
26
Content

• Introduction (problems and solutions)
• Model guided therapy with TIMMS
• Classification and model classes
• Virtual human model examples
• Conclusion

27
Solutions and Research Focus(medical)

• Transition from image guided to model guided
  therapy (e.g. through workflow and use case
  selection/creation/repositories)
• Concepts and specification of patient specific
  models in a multiscale domain of discourse
• Concepts and design of a canonical set of low
  level surgical functions
• Prototyping

28
Interventional Cockpit/SAS modules
IT Model-Centric World View
Repo- sitory
Engine
Data Exch.
Control
Therapy Imaging and Model Management System
(TIMMS)
Prototyping
Modelling
Simulation
Kernel for WF and KD Management
Visualisation Rep. Manager
Intervention
Validation
IO Imaging and Biosensors
Therapy Imaging and Model Management System
(TIMMS) ICT infrastructure (based on DICOM-X) for
data, image, model and tool communication for
patient model-guided therapy
Models and intervention records
Data and information
29
Solutions and Research Focus(technical)

• Concepts and data structure design of patient
  specific models (e.g. with archetypes and
  templates)
• Model management with open architectures (e.g.
  SOA)
• SOA modulariation with repositories, engines,
  LLMs and HLMs
• LLMs as adaptive (cognitive/intelligent) agents
• HLMs as application modules (competitive
  differentiation)
• LLMs possibly as open source
• Kernel (engine and repository) for adaptive
  workflow and KD management
• Cooperative and competitive RD framework for
  engine and repository building
• Therapy based open standard ( e.g. S-DICOM)
• Transition from CAD to CAT modelling

30
Interventional Cockpit/SAS modules
IT Model-Centric World View
Repo- sitory
Engine
Data Exch.
Control
Therapy Imaging and Model Management System
(TIMMS)
Archetypes and Templates
Modelling
Simulation
Kernel for WF and KD Management
Visualisation Rep. Manager
Intervention
Validation
IO Imaging and Biosensors
Therapy Imaging and Model Management System
(TIMMS) ICT infrastructure (based on DICOM-X) for
data, image, model and tool communication for
patient model-guided therapy
Models and intervention records
Data and information
31
Solutions and Research Focus(medical and
technical)

• Transition from image guided to model guided
  therapy (e.g. through workflow and use case
  selection/creation/repositories)
• Use cases for adaptive workflow, exception
  handling and KD management for selected
  interventions
• Cooperative and competitive RD framework for low
  (open source) and high level (competitive
  differentiation) surgical function
  computerisation
• Information/model flow from diagnosis (e.g. CAD)
  to CAT (i.e. interdisciplinary cooperation)
• Development of standards for patient modelling in
  WG24 DICOM in Surgery

32
Interventional Cockpit/SAS modules
IT Model-Centric World View
Repo- sitory
Engine
Data Exch.
Control
Candidate components for open source
Open Source
Modelling
Simulation
Kernel for WF and KD Management
Visualisation Rep. Manager
Intervention
Validation
IO Imaging and Biosensors
Therapy Imaging and Model Management System
(TIMMS) ICT infrastructure (based on DICOM-X) for
data, image, model and tool communication for
patient model-guided therapy
Models and intervention records
Data and information
33
WG 24 DICOM in Surgery Project Groups

• PG1 WF/MI Neurosurgery
• PG2 WF/MI ENT and CMF Surgery
• PG3 WF/MI Orthopaedic Surgery
• PG4 WF/MI Cardiovascular Surgery
• PG5 WF/MI Thoraco-abdominal Surgery
• PG6 WF/MI Interventional Radiology
• PG7 WF/MI Anaesthesia
• PG8 S-PACS Functions
• PG9 WFMS Tools
• PG10 Image Processing and Display
• PG11 Ultrasound in Surgery

34
Definition of Surgical Workflows (S-WFs)

• Micro Laryngeal Surgery (MLS) (PG2 ENT/CMF)
• Foreign Body Excision (PG2 ENT/CMF)
• Total Hip Replacement Surgery (PG3 Orthopaedic)
• Total Endoscopic Coronary Artery Bypass (TECAB)
  (PG4 Cardiovascular)
• Mitral Valve Reconstruction (MVR) (PG4
  Cardiovascular)
• Laparoscopic Splenectomy (PG5
  Thoraco-abdominal)
• Laparoscopic Cholecystectomy (PG5
  Thoraco-abdominal)
• Laparoscopic Nephrectomy left (PG5
  Thoraco-abdominal)
• Angiography with PTA and Stent (PG6
  Interventional Radiology)
• Hepatic Tumor Radio Frequency Ablation (PG6
  Interventional Radiology)
• Trajugular Intrahepatic Portosystemic Shunt (PG6
  Interventional Radiology)

35
CARS 2008 Computer Assisted Radiology and Surgery
CARS / SPIE / EuroPACS 9th Joint Workshop
onSurgical PACS and the Digital Operating
RoomBarcelona, 28 June, 2008

12th Meeting of the DICOM Working Group WG 24
DICOM in Surgery Barcelona, 28 June 2008
http//www.cars-int.org
36
(No Transcript)
37
WG24 DICOM in Surgery

• Secretariat Howard Clark, NEMA
• Secretary Franziska Schweikert,
  CARS/CURAC Office
  fschweikert_at_cars-int.org
  
• General Chair Heinz U. Lemke, ISCAS/CURAC,
  Germany
• Co-Chair Ferenc Jolesz, Harvard Medical
  School, Boston(Surgery/Radiology)
• Co-Chair tbd
• (Industry)