Difference between revisions of "2012 Progress Report DBP Huntington's Disease"

From NAMIC Wiki
Jump to: navigation, search
Line 3: Line 3:
 
=Predict HD=
 
=Predict HD=
 
A. Introduction
 
A. Introduction
PREDICT-HD (Neurobiological Predictors of Huntington’s Disease) is a NIH-funded project to quantify the neurologic and morphologic changes in pre-symptomatic Huntington’s gene-positive carriers so that drug-therapy trials can be performed in patients before symptomatic onset.  The aims of our work are to 1) create longitudinal neurological morphometric analysis on individual subjects from multimodal data, 2) perform full brain diffusion tensor imaging (DTI) tractography analysis on individual subjects, and 3) to create rigorous and reproducible results through the deployment of extensible tools for sharing source data, derived results, algorithms, and methods to external, multi-site analysis groups (Figure 1).
+
PREDICT-HD (Neurobiological Predictors of Huntington’s Disease) is a NIH-funded project to quantify the neurologic and morphologic changes in pre-symptomatic Huntington’s gene-positive carriers so that drug-therapy trials can be performed in patients before symptomatic onset.  The aims of our work are to 1) create longitudinal neurological morphometric analysis on individual subjects from multimodal data, 2) perform full brain diffusion tensor imaging (DTI) tractography analysis on individual subjects, and 3) to create rigorous and reproducible results through the deployment of extensible tools for sharing source data, derived results, algorithms, and methods to external, multi-site analysis groups (Figure 1).<br>
Figure 1
+
Figure 1<br>
B. Research Progress
+
B. Research Progress<br>
During 2011-2012 we were able to accomplish many important advancements in the analysis of HD.  In addition, we were able to contribute many new tools to the NA-MIC community as well as provide training and exposure to researchers both within and outside the NA-MIC community.   
+
During 2011-2012 we were able to accomplish many important advancements in the analysis of HD.  In addition, we were able to contribute many new tools to the NA-MIC community as well as provide training and exposure to researchers both within and outside the NA-MIC community.  <br>
B.1 Year 2 Achievements: Aim 1
+
B.1 Year 2 Achievements: Aim 1<br>
Our goal for this year was to progress from the preliminary longitudinal shape analysis tools developed in Year 1 to more robust and mature versions appropriate for large cohort studies with increased automation and reliability.  The tool DTIPrep , developed collaboratively with Martin Styner’s lab at UNC-Chapel Hill, has been improved to correct for motion/eddy current artifacts, estimate and filter noise in DTI, achieve better DTI estimation, and generates a property map to provide comprehensive information about specific scan quality failures, thus allowing for meta-analysis of failures across site, scanner, and protocol.  In addition, DTI-Reg has been created by the UNC group to pipeline pair-wise DTI registration using scalar FA maps.  Individual steps of the pair-wise registration pipeline are performed via external applications - some of them being Slicer modules. Registration is performed between these FA maps via BRAINSDemonWarp or Advanced Normalization Toolkit (ANTS), which provide different registration schemes: rigid, affine, BSpline, diffeomorphic, and logDemons.
+
Our goal for this year was to progress from the preliminary longitudinal shape analysis tools developed in Year 1 to more robust and mature versions appropriate for large cohort studies with increased automation and reliability.  The tool DTIPrep , developed collaboratively with Martin Styner’s lab at UNC-Chapel Hill, has been improved to correct for motion/eddy current artifacts, estimate and filter noise in DTI, achieve better DTI estimation, and generates a property map to provide comprehensive information about specific scan quality failures, thus allowing for meta-analysis of failures across site, scanner, and protocol.  In addition, DTI-Reg has been created by the UNC group to pipeline pair-wise DTI registration using scalar FA maps.  Individual steps of the pair-wise registration pipeline are performed via external applications - some of them being Slicer modules. Registration is performed between these FA maps via BRAINSDemonWarp or Advanced Normalization Toolkit (ANTS), which provide different registration schemes: rigid, affine, BSpline, diffeomorphic, and logDemons.<br>
We contributed a Slicer3 compatible build system for the ANTS package, modularized the registration framework, and created a Slicer3 compatible utility called CompositeTransformUtil. This program can create a composite transform from a list of individual transform files in the ITKv4 compatible format, and save it to a single transform file. It can also read a composite transform and save each of its constituent transforms to a separate file.  This will allow us to bridge the ITKv3 based tools and the performance enhanced ITKv4 tools to create cross-sectional or longitudinal atlases from large cohort studies and measure subjects against a statistically valid mean anatomy.
+
We contributed a Slicer3 compatible build system for the ANTS package, modularized the registration framework, and created a Slicer3 compatible utility called CompositeTransformUtil. This program can create a composite transform from a list of individual transform files in the ITKv4 compatible format, and save it to a single transform file. It can also read a composite transform and save each of its constituent transforms to a separate file.  This will allow us to bridge the ITKv3 based tools and the performance enhanced ITKv4 tools to create cross-sectional or longitudinal atlases from large cohort studies and measure subjects against a statistically valid mean anatomy.<br>
B.2 Year 2 Achievements: Aim 2
+
B.2 Year 2 Achievements: Aim 2<br>
 
