-1- ctsa at rsna 2009 measuring volume change in tumors jeffrey yap, phd ron kikinis, md wendy...

-1-CTSA at RSNA 2009

Measuring Volume

Change in Tumors

Jeffrey Yap, PhDRon Kikinis, MD

Wendy Plesniak PhDKilian M Pohl, PhD

Ender Konugolu, PhDAndriy Fedorov, PhD

Slicer3 Training Compendium


Data

This course is built upon two scans of a patient with meningioma:

MR Scan 1 MR Scan 2Please note: we have pre-adjusted the window and level settings for these volumes so that they are appropriate for display on most laptops. If display is not satisfactory on your machine or projector, the Volumes Module may be used to refine these settings.


DataClinical context

Baseline radiologist’s impression:•large falcine lesion is identified. •measures 3.1 cm anteroposteriorly, 3.05 cm from side-to-side, 3.5 cm in height. •enhances moderately on post gadolinium imaging.

Follow-up radiologist’s impression: •left frontal lobe mass appears unchanged on all series. •measures 3.3 x 3.2 cm in maximum dimension. •enhances moderately on post gadolinium imaging.

How has the tumor changed?

Baseline: June 2006 Follow-up: June 2007


Overview

Part I: Applying RECIST methodology

Part II: Measuring volume change in tumors using ChangeTracker


Part IStandard Response Evaluation Criteria in Solid Tumors (RECIST) using 3D Slicer

Jeffrey Yap, PhDRon Kikinis, MDWendy Plesniak, PhD


Response Evaluation Criteria in Solid Tumors (RECIST)

Methodologies:

WHO: Largest diameter and its perpendicularRECIST: Largest diameter (LD) only.

Assessment criteria:

•CR (complete response) = disappearance of all target lesions•PR (partial response) = 30% decrease in the sum of the longest diameter of target lesions•PD (progressive disease) = 20% increase in the sum of the longest diameter of target lesions•SD (stable disease) = small changes that do not meet above criteria


Learning objective

Following Part I of this tutorial, you’ll be able to use 3D Slicer to:

• Load a MRML Scene file,

• Visually assess the response to therapy,

• Make quantitative linear measurements across tumor cross sections, and

• Make quantitative volume measurements of tumor anatomy

• Assess according to RECIST criteria.


The Graphical User Interface (GUI) of Slicer3 integrates five components:

• the Menu Toolbar

•the Module GUI Panel

•the 3D Viewer

•the Slice Viewer

•the Slice and 3D View Controller

Slice Viewer

3DViewerModule GUI Panel

Slice and 3D View

Controller

Menu Toolbar

Quick overview of Slicer3 GUI


Select File->Close Scene:

Close any previous scene

This removes any datasets previously loaded into Slicer.


Loading data using a MRML scene file

Using File->Load Scene:

This raises the Load Scene Interface

• select the file RECIST-Simple.mrml

• double click the file, or click Open


Loading data: quick look at both image volumes

The image volume 2006-spgr should be displayed in the background BG layer.

The image volume 2007-spgr should be displayed in the foreground FG layer.

Use the FG/BG toggle button and the Fade Slider to see both volumes.

Toggle BGonly Fade FGonly


Explore the datasets

Scroll in any Slice Viewer to look for large tumor cross sections

Zoom in the Slice Viewers by Right-Clicking and dragging forward or back with the mouse.

Pan in the Slice Viewers by Middle-Clicking and dragging with the mouse.

Left-Click-and-drag in the 3D Viewer will rotate the view


Using Modules Menubutton: Expose the menu and select the Fiducials Module.

Side Note: The Fiducials Module is powerful and functionality-rich. Learn more about it here: http://www.slicer.org/slicerWiki/index.php/Modules:Fiducials-Documentation-3.4

Loading data: quick look at both image volumes


Using Fiducials for linear measurementPick Place Transform

Notice the Mouse Mode Toolbar for specifying the Mode of Interation

Slicer is currently in “Transform Mode”


Using Fiducials for linear measurement

Select Place Mode, to Place Fiducial Points

NOTE: In Place Mode

any Left-mouse-button click in a viewer will create a new fiducial point at the spatial location within the volumetric frame of reference.



Scroll to the –5.1565 mm position in the Sagittal Slice Viewer

Click and Drag the Right Mouse Button to zoom in.

Click and Drag the Middle Mouse button to center the tumor.



Click with the left mouse button to place two fiducial points along the long diameter.

THEN

Change back to Transform Mode



Note the magnified region around the cursor in the bottom-most GUI panel.

Move the mouse over either fiducial point.

When it turns yellow, you can move it by clicking and dragging with the left mouse button.


The distance measurement (mm) will be updated in the Fiducial’s GUI panel.



Toggle between foreground and background to observe the corresponding slice

and adjust the measurement for the follow-up study.



Compute tumor volume using segmentation

Link all Slice Viewers using the Link Toggle if they are unlinked

linked

Unlinked

Select the 2006-spgr-label Tumor Mask in the LB layer

Select the None volume in the FG layer

The tumor segmentation for the baseline volume is displayed


Using Modules Menubutton:

Expose the menu and select the LabelStatistics module under the Statistics group.

Compute tumor volume


Compute tumor volume

Specify 2006-spgr as the Input Greyscale Volume

Specify 2006-spgr-label as the Input Labelmap

Click Apply.

The tumor volume for label 6 (green) is shown in mm3


Assessment

The same analysis may be done for the follow-up study:

Specify 2007-spgr as the Input Greyscale Volume

Specify 2007-spgr-label as the Input Labelmap

And click Apply.

