Pohl K, Konukoglu E

-1-National Alliance for Medical Image Computing

Measuring Volume Change in Tumors

Kilian M Pohl, PhDEnder Konugolu

Slicer3 Training Compendium

Pohl K, Konukoglu E

-2-National Alliance for Medical Image Computing

Material

This course requires the following installation:

•The current version of Slicer 3 Software, which can be installed from http://www.na-mic.org/Slicer/Download/Snapshots/<OS>/Slicer3-3.3-alpha-2008-09-11-*

•The TumorGrowth-Tutorial-Data.xcat, which can be downloaded from http://www.na-mic.org/Wiki/index.php/Slicer3:Training

•A reliable internet connection for downloading the data

Disclaimer

It is the responsibility of the user of 3DSlicer to comply with both the terms of the license and with the applicable laws, regulations and rules.

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)

Pohl K, Konukoglu E

-3-National Alliance for Medical Image Computing

Data

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

MR Scan 1 MR Scan 2

Pohl K, Konukoglu E

-4-National Alliance for Medical Image Computing

Learning objective

Following this tutorial, you’ll be able to load two scans into Slicer3, and measure and visualize the change in volume of the meningioma between the two scans.

Pohl K, Konukoglu E

-5-National Alliance for Medical Image Computing

Overview

• Loading a scene

• Measuring volume change in tumors

Pohl K, Konukoglu E

-6-National Alliance for Medical Image Computing

Part 1: Loading the tutorial data

via a scene file

Pohl K, Konukoglu E

-7-National Alliance for Medical Image Computing

Slicer3 GUI

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

Pohl K, Konukoglu E

-8-National Alliance for Medical Image Computing

Loading a SceneSelect FileLoad Scene in the Main Menu.

Load the file ChangeTracker-Tutorial-Data.xcat

Data is downloaded from www.xnat.org

Pohl K, Konukoglu E

-9-National Alliance for Medical Image Computing

Loading a Scene

The MRML Tree in the Data module shows the two scans (2006-spgr and 2007-spgr) in this scene.

Pohl K, Konukoglu E

-10-National Alliance for Medical Image Computing

Modifying Layout of Viewer

Select and Four-up layout.

Pohl K, Konukoglu E

-11-National Alliance for Medical Image Computing

New Layout of Viewer

Slice Viewers

3D Viewer

Modify size of windows as needed

Pohl K, Konukoglu E

-12-National Alliance for Medical Image Computing

Part 2: Measuring Volume Change

Pohl K, Konukoglu E

-13-National Alliance for Medical Image Computing

New Layout of Viewer

Select Modules

Select ChangeTracker

Pohl K, Konukoglu E

-14-National Alliance for Medical Image Computing

Workflow Wizard

The Workflow Wizard of Slicer3 guides the user through a sequence of steps and is defined by the following components:

•the Step Panel

•the User Panel

•the Navigation Panel

Step Panel

User Panel

Navigation Panel

Pohl K, Konukoglu E

-15-National Alliance for Medical Image Computing

Select Scans

Select 2006-spgr and 2007-spgr

Press Next

Pohl K, Konukoglu E

-16-National Alliance for Medical Image Computing

Identify Volume of Interest

Move sliders to see tumor

Pohl K, Konukoglu E

-17-National Alliance for Medical Image Computing

Identify Volume of Interest

Identify VOI by clicking left mouse button inside tumor

Press Show VOI to view region

Pohl K, Konukoglu E

-18-National Alliance for Medical Image Computing

Identify Volume of Interest

VOI is shown in blue

Lower and Upper bound of VOI

Pohl K, Konukoglu E

-19-National Alliance for Medical Image Computing

Identify Volume of Interest

Move sliders or enter:X: 105 155Y: 76 121Z: 62 90

Press Next

Pohl K, Konukoglu E

-20-National Alliance for Medical Image Computing

Segment Tumor

Slice Viewers show VOI with current Segmentation

3D Viewer shows model of tumor with additional structures

Change Segmentation by moving Slider to 125

Pohl K, Konukoglu E

-21-National Alliance for Medical Image Computing

Segment TumorPress Grid to view grid with original voxel size

Pohl K, Konukoglu E

-22-National Alliance for Medical Image Computing

Segment TumorPress Slice to see the first scan in the 3D Viewer

Press Next

Pohl K, Konukoglu E

-23-National Alliance for Medical Image Computing

Choose Metric TypeChoice of Metric:

• Detect change by analyzing intensity pattern (fast)

• Measure change by analyzing deformation map (slow)

Select fast and press Analyze

Pohl K, Konukoglu E

-24-National Alliance for Medical Image Computing

Analysis I: Volume Preserving Registration

Before Afterwards

Pohl K, Konukoglu E

-25-National Alliance for Medical Image Computing

Analysis II: Normalize Intensities

Scan 1 Scan 2 Scan 2 - Norm

Pohl K, Konukoglu E

-26-National Alliance for Medical Image Computing

)(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

Pohl K, Konukoglu E

-27-National Alliance for Medical Image Computing

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

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

Pohl K, Konukoglu E

-28-National Alliance for Medical Image Computing

Analysis III: Detect Change–Deformation Map

=

Determine change via mapping:

Compute diffeomorphic mapping between scans

Compute Jaccobian

Apply mapping to segmentation

or

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

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

Pohl K, Konukoglu E

-29-National Alliance for Medical Image Computing

Show Change in Pathology

Define sensitivity of analysis

Show change in volume: purple = growthgreen = shrinkage

Pohl K, Konukoglu E

-30-National Alliance for Medical Image Computing

Define Working Directory

Press directory button

Select to create directory

Pohl K, Konukoglu E

-31-National Alliance for Medical Image Computing

Define Working Directory

Enter tutorial and press OK

Select tutorial and press OK

Pohl K, Konukoglu E

-32-National Alliance for Medical Image Computing

Save Screenshot

Press ScreenshotScreenshot is saved to tutorial directory

View TG_Screenshot_001.png

Pohl K, Konukoglu E

-33-National Alliance for Medical Image Computing

Save Analysis Result

Press AnalysisOutcome of analysis is saved to tutorial directory

View AnalysisOutcome.log

Pohl K, Konukoglu E

-34-National Alliance for Medical Image Computing

Save Intermediate Results

Press Data

Press Save All and Save

Pohl K, Konukoglu E

-35-National Alliance for Medical Image Computing

Finish Workflow

Press OK to return to the first step of the workflow

Pohl K, Konukoglu E

-36-National Alliance for Medical Image Computing

Overview

• Loading a scene

• Measuring volume change in tumors

Pohl K, Konukoglu E

-37-National Alliance for Medical Image Computing

Summary

Conclusion:

• Interactive interface to load scene

• Workflow technology to semi-automatically detect change in pathology

In-depth description of technology:

• Konukoglu et al. ,“Monitoring Slowly Evolving Tumors”, IEEE ISBI, p. 812 – 815, 2008 http://www.spl.harvard.edu/pages/Special:PubDB_View?dspaceid=1430

• Pohl et al. , “Automatic Tumor Growth Detection” in Meningiomas A Comprehensive Text, In Press.

Pohl K, Konukoglu E

-38-National Alliance for Medical Image Computing

Information

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

• We expect the tool to work also on set of scans that fulfill the following requirements:– same patient was scanned– scans were acquired with the same

acquisition protocol – scans have isotropic resolution

Pohl K, Konukoglu E

-39-National Alliance for Medical Image Computing

Acknowledgments

National Alliance for Medical Image ComputingNIH U54EB005149

add nac, ncigt and spl acknowledgements

Brain Science Foundation

INRIA, France