Joy Matsui and Mark Scully have integrated GTRACT  and DTIPrep into our longitudinal white matter analysis pipeline.  Additionally, Joy is working with Demian Wasserman (BWH) on developing appropriate data analysis for the pipeline results.  We currently have several papers in revision on her work with DTI analysis applied to subjects with apparent pathological changes from disease onset.
 
Joy Matsui and Mark Scully have integrated GTRACT  and DTIPrep into our longitudinal white matter analysis pipeline.  Additionally, Joy is working with Demian Wasserman (BWH) on developing appropriate data analysis for the pipeline results.  We currently have several papers in revision on her work with DTI analysis applied to subjects with apparent pathological changes from disease onset.
GTRACT has been integrated as an optional component in both the ITKv3 and the ITKv4 versions of Slicer.
+
GTRACT has been integrated as an optional component in both the ITKv3 and the ITKv4 versions of Slicer.<br>
B.3 Year 2 Achievements: Aim 3
+
B.3 Year 2 Achievements: Aim 3<br>
XNAT pipeline development has continued on schedule with the implementation of our BRAINSImageEval tool.  This application notifies our QA personnel when new scans have been uploaded to our server from our PREDICT-HD clinical sites, allows them to verify and score the usability of the data, and loads the image reviews on to our XNAT server.  The quality reports of the data are then used to identify the best processing strategies.   
+
XNAT pipeline development has continued on schedule with the implementation of our BRAINSImageEval tool.  This application notifies our QA personnel when new scans have been uploaded to our server from our PREDICT-HD clinical sites, allows them to verify and score the usability of the data, and loads the image reviews on to our XNAT server.  The quality reports of the data are then used to identify the best processing strategies.  <br>
In addition, Hans has worked closely with Satra Ghosh (BWH) on integrating Slicer modules with NiPype , a neuroimaging pipeline Python package, and improving the NiPype cluster processing for use on the University of Iowa’s High-Performance Computing (HPC) cluster.
+
In addition, Hans has worked closely with Satra Ghosh (BWH) on integrating Slicer modules with NiPype , a neuroimaging pipeline Python package, and improving the NiPype cluster processing for use on the University of Iowa’s High-Performance Computing (HPC) cluster.<br>
B.4 Year 2 Achievements: Additional
+
B.4 Year 2 Achievements: Additional<br>
Hans has been working closely with the ITK community to integrate the newest version of ITK (version 4) with Slicer, as well as integrating additional support for Python scripting functionality within Slicer (SimpleITK).  A bridging mechanism that allows convenient integration of SimpleITK and Slicer (sitkUtils) has been contributed.  At MICCAI 2011, the 2011 Winter Project Week, the March 21st-22nd Iowa Training conference, and during a 7 part lecture series at the University of Iowa he gave a tutorial presentations on Slicer, NAMIC, ITKv4 and SimpleITK.
+
Hans has been working closely with the ITK community to integrate the newest version of ITK (version 4) with Slicer, as well as integrating additional support for Python scripting functionality within Slicer (SimpleITK).  A bridging mechanism that allows convenient integration of SimpleITK and Slicer (sitkUtils) has been contributed.  At MICCAI 2011, the 2011 Winter Project Week, the March 21st-22nd Iowa Training conference, and during a 7 part lecture series at the University of Iowa he gave a tutorial presentations on Slicer, NAMIC, ITKv4 and SimpleITK.<br>
 
Hans, Joy, Regina, Mark, and Dave Welch participated in the 2010 Winter Project Meeting.  During the meeting, Dave began work with Ron Kikinis and Nicole Aucoin (BWH) on a fast registration module for the AMIGO  surgery suite (to be completed by June 2012), Regina and Mark implemented a prototype SPHARM  pipeline, Joy investigated fiber tracking methods in Slicer, and Hans met with NiPype’s lead developer.
 
