NA-MIC National Alliance for Medical Image Computing BRAINSCut General Tutorial Eun Young(Regina) Kim University of Iowa

Slides:



Advertisements
Similar presentations
Slicer3 for developers – Sonia Pujol, Ph.D. -1- National Alliance for Medical Image Computing Programming into Slicer3 Sonia Pujol, Ph.D. Surgical Planning.
Advertisements

NA-MIC National Alliance for Medical Image Computing © 2010, All Rights Reserved Diffusion Tensor Imaging Tutorial Sonia Pujol, Ph.D.
National Alliance for Medical Image Computing © 2010, All Rights Reserved Sonia Pujol, PhD Slicer3Minute Tutorial Sonia.
NA-MIC National Alliance for Medical Image Computing Diffusion Tensor Imaging tutorial Sonia Pujol, PhD Surgical Planning Laboratory.
NA-MIC National Alliance for Medical Image Computing © 2010, All Rights Reserved Diffusion Tensor Imaging Tutorial Sonia Pujol, Ph.D.
NA-MIC National Alliance for Medical Image Computing Connected Threshold Image Filter Salma Bengali, Alan Morris, Josh Cates, Rob.
NA-MIC National Alliance for Medical Image Computing CARMA Inhomogeneity Correction Filter Alan Morris, Eugene Kholmovski, Josh Cates,
NA-MIC National Alliance for Medical Image Computing [Tutorial Name] [List of authors] [Institution] [ of the first author]
A Brief Overview of Neural Networks By Rohit Dua, Samuel A. Mulder, Steve E. Watkins, and Donald C. Wunsch.
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial / Registration Library: Case 03 - DTI aligning low-resolution diffusion.
-1- Massachusetts General Hospital National Alliance for Medical Image Computing Using Plastimatch for Landmark-Based Registration Nadya Shusharina Department.
Slicer Welcome Tutorial
NA-MIC National Alliance for Medical Image Computing [Tutorial Name] [List of authors] [Institution] [ of the first author]
November 19, 2009Introduction to Cognitive Science Lecture 20: Artificial Neural Networks I 1 Artificial Neural Network (ANN) Paradigms Overview: The Backpropagation.
1 Chapter 11 Neural Networks. 2 Chapter 11 Contents (1) l Biological Neurons l Artificial Neurons l Perceptrons l Multilayer Neural Networks l Backpropagation.
Neural Networks. R & G Chapter Feed-Forward Neural Networks otherwise known as The Multi-layer Perceptron or The Back-Propagation Neural Network.
(Page 554 – 564) Ping Perez CS 147 Summer 2001 Alternative Parallel Architectures  Dataflow  Systolic arrays  Neural networks.
NA-MIC National Alliance for Medical Image Computing CARMA Registration Alan Morris, Greg Gardner, Salma Bengali, Josh Cates, Rob.
NA-MIC National Alliance for Medical Image Computing CMR Toolkit Threshold Model Module Salma Bengali, Alan Morris, Josh Cates, Rob.
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial: Registration Library Case 19 Multi-stage registration for group.
NA-MIC National Alliance for Medical Image Computing Non-rigid MR-CT Image Registration Atsushi Yamada, Dominik S. Meier and Nobuhiko.
Traffic Sign Recognition Using Artificial Neural Network Radi Bekker
NA-MIC National Alliance for Medical Image Computing CARMA Registration Salma Bengali, Alan Morris, Greg Gardner, Josh Cates, Rob.
NA-MIC National Alliance for Medical Image Computing Interactive Editor tutorial Sonia Pujol, Ph.D. Surgical Planning Laboratory Harvard.
-1- Pujol S et al. National Alliance for Medical Image Computing 3D Visualization of FreeSurfer Data Sonia Pujol, Ph.D. Silas Mann, B.Sc. Randy Gollub,
1 Introduction to Artificial Neural Networks Andrew L. Nelson Visiting Research Faculty University of South Florida.
NA-MIC National Alliance for Medical Image Computing shapeAnalysisMANCOVA_Wizar d Lucile Bompard, Clement Vacher, Beatriz Paniagua, Martin.
DIGITAL IMAGE PROCESSING Dr J. Shanbehzadeh M. Hosseinajad ( J.Shanbehzadeh M. Hosseinajad)
NA-MIC National Alliance for Medical Image Computing Fiducials Nicole Aucoin Brigham and Women's Hospital
-1- 3D Visualization. Sonia Pujol, Ph.D., Harvard Medical School National Alliance for Medical Image Computing 3D Visualization Sonia Pujol, Ph.D. Surgical.
NA-MIC National Alliance for Medical Image Computing Non-rigid Registration of Pre- procedural MRI and Intra-procedural CT for CT-guided.
