2011 Winter Project Week:Breakout Slicer with Ron

 Back to  Project Week Agenda

Moderator: Ron Kikinis

Introduction

• This is an advanced topic presentation on Slicer abilities. The intention is to familiarize the NA-MIC community with Slicer abilities.
• You have mastered the introductions and you are hungry for more.... :)
  • This means that we expect you to at least have worked through the Core Tutorials before this class.
• Ron and the creators of modules will show you advanced capabilities in Slicer 3.6

Logistics

• What: Advanced capabilities in Slicer
• Who: All the new DBP engineers, all the participants new to NA-MIC
• When: Monday afternoon: 3pm through 5pm
• Where: Main Event Room

Data

• Image data used
• 

  Case 40 from the fMRI neurosurgery data set on central.xnat.org

• 

  T1 Weighted SPGR image

• 

  T2 weighted Bzero image derived from the DWI acquisition

• 

  FLAIR image

• 

  SPGR after injection of Gd

• 

  DTI color by orientation overlayed over T2 baseline

• 

  Annotation of some of the structures

Preliminary Program

• 10 minutes or less per segment
• Comply with Rons Rules for Tools

• Registration (Dominik Meier, Hans Johnson)
  • Update: revised dataset and Slicer Scenes for demo here:Media:RegDemo_AHM11.zip‎
  • focusing on DTI only: issues
    • T1 is too high resolution to resample DTI into (will cause memory problems); use subsampled version T1_sub instead.
    • initial misalignment is regional only, BSpline portion is sufficient (no affine required)
    • masking is required for successful registration: BRAINSfit requires two masks: moving mask for the DTI_base we get from the initial DTI_estimation; for the fixed mask T1_sub we use the same one but resampled (T1_sub_mask).
    • BRAINSfit registration will take ca. 90seconds with default settings and parameters as suggested below
    • BRAINSfit parameters: fixed volume: T1_sub, moving volume: DTI_base; check: "include BSpline Registration Phase"; output: Slicer BSpline Transform : create new; output image volume: create new; output image pixel type: short; Registration parameters: number of grid subsdivisions; = 7,5,5 ; Control of mask processing: check ROI; input fixed mask = T1_sub_mask; input moving mask: DTI_maskj
  • DTI resampling is done via ResampleDTI Volume module. Parameters
    • Input Volume: DTI, Output Volume: create new; Reference Volume: "T2-sub"; Transform Node: Bspline Xform produced by BRAINSfit, check output-to-input" box.
  • for the procedures to register everything to the T1 (incl. contrast, FLAIR, T2) , see here: Reg.Lib Case 33
  • Resample as needed (DTI, T2 Baseline):
    • transforms available here: Reg.Lib Case 33
• Compareview (Jim Miller)
• Scenesnapshots (Wendy Plesniak, Alex Yarmarkovich) Sample scene (large) including scene snapshots: fMRICase040.zip
• Simple measurements and fiducials (Nicole Aucoin) Sample scene and data with snapshots for fiducials, seeding, measurements
  • Fiducials tutorial and Data
    • Adding and Deleting Fiducials
    • Editing Fiducials
    • Display properties
    • Linking (control key+mouse move) and jumping slices (right click in table)
    • Passing Fiducials to Command Line modules
    • Fiducial Seeding
  • Measurements
    • Ruler (put down two fiducials, Control-m to make a ruler)
    • Angle
    • Constraining to slices or models
• Interactive editor (Steve Pieper)
  • Editor Documentation
  • Description of individual effects
• Fast segmenters: Fast Marching, RSS, GrowCutSegment (Andrey Fedorov, Yi Gao, Harini Veeraraghavan)
  • Fast Marching demo scene
    • inner part of the tumor segmented from SPGR
    • outer part of the tumor segmented from post-Gad
    • WM and GM segmented from N4-processed SPGR (WM segmentation of the original SPGR volume is included, note under-segmented WM in the skull base)
    • all segmentations were done using FastMarching, fiducials are included for each of the segmentations
    • no fine-tuning of the fiducial locations was done -- this is an example result one can get almost right away
    • rule of thumb in placing fiducials: try to cover uniformly the volume you are trying to segment; this is particularly important for large structures like WM/GM
  • RSS demo scene
    • Segment the inner part of the tumor from SPGR using RSS
• DTI processing (Alex Yarmarkovich, Demian Wasserman)
• Volume Cropping (Andrey Fedorov)
• Keyboard and mouse shortcuts (Steve Pieper)
• Volume Rendering (Yanling Liu/Alex Yarmarkovich/Curtis Lisle)
• EM segmentation (Kilian Pohl)

Attendees

1. Ron Kikinis
2. Steve Pieper
3. UCLA DBP Engineer: Andrei Irimia
4. Iowa DBP Engineer: Mark Scully
5. MGH DBP Engineer: Nadya Shusharina
6. Utah DBP Engineer: Josh Cates
7. Utah DBP Student: Josh Blauer
8. Utah DBP PI: Rob MacLeod
9. Utah DBP Student: Chris Gloschat