Scribd
Upload
11 views55 pages

ImagePhenotypingTutorial SoniaPujol

This document provides an introduction to image phenotyping using the 3D Slicer platform. It describes how imaging phenotypes can describe disease features detected through medical imaging and statistical analysis. The tutorial then demonstrates how to load a meningioma dataset into 3D Slicer, segment the tumor manually using the Segment Editor module, and calculate the tumor volume using the Segment Statistics module. The overall goal is to introduce basic image phenotyping techniques for tumor characterization.

Uploaded by

t
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
11 views55 pages

ImagePhenotypingTutorial SoniaPujol

This document provides an introduction to image phenotyping using the 3D Slicer platform. It describes how imaging phenotypes can describe disease features detected through medical imaging and statistical analysis. The tutorial then demonstrates how to load a meningioma dataset into 3D Slicer, segment the tumor manually using the Segment Editor module, and calculate the tumor volume using the Segment Statistics module. The overall goal is to introduce basic image phenotyping techniques for tumor characterization.

Uploaded by

t
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 55

An Introduc+on to Image

Phenotyping

Sonia Pujol, Ph.D.

Director of Training and Educa+on, 3D Slicer


Assistant Professor of Radiology
Brigham and Women’s Hospital
Harvard Medical School
Imaging Phenotypes
• Imaging phenotypes describe features of
disease that can be detected through medical
imaging combined with feature detec+on,
machine learning and sta+s+cal analysis, and
correlated with other indicators of disease.

Gillies et al. Radiology 2015


Sonia Pujol, PhD - All Rights Reserved 2011-2019
Clinical Case: Meningioma
• Meningiomas are slow-growing
extra axial brain tumor arising
from arachnoidal cells
• Treatment op+ons include
observa+on, surgery and
radia+on therapy
• Image-based predictors of tumor
grades have the poten+al to
enhance clinical decision-making

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Overall Goal
This tutorial is an introduc+on
to the basics of image
phenotyping for tumor
characteriza+on using the 3D
Slicer plaVorm.

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Imaging Features
Quan+ta+ve features derived
from imaging data have the
poten+al to provide clinically
relevant informa+on for
predic+ng tumor grade and
evalua+ng response treatment

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Tutorial materials

• 3D Slicer release version 4.10.1

• Slicer Radiomics Extension

• Meningioma dataset

Sonia Pujol, PhD - All Rights Reserved 2011-2019


3D Slicer installa+on
• To install and start the 3D Slicer soZware on your
computer, follow the instruc+ons of the Quick Start
Guide tutorial available at
h[ps://www.slicer.org/wiki/Documenta+on/4.10/
Training

Sonia Pujol, PhD - All Rights Reserved 2011-2019


Disclaimer
• 3D Slicer is a free open source soZware for medical
image compu+ng research distributed under a BDS
style license.

• The soZware is not FDA approved or CE-Marked, and


is for research use only.

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Workflow Overiew

Step 1: Data Step 2: Tumor Step 3: Tumor Step 4: Imaging Features


Loading Segmenta+on Volume Calcula+on Extrac+on

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Part 1: Data
Loading and tumor
diameter
measurements

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Data Loading

Drag and drop the file Meningioma.nrrd


into the Slicer

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Data Loading

Click on OK to load
the files in Slicer

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Data Loading

Slicer displays the axial, sagi[al and coronal


images of the meningioma dataset.

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Data Loading

Select the Red Slicer Only


Layout from the Layout Menu
Sonia Pujol, PhD - All Rights Reserved 2019-2020
Data Loading

Select the Ruler tool in


the main menu

Find the slice that you


want to use to
measure the largest
diameter
Posi+on two fiducials
using the Ruler tool to
measure the largest
diameter on an axial
image

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Data Loading

Slicer displays the


value of the tumor
d i a m e t e r a s
measured on an axial
slice

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Part 2: Tumor
Segmenta+on

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Image Segmenta+on
• Image Segmenta+on is the process of
iden+fying the contours of structures of
interest in imaging data

• Image Segmenta+on can be performed by


manual contouring or by using automated
segmenta+on algorithms

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Image Segmenta+on
• The Segment Editor module of 3D Slicer
provides powerful tools for manual and semi-
automated segmenta+on

• The module takes a reference image (Master


Volume) as input and produces a segmented
image (Segmenta+on) in output

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Terminology

Master Volume Segmenta:on Master Volume


&
Segmenta:on
In this tutorial,
- The Master Volume Meningioma is the ini+al brain MRI dataset
- The Segmenta+on Volume Segmenta+on is a segmenta+on
object with the same dimensions and origin as the Master
Volume
Sonia Pujol, PhD - All Rights Reserved 2019-2020
Segment Editor Module