1 Chapter 11 Neural Networks. 2 Chapter 11 Contents (1) l Biological Neurons l Artificial Neurons l Perceptrons l Multilayer Neural Networks l Backpropagation.
NA-MIC National Alliance for Medical Image Computing DTI Atlas Registration via 3D Slicer and DTI-Reg Martin Styner, UNC Clement Vachet,
DICOM to NRRD Conversion Tutorial Martin Styner 1 University of North Carolina Neuro Image Research and Analysis Lab.
Sonia Pujol, PhD National Alliance for Medical Image Computing © 2010, ARR.
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial: Registration Library Case 14 Intra-subject Brain PET-MRI fusion.
NA-MIC National Alliance for Medical Image Computing Non-rigid MR-CT Image Registration Atsushi Yamada, Dominik S. Meier and Nobuhiko.
NA-MIC National Alliance for Medical Image Computing BRAINSCut General Tutorial Eun Young(Regina) Kim University of Iowa
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial Registration Library Case 05: Knee MRI: model/surface registration.
Slicer3 for developers – S.Pujol -1- National Alliance for Medical Image Computing Slicer3 Course for developers Sonia Pujol, Ph.D. Surgical Planning Laboratory.
Exploring Peritumoral White Matter Fibers for Neurosurgical Planning Sonia Pujol, Ph.D. Ron Kikinis, M.D. Surgical Planning Laboratory Harvard University.
NA-MIC National Alliance for Medical Image Computing Shape analysis using spherical harmonics Lucile Bompard, Clement Vachet, Beatriz.
NA-MIC National Alliance for Medical Image Computing Using Annotations in Slicer 4.0 Yong Zhang, Kilian Pohl June 2010.
Programming into Slicer3. Sonia Pujol, Ph.D., Harvard Medical School National Alliance for Medical Image Computing ©
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial: Registration Library Case 15 AC-PC Alignment Dominik Meier, Ron.
NA-MIC National Alliance for Medical Image Computing Diffusion Tensor Imaging tutorial Sonia Pujol, PhD Surgical Planning Laboratory.
NA-MIC National Alliance for Medical Image Computing PV Antrum Cut Filter Alan Morris, Salma Bengali, Greg Gardner, Josh Cates, Rob.
Introduction to Neural Networks Introduction to Neural Networks Applied to OCR and Speech Recognition An actual neuron A crude model of a neuron Computational.
-1- Massachusetts General Hospital National Alliance for Medical Image Computing Using Plastimatch for DICOM / DICOM-RT Import Gregory C. Sharp Department.
Hazırlayan NEURAL NETWORKS Backpropagation Network PROF. DR. YUSUF OYSAL.
Diffusion Tensor Analysis in Slicer3
Sonia Pujol, PhD -1- National Alliance for Medical Image Computing Neuroimage Analysis Center Diffusion Tensor Imaging tutorial Sonia Pujol, Ph.D. Surgical.
COSC 4426 AJ Boulay Julia Johnson Artificial Neural Networks: Introduction to Soft Computing (Textbook)
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial Nonrigid Atlas Registration Dominik Meier, Ron Kikinis February.
NA-MIC National Alliance for Medical Image Computing Slicer3 Tutorial Registration Library Case 06: Breast Cancer Follow-up Dominik Meier,
Bab 5 Classification: Alternative Techniques Part 4 Artificial Neural Networks Based Classifer.
Standard Response Evaluation Criteria in Solid Tumors (RECIST) using 3D Slicer Slicer3 Training Compendium Image here. Jeffrey Yap, PhD Wendy Plesniak,
Department of Psychiatry, Department of Computer Science, 3 Carolina Institute for Developmental Disabilities 1 Department of Psychiatry, 2 Department.
Kim HS Introduction considering that the amount of MRI data to analyze in present-day clinical trials is often on the order of hundreds or.
Surgical Planning Laboratory -1- Brigham and Women’s Hospital Slicer Tutorial 4 Module: DTMRI Data: Sample Slicer DTI Sonia Pujol,
3D Visualization Sonia Pujol, Ph.D. Instructor of Radiology
Fiber Bundle Volume Measurement
Cardiac Agatston Scoring
Non-rigid MR-CT Image Registration
shapeAnalysisMANCOVA_Wizard
3D Visualization of FreeSurfer Data
Face Recognition with Neural Networks
[Tutorial Name] [List of authors] [Institution]
Automatic SPHARM Shape Analysis in 3D Slicer
Interactive Editor tutorial
Presentation transcript:

