A Dynamic Data Driven Grid System for Intraoperative Image Guided Neurosurgery

1
A Dynamic Data Driven Grid System for
Intra-operative Image Guided Neurosurgery

• A Majumdar1, A Birnbaum1, D Choi1, A Trivedi2,
  S. K. Warfield3, K. Baldridge1, and Petr Krysl2
• 1 San Diego Supercomputer Center University of
  California San Diego
• 2 Structural Engineering Dept University of
  California San Diego
• 3 Computational Radiology Lab Brigham and
  Womens HospitalHarvard Medical School
• Grants NSF ITR 0427183,0426558 NIHP41
  RR13218, P01 CA67165, LM0078651, I3 grant (IBM)

2
TALK SECTIONS

• PROBLEM DESCRIPTION AND DDDAS
• GRID ARCHITECTURE
• ADVANCED BIOMECHANICAL MODEL
• PARALLEL AND END-to-END TIMING
• SUMMARY

3
1. PROBLEM DESCRIPTION AND DDDAS

4
Neurosurgery Challenge

• Challenges
• Remove as much tumor tissue as possible
• Minimize the removal of healthy tissue
• Avoid the disruption of critical anatomical
  structures
• Know when to stop the resection process
• Compounded by the intra-operative brain shape
  deformation that happens as a result of the
  surgical process preoperative plan diminishes
• Important to be able to quantify and correct for
  these deformations while surgery is in progress
  by dynamically updating pre-operative images in a
  way that allows surgeons to react to these
  changing conditions
• The simulation pipeline must meet the real-time
  constraints of neurosurgery provide images
  approx. once/hour within few minutes during
  surgery lasting 6 to 8 hours

5
Intraoperative MRI Scanner at BWH
6
Brain Shape Deformation

Before surgery
After surgery
7
Overall Process

• Before image guided neurosurgery
• During image guided neurosurgery

8
Timing During Surgery

Time (min)
0
20
10
30
40
Before surgery
During surgery
Preop segmentation
Intraop MRI
Segmentation
Registration
Surface displacement
Biomechanical simulation
Visualization
Surgical progress
9
Current Prototype DDDAS Inside Hospital
10
Current Prototype DDDAS System

• Receives 3-D MRI from operating room once/hour or
  so
• Uses displacement of known surface points as BC
  to solve a crude linear elastic biomechanical FEM
  material model on compute system located at BWH
• This crude inaccurate model is solvable within
  the time constraint of few minutes once an hour
  on local computers at BWH
• Dynamically updates pre-op images with
  biomechanical volumetric simulation based
  intra-op images
• Time critical updates shown to surgeons for
  intra-op surgical navigation

11
Two Research Aspects

• Grid Architecture grid scheduling, on demand
  remote access to multi-teraflop machines, data
  transfer
• Data transfer from BWH to SDSC, solution of
  detail advanced biomechanical model, transfer of
  results back to BWH for visualization need to be
  performed in a few minutes
• Development of detailed advanced non-linear
  scalable viscoelastic biomechanical model
• To capture detail intraoperative brain deformation

12
Example of visualization Intra-op Brain Tumor
with Pre-op fMRI
13
2. GRID ARCHITECTURE

14
Queue Delay Experiment on TeraGrid Cluster

• TeraGrid is a NSF funded grid infrastructure
  across multiple research and academic sites
• Queue delays at SDSC and NCSA TG were measured
  over 3 days for 5 mins wall clock time on 2 to 64
  CPUs
• Single job submitted at a time
• If job didnt start within 10 mins, job
  terminated, next one processed
• What is the likelihood of job running
• 313 jobs to NCSA TG cluster and 332 to SDSC TG
  cluster 50 to 56 jobs of each size on each
  cluster

15
of submitted tasks that run, as a fn of CPUs
requested

16
Average queue delay for tasks that began running
within10 mins

17
Queue Delay Test Conclusion

• There appears to be a direct relationship between
  the size of request and the length of the queue
  delay
• Two clusters exhibit different performance
  profiles
• This behavior of queue systems clearly merits
  further study
• More rigorous statistical characterization ongoig
  on much larger data sets

18
Data Transfer

• We are investigating grid based data transfer
  mechanisms such as globus-url-copy, SRB
• All hospitals have firewalls for security and
  patient data privacy single port of entry to
  internal machines

Transfer time in seconds for 20 MB file
19
3. ADVANCED BIOMECHANICAL MODEL

20
Mesh Model with Brain Segmentation
21
Current and New Biomechanical Model

• Current linear elastic material model RTBM
• Advanced model under development - FAMULS
• Advanced model is based on conforming adaptive
  refinement method FAMULS package (AMR)
• Inspired by the theory of wavelets this
  refinement produces globally compatible meshes by
  construction
• First task is to replicate the linear elastic
  result produced by the RTBM code using FAMULS

22
FEM Mesh FAMULS RTBM

RTBM (Uniform)
FAMULS (AMR)
23
Deformation Simulation After Cut

No AMR FAMULS
RTBM
3 level AMR FAMULS
24
Advanced Biomechanical Model

• The current solver is based on small strain
  isotropic elastic principle
• The new biomechanical model will be inhomogeneous
  scalable non-linear viscoelastic model with AMR
• We also want to increase resolution close to the
  level of MRI voxels i.e. millions of FEM meshes
• Since this complex model still has to meet the
  real time constraint of neurosurgery it requires
  fast access to remote multi-tflop systems

25
4. PARALLEL AND END-to-END TIMING

26
Parallel Registration Performance
27
Parallel Rendering Performance
28
Parallel RTBM Performance
(43584 meshes, 214035 tetrahedral elements)

60.00
50.00
IBM Power3
40.00
Elapsed Time (sec)
30.00
IA64 TeraGrid
20.00
IBM Power4
10.00
-
1
2
4
8
16
32
of CPUs
29
End to End (BWH ? SDSC?BWH) Timing

• RTBM not during surgery
• Rendering - during Surgery

30

End-to-end Timing of RTBM

• Timing of transferring 20 MB files from BWH to
  SDSC, running simulations on 16 nodes (32 procs),
  transferring files back to BWH 9 (60
  7) 50 124 sec.
• This shows that the grid infrastructure can
  provide biomechanical brain deformation
  simulation solutions (using the linear elastic
  model) to surgery rooms at BWH within 2 mins
  using TG machines
• This satisfies the tight time constraint set by
  the neurosurgeons

31
End-to-end Timing of Rendering

• MRI data from BWH was transferred to SDSC during
  a surgery
• Parallel rendering was performed at SDSC
• Rendered viz was sent back to BWH (but not shown
  to surgeons)
• Total time (for two sets of data) in sec 253
  (BWH to SDSC) 2 7.4 (render on 32 procs) 0.2
  (overlapping viz) 13.7 (SDSC to BWH) 148.4
  sec

32
5. SUMMARY

33
Ongoing and Future DDDAS Research

• Continuing research and development in grid
  architecture, on demand computing, data transfer
• Continuing development of advanced biomechanical
  model and parallel algorithm
• Moving towards near-continuous DDDAS instead of
  once an hour or so 3-D MRI based DDDAS
• Scanner at BWH can provide one 2-D slice every 3
  sec or three orthogonal 2-D slices every 6 sec
• Near-continuous DDDAS architecture
• Requires major research, development and
  implementation work in the biomechanical
  application domain
• Requires research in the closed loop system of
  dynamic image driven continuous biomechanical
  simulation and 3-D volumetric FEM results based
  surgical navigation and steering