DIPA: Diffusion Image Processing and Analysis - PowerPoint Presentation

343 views
Uploaded On 2020-06-29

DIPA: Diffusion Image Processing and Analysis - PPT Presentation

Andrew Schoen Center for Healthy Minds at the Waisman Center University of Wisconsin Madison What do we do What this looks like Waisman Center Center for Healthy Minds MIDUS Study Wide variety of assessments ID: 788997

dipa config waisman processing config dipa processing waisman workflow data code files handling file dmri midus htcondor model component

Link:

Copy

Embed:

<iframe width="560" height="315" src="https://www.docslides.com/embed/788997" frameborder="0" allowfullscreen></iframe>

Download Presentation from below link

Download The PPT/PDF document "DIPA: Diffusion Image Processing and An..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.

Presentation Transcript

Slide1

DIPA: Diffusion Image Processing and Analysis

Andrew Schoen

Center for Healthy Minds at the Waisman Center

University of Wisconsin - Madison

Slide2

What do we do, What this looks like:Waisman CenterCenter for Healthy Minds / MIDUS StudyWide variety of assessments

dMRI

Diffusion of water molecules in the brain

Used to model pathways throughout the brainDifficultiesHighly susceptible to noise, artifact, motionMultiple levels of processing, many optionsPoorly streamlined, poorly standardized processing

Slide3

Model Fitting

Normalization

Extraction

Tractography

Workflow

An extreme simplification of our workflow, with two participants’ data.

Slide4

Gotchas with moving to HTCondorSometimes slow rates of data collectionNature of individual studies determines workflow.Exchangeability of different processing stepsComplicated code distributionsIterative processingData QualityAccessibility

Slide5

The Start of DIPADTITK -> SetupCondorDTITKVery simple condorization of a linear, iterative processing stepI tore apart an existing package.Generated a DAG.Only worked on shared file systemsDIPALet’s get it working with transfersRinse and repeat:

Hey, this works pretty well. Let’s add things before and after.

Wait, this is getting a bit too complicated. Improve code base to make it easier.

Slide6

DIPA -h

Slide7

DIPA Correct -a

Slide8

DIPA StructureSubclass “Component”, and overwrite necessary methods.setComponentConfiguration(self, inputs, PreviousConfig, config)setComponentInputSpecFile(self, config)setComponentMatrices(self, config

)

setComponentDirectories

(self,

config)setComponentSourceFileManifest(self, config)setComponentFiles(self, config)

setComponentScripts

(self,

config

)

performComponentSetup(self, config)You provide the “Component”, add it to a json file, and DIPA handles the rest.

Slide9

DIPA in the FutureSimplified developer interfaceBetter handling of files, variablesSmarter handling of input files, intermediates, and skipped processing stepsMore informative help pagesPegasusImproved restarting/expansion of previously run workflowsImproved multi-site implementationGranular monitoring

0.4.0 BETA

Slide10

CreditsNagesh Adluru : dMRI specialist, theory consultant.Nate Vack : General programming consultantMichael Stoneman : Undergrad assistantStacey Schaefer : MIDUS lead scientistDavid Thompson : Waisman

Sys-Admin

Andy Pohl :

HTCondor

Waisman contactKaran Vahi : PegasusThe whole HTCondor Team

Slide11

Questions, suggestions, comments