NA-MIC National Alliance for Medical Image Computing BRAINSCut General Tutorial Eun Young(Regina) Kim University of Iowa NA-MIC Tutorial Contest: Summer 2011 © 2011, All Rights Reserved

National Alliance for Medical Image Computing Learning Objective BRAINSCut This tutorial shows how to segment regions of interest(ROI) by using BRAINSCut. The tutorial example includes sub-cortical structure segmentation from human MR brain Image. © 2011, All Rights Reserved BRAINSCut Output example for sub-cortical structures

National Alliance for Medical Image Computing This is NOT mandatory but would be helpful when you run command line module for BRAINSCut –Tutorial Name: Programming in Slicer3 Tutorial –Author: Sonia Pujol, Ph.D –Tutorial Link: gIntoSlicer3_SoniaPujol.pdf Pre-requisite © 2011, All Rights Reserved

National Alliance for Medical Image Computing This tutorial requires the installation of the Slicer3.6 release or Slicer4 and the tutorial dataset. They are available at the following locations: Slicer download page Tutorial dataset: [name of dataset] [Add a link to the tutorial dataset] Material © 2011, All Rights Reserved

National Alliance for Medical Image Computing [Add the results of the tests on Linux, Mac and Windows] Platforms © 2011, All Rights Reserved

National Alliance for Medical Image Computing The tutorial starts with brief background for BRAINSCut and followed by step-by-step instruction how to get sub-cortical volumetric segmentation. BRAINSCut Background BRAINSCut Process 1.BRAINSCut I/O: Understand and preparation 2.Train BRAINSCut: Tell BRAINSCut what’s your ROI 3.Apply BRAINSCut: Get the actual volumetric segmentation Overview © 2011, All Rights Reserved

National Alliance for Medical Image Computing BRAINSCut Background BRAINSCut uses Artificial Neural Net(ANN) learning mechanism in its segmentation process. Here briefly introduce how ANN works for image segmentation. BIOLOGICAL NEURON PERCEPTRONS ARTIFICIAL NEURAL NETWORK Inspired by human brain signaling of neuron, ANN learns from training set, which is paired set of input and output. This training process is simply adjusting process of network, including size of network(e.g. number of perceptrons) and weights to match computed output from input as similar as possible to the paired output.

