Transcript NA-MIC National Alliance for Medical Image Computing http://na-mic.org Skull Stripper Extension Jim Miller GE Research.

NA-MIC
National Alliance for Medical Image Computing
http://na-mic.org
Skull Stripper Extension
Jim Miller
GE Research
Part of a NA-MIC Collaboration Grant
PAR-07-249: R01EB006733 Development
and Dissemination of Robust Brain MRI
Measurement Tools
PI - Dinggang Shen, UNC
NA-MIC versions of
HAMMER
White matter lesion detection for MS
National Alliance for Medical Image Computing
http://na-mic.org
Skull Stripping
• Standard step in many neuro applications
– Removes anatomy not relevent to the analysis
– Improves robustness
• Preliminary results for grant used FSL
– Needed a skull stripper with a NA-MIC
compatible license
National Alliance for Medical Image Computing
http://na-mic.org
Skull Stripper Method
• Similar to BET
• Deformable model
• Fuzzy tissue classification
• Tested on OASIS database
– DICE comparison to BET 0.948
• Xiaodong Tao ; Ming-Ching Chang; A skull stripping method
using deformable surface and tissue classification. Proc. SPIE
7623, Medical Imaging 2010: Image Processing, 76233L
(March 12, 2010); doi:10.1117/12.844061.
National Alliance for Medical Image Computing
http://na-mic.org
Skull Stripper Extension
• Single CLI module
• Not a complete neuro analysis workflow
• Packaged for re-use
National Alliance for Medical Image Computing
http://na-mic.org
Extensions manager
National Alliance for Medical Image Computing
http://na-mic.org
Skull Stripper Module
National Alliance for Medical Image Computing
http://na-mic.org
SlicerRT
3D Slicer extension for image-guided
radiation therapy research
SparKit project overview
• Funding by Cancer Care Ontario till 2016
• PI & co-PI: Gabor Fichtinger (Queen’s), David
Jaffray (Toronto UHN), Terry Peters (Robarts)
• Themes:
– SlicerRT: radiotherapy toolkit for 3D Slicer
– SlicerIGT: Image-guided therapy with 3D Slicer
• SlicerRT team: Csaba Pinter, Andras Lasso
(Queen’s), Kevin Wang (Toronto UHN)
Laboratory for Percutaneous Surgery – Copyright © Queen’s University, 2013
Theme
• Motivation:
– commercial treatment planning
software: closed, expensive
– existing research tools
RT export
(CERR, PLUNC,
(DICOM,
Matlab, …)
dicompyler, etc): limited
• Goal:
– “hub” for RT data analysis
RT-specific
and comparison
analysis
(dose
– Free, open-source, BSD
comparison,
license
DVH, …)
• Approach:
– leverage existing tools and
parallel efforts
RT import
(DICOM)
Visualization
(volumes,
models,
2D/3D, etc.)
3D
Slicer
Analysis
(error metrics,
measurements,
image fusion,
etc.)
Laboratory for Percutaneous Surgery – Copyright © Queen’s University, 2013
Processing
(registration,
segmentation,
etc.)
Features overview
• Import/Export: DICOM-RT
import, Matlab bridgeE
• Contours analysis: multiple
representations (model,
labelmap; auto convert),
editing, contour comparisonE,
grow/shrinkE, combineE
• Dose analysis: dose volume
histogram, accumulation,
comparison (gamma), isodose
contours visualization, proton
dose computationE
• Registration: BSpline
registrationP, Landwarp registrationP
E = Experimental
P = Through Plastimatch
Laboratory for Percutaneous Surgery – Copyright © Queen’s University, 2013
SlicerRT extension for 3D Slicer
• Collection of RT-specific modules, includes
• Distributed as a 3D Slicer extension: can be downloaded,
installed, upgraded using the extension manager in Slicer
SlicerRT extension in the 3D Slicer app store (free)
Laboratory for Percutaneous Surgery – Copyright © Queen’s University, 2013
References
• Overview paper: Csaba Pinter, Andras Lasso, An Wang, David Jaffray,
and Gabor Fichtinger, “SlicerRT: Radiation therapy research toolkit for
3D Slicer”, Med. Phys. 39 (10), October 2012
• Project homepage: https://www.assembla.com/spaces/slicerrt/
• Contact: Andras Lasso ([email protected])
Laboratory for Percutaneous Surgery – Copyright © Queen’s University, 2013
iGyne
3D Slicer extension
for needle detection and labeling
in MR-guided interstitial brachytherapy
for gynecologic cancer
17
Funding Acknowledgement
P41 EB 015898
National Center for Image Guided
Therapy (Jolesz, Tempany)
2005-2015
R03 EB 013792
Segmentation for Gynecologic
Brachytherapy (Kapur)
2011-2013
Slide 18
Investigators
Radiology
Computer Science
MRI Physics
Mechanical Design
Radiation Oncology
Radiation Physics
Slide 19
Clare Tempany, MD
Tina Kapur, PhD
Jan Egger, PhD, PhD
Guillaume Pernelle, MS
Xaiojun Chen, PhD
Ehud Schmidt, PhD
Sam Song, PhD
Tobias Penzkofer, MD
Akila Viswanathan, MD,
MPH
Robert Cormack, PhD
Antonio Damato, PhD
Jorgen Hansen, MS
Yi Gao, PhD
Wei Wang, PhD
Alireza Mehrtash, MS
Gynecologic Cancers
• 500,000 cases per year
worldwide: Cervical,
Uterine, Vaginal, Vulvar,
Ovarian
• 4th leading cause of death
in women in the US
• Treated with chemoradiation, brachytherapy
Slide 20
The Problem: Which needle is which?