Final RECIST tumor metrics:

2007-spgr

2006-spgr

V mm3CD (mm)LD (mm)

17458.24

17432.05

31.9736.67

32.2437.72

% change: -2.78% -0.84% 0.15%


RECIST methodology:

Is an effective approach for measuring sizeable changes in tumor size and assessing tumor response to therapy.

We are developing new analysis tools in 3D Slicer for assessing change over time, including when the changes are small.

Summary


3D Slicer’s LabelDiameterEstimation module (available currently only in the Slicer nightly builds), will automatically compute the largest tumor diameter given an initial segmentation.

Baseline: June 2006 Follow-up: June 2007

New tools


3D Slicer’s Measurements module, is developing a measurement widget that operates in the 3D Viewer and the Slice Viewers. (the prototype version is available in Slicer’s nightly builds).

New tools


3D Slicer’s ChangeTracker module can assess volumetric changes at the voxel-level.

New tools


Part II

The module described in this tutorial was tested on Axial 3D SPGR T1 post Gadolinium scans (Voxel dimension: 0.94mm x 0.94mm x 1.20mm, FOV: 240mm, Matrix: 256 x 256)

Analyzing Small Volumetric Changes using the ChangeTracker Module

Kilian M Pohl, PhDEnder Konugolu, PhD Andriy Fedorov, PhD


Select File->Close Scene:

Close any previous scene

This removes any datasets previously loaded into Slicer.


Loading training dataset

Select File->Load Scene:

This raises the Load Scene Interface

• select the file: ChangetrackerTutorial2009.mrml

• double click the file, or click Open


Using the Layout MenuButton

Change Layout to Four-up

Select Four-up layout.


Select ChangeTracker Module

Using the Modules Menubutton,

Select the ChangeTracker Module from the Wizards category.


Workflow Wizard

The Workflow Wizard guides the user through a sequence of steps and has the following components:

•the Step Panel

•the User Panel

•the Navigation Panel

Step Panel

User Panel

Navigation Panel


Select Scans

Select: Scan 1 = 2006-spgr

Scan 2 = 2007-spgr


Zoom in Move sliders to see tumor

Zoom in (Right mouse down and push/pull)

Press Next


Identify Volume of Interest

A VOI Box Widget is positioned within the volume, and its blue extent is displayed in each Slice Viewer

Press Display volume rendering and VOI label

Right mouse click in the tumor center.


VOI box widget and volume rendering of tumor in yellow


Right mouse click in any Slice Viewer to adjust VOI center

Left mouse click in any Slice Viewer to change the VOI extent


VOI Widget range sliders are color-coded to match VOI box Widget handles in 3D Viewer

Fine-tune the VOI using the VOI Widget range sliders or by moving the VOI Widget handles in 3D view




Scroll through slices to ensure that tumor boundaries are included in the VOI

Click Next


Segment Tumor

Slice Viewers show VOI with current segmentation overlay in light yellow

3D Viewer shows model of tumor

Change Segmentation by moving range slider


Press Grid to view grid with original voxel size

(You may need to pan/zoom in slice viewers to refresh their display)

Segment Tumor


Press Slice to see the first scan in the 3D Viewer

Note: The 3D View can be reset using the Axis Navigator Widget

Segment Tumor


Press Next

Segment Tumor


Choose Metric Type

Choice of Metric:

• Detect change by analyzing intensity pattern (fast)

• Measure change by analyzing deformation map (slow)

Select fast and press Analyze


Analysis I: Volume Preserving Registration

Before Afterwards

Konukoglu et al. ,“Monitoring Slowly Evolving Tumors”, ISBI, 2008


Analysis II: Normalize Intensities

Scan 1 Scan 2 Scan 2 - Norm



)(absxI

PDF of Dormant Tissue

Determine probability density function

Analysis III: Detect Change – Intensity

Based

- =

Compute statistics of static tissue:

Subtract Scan2 from Scan1

Define region of static tissue


Analysis III: Detect Change – Intensity Based

Mode : Sensitive Moderate Robust False Positive Risk: High Medium LowGrowth (mm3) : 1301 547 100Growth (voxel): 1057 444 81

Ana

lysi

sD

ata



Analysis III: Detect Change–Deformation Map

=

Determine change via mapping:

Compute diffeomorphic mapping between scans

Compute Jacobian

Apply mapping to segmentation

or

Mode : Jaccobian Growth (mm3) : 1291 Growth (voxels): 1049

Mode : Segmentation Growth (mm3) : 1386 Growth (voxels): 1126


Show Change in Pathology

Change in volume is shown: magenta = growthgreen = shrinkage

Select visibility icon to show slice in the 3D Viewer



Lightbox View shows five consecutive slices for the VOI in the first scan (top), and in the second scan (bottom)



Define sensitivity of analysis



Click on the colored bar at the top of any Slice Viewer to show or hide the controls.

This allows more display space for viewing the image data.


Appropriate use

• Tested on Axial 3D SPGR T1 post Gadolinium scans (Voxel dimension: 0.94mm x 0.94mm x 1.20mm, FOV: 240mm, Matrix: 256 x 256)

• Tumor boundary should be clear

• Only for contrast enhanced images

• Need homogenous enhancement across timepoints

• Not tested for tumors with necrosis in one timepoint


Acknowledgments

Harvard Clinical and Translational Science Center

National Alliance for Medical Image ComputingNIH U54EB005149

Brain Science Foundation

INRIA, France

Neuroimage Analysis Center (NAC)

National Center for Image-Guided Therapy (NCIGT)

Surgical Planning Laboratory, Brigham and Women’s Hospital

-1- ctsa at rsna 2009 measuring volume change in tumors jeffrey yap, phd ron kikinis, md wendy...

Documents