National Alliance for Medical Image Computing BRAINSCut Process BRAINSCut process starts with preparation for learning/training for ROI. To train BRAINSCut, as stated in the previous slide, a paired input-output training set has to be provided. Processes of Generating probability map and Creating Vector together produces training data set for your ROI. Train Model process is followed by those step to learn about the structure and then completed with Apply Model step to get volumetric segmentation. Generate Probability Create Vector Train Model Apply Model Create a paired training set for ROI Train BRAINSCut with ANN Get a segmentation!

National Alliance for Medical Image Computing BRAINSCut Process BRAINSCut I/O: Net Configuration File : All the image volumes for BRAINSCut has to be indicated in the net configuration file(XML file format). Please download Tutorial Data File and see the following files. -File name: BRAINSCutGeneralTutorialNetConfiguration.xml This file includes general setting for BRAINSCut with tutorial data set in it. After following the tutorial, you should be able to modify the net configuration file to fit your purpose.

National Alliance for Medical Image Computing BRAINSCut Process BRAINSCut I/O: Net Configuration File 1.Open the net configuration file (BRAINSCutGeneralTutorialNetConfiguration.xml) with your favorite text editor 2.Here we need to change two variables for input and output directory name. 3.As shown in below figure, please change all [BRAINSCutInputDir] to your downloaded location. Original provided file Modified file example

National Alliance for Medical Image Computing BRAINSCut Process BRAINSCut I/O: Net Configuration File 4.In a same way, please change all [BRAINSCutOutputDir] to your preferred location to store BRAINSCut output. Original provided file Modified file example Now, we are ready to go…

National Alliance for Medical Image Computing BRAINSCut Process Generate Probability Create Vector Train Model Apply Model $>[BRAINSCut Build Directory]/BRAINSCut \ --netConfiguration BRAINSCutGeneralTutorialNetConfiguration.xml \ --generateProbability Command Line Example -Probability map is to identify rough region of segmentation. By registering each training data to the atlas data set, probability map(spatial prior) of each ROI could be created. -ROI could be specified in the net configuration file in the section. Right CaudateLeft PutamenLeft Globus

National Alliance for Medical Image Computing BRAINSCut Process Generate Probability Create Vector Train Model Apply Model [BRAINSCut Build Directory]/BRAINSCut \ --netConfiguration BRAINSCutGeneralTutorialNetConfiguration.xml \ --createVectors Command Line Example Neighbors along the Gradient Descents Training data, input-output paired set, will be generated by above command. When creating training vector, BRAINSCut will use all image types specified in the Atlas Data Set Use these type image

National Alliance for Medical Image Computing BRAINSCut Process Generate Probability Create Vector Train Model Apply Model [BRAINSCut Build Directory]/BRAINSCut \ --netConfiguration BRAINSCutGeneralTutorialNetConfiguration.xml \ --trainModel Command Line Example Input Layer Hidden Layer Output Layer z ρ φ θ After creating training vector set, BRAINSCut will train for your ROI. The figure below shows the architecture of BRAINSCut.

National Alliance for Medical Image Computing BRAINSCut Process Generate Probability Create Vector Train Model Apply Model [BRAINSCut Build Directory]/BRAINSCut \ --netConfiguration BRAINSCutApplyPutamenTutorialNetConfiguration.xml \ --applyModel Command Line Example Now, it is time to apply our trained data set. Since trained model from previous section could be quite ugly, please open up the BRAINSCutApplyTutorialNetConfiguration.xml file. The file includes finely trained ANN model for Caudate and Putamen for now. For other structures we are working on and will be updated as soon as we get. By doing above line, segmentation output will be generated as in the directory specified at the net configuration file. In this tutorial example, the result file would go into the same directory as BRAINSCutOutputDir. [BRAINSCut Build Directory]/BRAINSCut \ --netConfiguration BRAINSCutApplyCuadateTutorialNetConfiguration.xml \ --applyModel

National Alliance for Medical Image Computing Conclusion © 2011, All Rights Reserved

National Alliance for Medical Image Computing National Alliance for Medical Image Computing NIH U54EB [Other Acknowledgments] Acknowledgments © 2011, All Rights Reserved