Hans, Joy, Regina, Mark, and Dave Welch participated in the 2010 Winter Project Meeting.  During the meeting, Dave began work with Ron Kikinis and Nicole Aucoin (BWH) on a fast registration module for the AMIGO  surgery suite (to be completed by June 2012), Regina and Mark implemented a prototype SPHARM  pipeline, Joy investigated fiber tracking methods in Slicer, and Hans met with NiPype’s lead developer.
Iowa hosted a NA-MIC training session  this March that was a huge success, with an attendance of over 35 individuals from 6 departments during the two-day conference.  Later in April, Hans presented a seven-part lecture on 3D Slicer to the Iowa Institute of Biomedical Imaging (IIBI) with guest speakers from NiPype and Iowa’s HPC.
+
Iowa hosted a NA-MIC training session  this March that was a huge success, with an attendance of over 35 individuals from 6 departments during the two-day conference.  Later in April, Hans presented a seven-part lecture on 3D Slicer to the Iowa Institute of Biomedical Imaging (IIBI) with guest speakers from NiPype and Iowa’s HPC.<br>
 
Additionally, data sharing through XNAT has become a benchmark success for open-data methods in neuroimaging, with many researchers and institutions accessing to our database this year (Table 1).
 
Additionally, data sharing through XNAT has become a benchmark success for open-data methods in neuroimaging, with many researchers and institutions accessing to our database this year (Table 1).
Kent Williams updated and refactored the Dicom2Nrrd conversion module in Slicer and is waiting upon integration by the Slicer community.  This represents a major improvement to Slicer’s support for the DICOM file format, in particular for DTI and DWI data.  The previous version of Dicom2Nrrd was difficult to maintain and incompatible with ITKv4’s version of GDCM (Grassroots Dicom).  The current version is much better structured, easier to maintain and expand, compatible with ITKv4, and handles a wider variety of DICOM formats.
+
Kent Williams updated and refactored the Dicom2Nrrd conversion module in Slicer and is waiting upon integration by the Slicer community.  This represents a major improvement to Slicer’s support for the DICOM file format, in particular for DTI and DWI data.  The previous version of Dicom2Nrrd was difficult to maintain and incompatible with ITKv4’s version of GDCM (Grassroots Dicom).  The current version is much better structured, easier to maintain and expand, compatible with ITKv4, and handles a wider variety of DICOM formats.<br>
BRAINSTools multithreading currently depends on ITKv3 implementation of Mattes Mutual Information metric.  Significant effort was spent to debug and fix the multi-treading issues in ITK.  Where appropriate, those changes were included in ITKv3, otherwise the improvements are all available once the transition to ITKv4 occurs.  
+
BRAINSTools multithreading currently depends on ITKv3 implementation of Mattes Mutual Information metric.  Significant effort was spent to debug and fix the multi-treading issues in ITK.  Where appropriate, those changes were included in ITKv3, otherwise the improvements are all available once the transition to ITKv4 occurs. <br>
C. Plans for the Coming Year
+
C. Plans for the Coming Year<br>
Our lab has several papers in the progress that will be published during the next year.  We also have begun work on our longitudinal, large cohort study that will be completed within the year.  In addition, we continue with our efforts to integrate SPHARM with Slicer.
+
Our lab has several papers in the progress that will be published during the next year.  We also have begun work on our longitudinal, large cohort study that will be completed within the year.  In addition, we continue with our efforts to integrate SPHARM with Slicer.<br>
In year 3, David will create normative healthy and diseased models of subjects with the use of ANTS and SPHARM for Aim 1.  Aim 2 will be fulfilled with a NiPype-based workflow running our BRAINSStandAlone utilities in concert with the white matter tractography tools Joy has been developing.   
+
In year 3, David will create normative healthy and diseased models of subjects with the use of ANTS and SPHARM for Aim 1.  Aim 2 will be fulfilled with a NiPype-based workflow running our BRAINSStandAlone utilities in concert with the white matter tractography tools Joy has been developing.<br>  
Documentation and sharing of workflows will be done through the NA-MIC wiki to complete Aim 3.  In addition, the use of NiPype will make it possible to describe our methods in publications to such a degree that the effort needed for reproduction will be minimized.
+
Documentation and sharing of workflows will be done through the NA-MIC wiki to complete Aim 3.  In addition, the use of NiPype will make it possible to describe our methods in publications to such a degree that the effort needed for reproduction will be minimized.<br>
D. Papers that Acknowledge NA-MIC
+
D. Papers that Acknowledge NA-MIC<br>
• Magnotta VA, Matsui JT, Liu D, Johnson HJ, Long J, Bolster BD, Mueller BA, Lim K, Mori S, Helmer K, Turner JA, Reading S, Lowe MJ, Aylward E, Flashman LA, Bonett G, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Multi-center reliability of diffusion tensor imaging.”
+
• Magnotta VA, Matsui JT, Liu D, Johnson HJ, Long J, Bolster BD, Mueller BA, Lim K, Mori S, Helmer K, Turner JA, Reading S, Lowe MJ, Aylward E, Flashman LA, Bonett G, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Multi-center reliability of diffusion tensor imaging.”<br>
o To be submitted to Brain Connectivity (April 2012)
+
o To be submitted to Brain Connectivity (April 2012)<br>
• Matsui JT, Vaidya JG, Johnson HJ, Magnotta VA, Rao SM, Smith MM, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Using rotationally invariant scalars in specific regions of the prefrontal cortex to distinguish different stages of prodromal Huntington’s disease.”
+
• Matsui JT, Vaidya JG, Johnson HJ, Magnotta VA, Rao SM, Smith MM, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Using rotationally invariant scalars in specific regions of the prefrontal cortex to distinguish different stages of prodromal Huntington’s disease.”<br>
 