21
iGyne Key Features
• Software workflow that matches
clinical workflow
• (robust) DICOM transfer from MR
• Model-to-model registration of
applicator CAD model to image
• Simulation of needle trajectories
• Novel needle detection and labeling
• Reformatting of MRI along needle
trajectory
22
Timeline: Concept to Extension 1 Year
2011
November:
2012 February:
• MeVis Lab
Prototype
(Jan Egger)
• Slicer 3 C++
Extension
Module
(Xiaojun
Chen)
2012
November:
• Slicer4
Python
Extension
Module
(Guillaume
Pernelle)
New Lines of Code Written for iGyne: 5000
23
iGyne detects and labels needles from
MR images
24
Results: ~3mm Hausdorff distance
Computation Time
50s for applicator registration
~1s per needle to segment
25
PerkNav
3D Slicer Extension for Navigation
during Percutaneous Interventions
Overview
• Funding: Cancer Care Ontario, part of SparKit
• PI: Gabor Fichtinger (Queen’s University)
• Team: Tamas Ungi, Andras Lasso, Adam Rankin
28
PerkNav Features
• Collect Fiducials – landmark registration
• Create Models – simple shapes to represent objects
• Ultrasound Snapshots – for intervention navigation
• Transform Recorder – save tool trajectories
• Perk Evaluator – motion economy metrics
• Image Overlay
29
Example applications
Training
Orthopedic surgery
Prostate brachytherapy
Urology
Anesthesia
30
References
• Current webpage: www.assembla.com/spaces/slicerigt
• Selected publications:
– Moult et al. Ultrasound-Guided Facet Joint Injection Training Using
Perk Tutor. IJCARS, in press.
– Ungi et al. Perk Tutor: A configurable, open-source training platform
for ultrasound-guided needle insertions. IEEE TBME, 2012.
– Fritz et al. Augmented Reality Visualization with Use of Image Overlay
Technology for MR Imaging-guided Interventions: Assessment of
Performance in Cadaveric Shoulder and Hip Arthrography at 1.5 T.
Radiology, 2012.
– Ungi et al. Spinal Needle Navigation by Tracked Ultrasound Snapshots.
IEEE TBME, 2012.
31
Reporting
and
Longitudinal PET/CT Analysis
3D Slicer extensions to support
imaging biomarker discovery
Funding
• U01 CA151261:
Quantitative Imaging
Network,
Multiparametric MRI for
Prostate Cancer imaging
(PI Fiona Fennessy,
Brigham and Women’s
Hospital)
Team
• Reporting:
– Andrey Fedorov, Nicole Aucoin, Steve Pieper
• Longitudinal PET/CT:
– Paul Mercea, Andrey Fedorov, Steve Pieper, Ron
Kikinis
– U. Iowa: Reinhard Beichel, Markus Van Tol, John
Buatti
• J.-C. Fillion-Robin & 3D Slicer community
Motivation
• QIN: quantitative imaging as a biomarker of cancer
treatment response
• Pressing need for enabling tools for imaging
biomarker development, validation and
dissemination
• Self-contained image annotation
and markup
• Support of interoperable formats
(DICOM, AIM)
• Support of SNOMED Clinical
Terminology for Slicer color maps
• Result:
– Improved integration with clinical
systems and PACS
– Standards-based formats for sharing
analysis results and collaboration
Reporting: Towards interoperability
AIM:
Markup
(points, bi-dimensional
measurements)
DICOM SEG:
Volumetric
segmentation
results
• Clinical researcher view:
– Research software for SUV
quantification from longitudinal PET
studies
• Biomedical engineer view:
– Platform for development and
dissemination of segmentation,
quantification, analysis tools
Adoption by QIN Community
• Reporting:
– MGH and
Stanford
• Longitudinal
PET/CT analysis:
– U. Iowa
Longitudinal PET/CT Timeline
• Lead developer: Paul Mercea, MS student
– Requirements analysis, prototyping: 2 months
– Code development: 3 months
– Expert evaluation, refinement: 1 month
– Lines of code: > 12,000