LeZ Click on Data in the


Modules menu and select
the Segment Editor module
in the list of Modules

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

The Master Volume is the Meningioma dataset


The Segmenta+on volume is the segmenta+on
object created by Slicer with the same
dimensions and origin as the Master Volume

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Click on Add to add


a segment to the
Segmenta+on and
rename it Tumor

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Paint


tool and draw a
small mark inside
the tumor on 2-3
consecu+ve slices

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Click on Add to add a


new segment and
rename it Background

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Paint


Tool and draw a
circle around the
tumor

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Four-Up


slice layout

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Paint Tool and


draw a circle around the
tumor in the sagi[al and
coronal slices

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Grow


from seeds tool

Click on Ini+alize

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Slicer shows a preview of


the segmenta+on result

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Click on Apply to
accept the result

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Slicer shows the


segmenta+on result

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Click on Show 3D

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Slicer shows the


segmenta+on result in 3D

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the 3D only


layout

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Scissors


tool and remove
the segmented
sec+ons which are
not part of the
tumor

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Scissors


tool and remove the
segmented sec+ons
which are not part of
the tumor

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Segment Editor Module

Select the Scissors


tool and remove the
segmented sec+ons
which are not part of
the tumor

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Part 3: Tumor Volume
Calcula+on

Sonia Pujol, PhD - All Rights Reserved 2019-2020


3D Measurements
This sec+on shows how to
compute the volume of the
segmented tumor using the
Segment Sta+s+cs module

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Tumor Measurements

Select the Segment


Sta+s+cs module in
the Quan+fica+on
category

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Tumor Measurements

Select the Segmenta+on Segmenta+on


Select the Scalar Volume Meningioma

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Tumor Measurements

Select the
Conven+onal layout
and click on Apply

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Tumor Measurements

Slicer displays the


computed tumor volume:
17.053 cm3

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Part 4: Quan+ta+ve
Imaging features
computa+on

Sonia Pujol, PhD - All Rights Reserved 2019-2020


PyRadiomics package
• Pyradiomics is an open source python package that enables the
automated extrac+on of over 1,500 quan+ta+ve features from
medical imaging data.

• The package includes tools for image pre-processing and filtering


based on SimpleITK

• Pyradiomics command line tools enable batchprocessing

h[ps://pyradiomics.readthedocs.io

h[p://www.radiomics.io/
van Griethuysen, J. J. M., Fedorov, A., Parmar, C., Hosny, A., Aucoin, N., Narayan, V., Beets-Tan, R. G. H., Fillon-Robin, J.
C., Pieper, S., Aerts, H. J. W. L. (2017). ComputaNonal Radiomics System to Decode the Radiographic Phenotype. Cancer
Research, 77(21), e104–e107.

Sonia Pujol, PhD - All Rights Reserved 2011-2019


Slicer Radiomics

The Slicer Radiomics extension provides a graphical


user interface to the pyradiomics library
Sonia Pujol, PhD - All Rights Reserved 2011-2019
Slicer Radiomics
The Slicer Radiomics
Extension enables interac+ve
imaging features explora+on
and configura+on of
extrac+on parameters

Sonia Pujol, PhD - All Rights Reserved 2011-2019


Radiomics Module
LeZ click on the module Data to display
the list of modules

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Radiomics Module Installa+on

Select the module Radiomics in the category Informa+cs

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Radiomics Module Installa+on

Select the Image Volume Meningioma


Select the Input Regions Segmenta+on

Select the feature class First Order


and glcm

Select Output Table à Create New


and click on Apply

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Features Computa+on

Slicer displays the result of the


features computa+on

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Features Computa+on

Slicer displays the values of the 18


features for the FirstOrder class

Sonia Pujol, PhD - All Rights Reserved 2019-2020


Conclusion
• This tutorial provides a basic introduc+on to
image phenotyping using the Slicer Radiomics
extension.
• The extension enables the computa+on of
feature classes implemented in pyradiomics
• The descrip+on of each class is available at
h[ps://pyradiomics.readthedocs.io/

Sonia Pujol, PhD - All Rights Reserved 2011-2019


Acknowledgments:
Neuroimaging Analysis Center

The mission of the Neuroimage Analysis


Center (NAC) (NIH P41 EB015902)
is to advance the role of neuroimaging in
health care.

Sonia Pujol, PhD - All Rights Reserved 2011-2019

You might also like