o To be submitted to Human Brain Mapping (June 2012)
 
o To be submitted to Human Brain Mapping (June 2012)
 
 

Revision as of 22:12, 18 April 2012

Home < 2012 Progress Report DBP Huntington's Disease

Back to NA-MIC_Progress Report_2012_Outline

Predict HD

A. Introduction PREDICT-HD (Neurobiological Predictors of Huntington’s Disease) is a NIH-funded project to quantify the neurologic and morphologic changes in pre-symptomatic Huntington’s gene-positive carriers so that drug-therapy trials can be performed in patients before symptomatic onset. The aims of our work are to 1) create longitudinal neurological morphometric analysis on individual subjects from multimodal data, 2) perform full brain diffusion tensor imaging (DTI) tractography analysis on individual subjects, and 3) to create rigorous and reproducible results through the deployment of extensible tools for sharing source data, derived results, algorithms, and methods to external, multi-site analysis groups (Figure 1).
Figure 1
B. Research Progress
During 2011-2012 we were able to accomplish many important advancements in the analysis of HD. In addition, we were able to contribute many new tools to the NA-MIC community as well as provide training and exposure to researchers both within and outside the NA-MIC community.
B.1 Year 2 Achievements: Aim 1
Our goal for this year was to progress from the preliminary longitudinal shape analysis tools developed in Year 1 to more robust and mature versions appropriate for large cohort studies with increased automation and reliability. The tool DTIPrep , developed collaboratively with Martin Styner’s lab at UNC-Chapel Hill, has been improved to correct for motion/eddy current artifacts, estimate and filter noise in DTI, achieve better DTI estimation, and generates a property map to provide comprehensive information about specific scan quality failures, thus allowing for meta-analysis of failures across site, scanner, and protocol. In addition, DTI-Reg has been created by the UNC group to pipeline pair-wise DTI registration using scalar FA maps. Individual steps of the pair-wise registration pipeline are performed via external applications - some of them being Slicer modules. Registration is performed between these FA maps via BRAINSDemonWarp or Advanced Normalization Toolkit (ANTS), which provide different registration schemes: rigid, affine, BSpline, diffeomorphic, and logDemons.
We contributed a Slicer3 compatible build system for the ANTS package, modularized the registration framework, and created a Slicer3 compatible utility called CompositeTransformUtil. This program can create a composite transform from a list of individual transform files in the ITKv4 compatible format, and save it to a single transform file. It can also read a composite transform and save each of its constituent transforms to a separate file. This will allow us to bridge the ITKv3 based tools and the performance enhanced ITKv4 tools to create cross-sectional or longitudinal atlases from large cohort studies and measure subjects against a statistically valid mean anatomy.
B.2 Year 2 Achievements: Aim 2
Joy Matsui and Mark Scully have integrated GTRACT and DTIPrep into our longitudinal white matter analysis pipeline. Additionally, Joy is working with Demian Wasserman (BWH) on developing appropriate data analysis for the pipeline results. We currently have several papers in revision on her work with DTI analysis applied to subjects with apparent pathological changes from disease onset. GTRACT has been integrated as an optional component in both the ITKv3 and the ITKv4 versions of Slicer.
B.3 Year 2 Achievements: Aim 3
XNAT pipeline development has continued on schedule with the implementation of our BRAINSImageEval tool. This application notifies our QA personnel when new scans have been uploaded to our server from our PREDICT-HD clinical sites, allows them to verify and score the usability of the data, and loads the image reviews on to our XNAT server. The quality reports of the data are then used to identify the best processing strategies.
In addition, Hans has worked closely with Satra Ghosh (BWH) on integrating Slicer modules with NiPype , a neuroimaging pipeline Python package, and improving the NiPype cluster processing for use on the University of Iowa’s High-Performance Computing (HPC) cluster.
B.4 Year 2 Achievements: Additional
Hans has been working closely with the ITK community to integrate the newest version of ITK (version 4) with Slicer, as well as integrating additional support for Python scripting functionality within Slicer (SimpleITK). A bridging mechanism that allows convenient integration of SimpleITK and Slicer (sitkUtils) has been contributed. At MICCAI 2011, the 2011 Winter Project Week, the March 21st-22nd Iowa Training conference, and during a 7 part lecture series at the University of Iowa he gave a tutorial presentations on Slicer, NAMIC, ITKv4 and SimpleITK.
Hans, Joy, Regina, Mark, and Dave Welch participated in the 2010 Winter Project Meeting. During the meeting, Dave began work with Ron Kikinis and Nicole Aucoin (BWH) on a fast registration module for the AMIGO surgery suite (to be completed by June 2012), Regina and Mark implemented a prototype SPHARM pipeline, Joy investigated fiber tracking methods in Slicer, and Hans met with NiPype’s lead developer. Iowa hosted a NA-MIC training session this March that was a huge success, with an attendance of over 35 individuals from 6 departments during the two-day conference. Later in April, Hans presented a seven-part lecture on 3D Slicer to the Iowa Institute of Biomedical Imaging (IIBI) with guest speakers from NiPype and Iowa’s HPC.
Additionally, data sharing through XNAT has become a benchmark success for open-data methods in neuroimaging, with many researchers and institutions accessing to our database this year (Table 1). Kent Williams updated and refactored the Dicom2Nrrd conversion module in Slicer and is waiting upon integration by the Slicer community. This represents a major improvement to Slicer’s support for the DICOM file format, in particular for DTI and DWI data. The previous version of Dicom2Nrrd was difficult to maintain and incompatible with ITKv4’s version of GDCM (Grassroots Dicom). The current version is much better structured, easier to maintain and expand, compatible with ITKv4, and handles a wider variety of DICOM formats.
BRAINSTools multithreading currently depends on ITKv3 implementation of Mattes Mutual Information metric. Significant effort was spent to debug and fix the multi-treading issues in ITK. Where appropriate, those changes were included in ITKv3, otherwise the improvements are all available once the transition to ITKv4 occurs.
C. Plans for the Coming Year
Our lab has several papers in the progress that will be published during the next year. We also have begun work on our longitudinal, large cohort study that will be completed within the year. In addition, we continue with our efforts to integrate SPHARM with Slicer.
In year 3, David will create normative healthy and diseased models of subjects with the use of ANTS and SPHARM for Aim 1. Aim 2 will be fulfilled with a NiPype-based workflow running our BRAINSStandAlone utilities in concert with the white matter tractography tools Joy has been developing.
Documentation and sharing of workflows will be done through the NA-MIC wiki to complete Aim 3. In addition, the use of NiPype will make it possible to describe our methods in publications to such a degree that the effort needed for reproduction will be minimized.
D. Papers that Acknowledge NA-MIC
• Magnotta VA, Matsui JT, Liu D, Johnson HJ, Long J, Bolster BD, Mueller BA, Lim K, Mori S, Helmer K, Turner JA, Reading S, Lowe MJ, Aylward E, Flashman LA, Bonett G, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Multi-center reliability of diffusion tensor imaging.”
o To be submitted to Brain Connectivity (April 2012)
• Matsui JT, Vaidya JG, Johnson HJ, Magnotta VA, Rao SM, Smith MM, Paulsen JS, PREDICT-HD Investigators of the Huntington Study Group. “Using rotationally invariant scalars in specific regions of the prefrontal cortex to distinguish different stages of prodromal Huntington’s disease.”
o To be submitted to Human Brain Mapping (June 2012)