Microsoft word - description_project_pipelinenew2.doc

 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab PARTIAL VOLUME ERROR .3
THE PIPELINE PROGRAM .4
THE USER-INTERFACE .4 Restrictions using project .7 Syntax and indices.8 The project field .9 The pipeline field and sub-fields .10 The taskDone field and its sub-fields .12 The handles field .13 The info field .14 REALISING THE PIPELINE PROGRAM.14
MATLAB AND THE PIPELINE PROGRAM.15 CONNECTION DIAGRAM OF THE PIPELINE PROGRAM.15 A FUNCTION_WRAPPER.16 A CONFIGURATOR_WRAPPER .17 THE MAIN_WRAPPER.17 checkIn .17 checkOut .17 PREPARATION: A TASK GUIDELINE .18 PREPARATION: A METHOD GUIDELINE.19 HOW TO… .20
HOW TO WRITE A FUNCTION WRAPPER .20 HOW TO WRITE A CONFIGURATOR WRAPPER .22 HOW TO MAKE A SET-UP FILE .22 FILE FORMATS AND DIRECTORIES.25  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 1 Partial Volume Error
Intro.
What is PVE project
Groups implementation of methods. Tasks in a PVE correction process.
Set-up at pipeline w. given methods, log all necessary information ´structure, and control a process…One success criteria for the PVEout (Partial Volume Effect) project within theEuropean 5th framework is to make a common interface for a product that can do afull partial volume correction of a PET/SPECT scan using a high resolution MR scan.
At the Firenze meeting, Italy, mid November 2002, the partners agreed to make aproposal on how the different methods in the PVE process could be integratedthrough a user-friendly graphic user interface (GUI). Further, the partners agreed inusing the Matlab program from MathWorks Inc. for implementing the GUI.
A major advantage using Matlab as platform is the non-dependency of the operatingsystem (Linux or Windows), the worldwide use of Matlab within medical community,and an easy way to realize the GUI.
The Copenhagen group, NRU, would make the preliminary description. The result isgiven in the following document where a full (preliminary) description on how anuser friendly interface could be realized, code be written and, different methodsintegrated. Further a proposed prototype of a GUI is created in Matlab.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 2 The pipeline program
In the following a pipeline program is introduced. The pipeline program offers an
easy way to set-up a pipeline containing one or more steps in a user-defined order. A
step in the pipeline is refereed as a task and within each task there exist none or more
programs each realising the given. Such a program for a task is refereed as a method.
The pipeline program is based on a documented data structure where information ofthe individual pipeline is stored. The data structure can be seen as a read/writeabledatabase refereed as the project for the actual pipeline. For each project a uniqueworking directory is automatically created, where files generated through a pipelineare stored. In this way files within different projects are not mixed up and morepipeline programs can be executed at the same time.
A user-friendly interface gives a good overview of a project and makes the pipelineprogram easy-to-use. Some features are as followed:- to select other of available methods than default selected- to save user defined settings for a actual pipeline- to load and continue a not finished project- inspect used settings and other information for a finished project The data structure is independent of the user-interface, which means that the pipelineprogram in principle can be executed without any user-interface (GUI). This gives thepossibility to run the pipeline program on clusters of computers. Of cause this featureexpect that methods in the actual pipeline are fully automatic and either does not needany user-interface.
To set-up a pipeline two-steps must be done:1) Each method must interface the pipeline program through a simple wrapper that shields a method from the pipeline program.
2) Create a set-up file defining execution order of the tasks and needed information of the different methods. Guideline to full-fill and examples are given.
Because …A user implementing a new pipeline can stay focused on the given two steps, if errorsare found/detected.
2.1 The user-interface
Users using a already existing set-up of a pipeline.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03

The pipeline program for PVElab  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03

The pipeline program for PVElab The PVElab offer methods to do a PVE correction of a low resolution PET or SPECTscan (functional image) given a high resolution MR scan (structural image).
This demo presents our ideas behind the PVElab design. This document explains howthe design and layout of the GUI should work.
The PVElab GUI is build about tasks. Each task is linked to a task button in the top.
Each task can be accomplished by a number of methods. Registration for example canbe done by IIO, IPS etc.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab The results of a method are displayed in the right frame. The idea with PVElab is thatit displays the selected method and offers a possibility to quickly inspect the results.
If more interaction from the user is required, a new figure can be created.
The method is selected from the pop up menu. The info field displays clues to guidethe user through all the tasks. When a method is selected a short method description isdisplayed in the info field.
A log is also created, so you can see and document how your results were obtained.
You can keep an eye on the log because it is also displayed at the bottom of thefigure.
Pressing the ‘Options' button can configure each method. In the options figure it ispossible to reset the method to default settings.
The tasks display their state by a small symbol. The states are I'm next - Completed- Error .
You can save all your work in a project. This project contains all the settings for alltasks. When you load your project you can change some options and run thecalculations again.
A summary of the current menu structure: 2.2 The project
As introduced each project contain a data structure where all information for apipeline are stored. Information regards settings for the tasks and available methods inthe pipeline, results, log information and status of the pipeline process.
In the following section all fields of the data structure in a project will be explained
shortly. This section is intended for programmers who want to implement new tasks
and/or methods. All so users who want a deeper understand of the data structure or
like to retrieve extended information from the project.
How to write wrapper functions that interface to the pipeline program and setting up
pipelines, see section Error! Reference source not found..
2.2.1 Restrictions using project The data structure of a project is in general a READ-ONLY structure, unless else isgiven and a field is marked with a start (*) in the following sections. A pipeline in thepipeline program is a collection of different tasks and methods implemented bydifferent programmers. It is therefore obviously, not respecting given restrictions maycause to instability, not generality and misunderstanding using the pipeline program.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab Do not store larges data amounts in the project such as images. Matlab controls theproject where pointers not are available. The data structure is therefore transferredbetween functions, each making a copy of the project. This can make the pipelineprogram slow and using unnecessary memory if lot of data is stored in the project.
2.2.2 Syntax and indices To navigate in a project one must understand the basic syntax of a data structure andused indices.
A data structure is build up of blocks containing one or more fields of those illustratedin Figure 5.
Type of fields:- field0: Contain one or more given type of fields: field0 to field4.
- field1: One variable of type string, number or a field.
- field2(index): Vector a specific type of variables, as numbers or strings with the same length.
- field3{index}:Vector of none-specific type of variables or fields.
- field{index1,index2}: As field3, but with an extra dimension.
To access field1 to field4 in structure field0 do:Reading a field into variable A: Writing variable to a field: Used indices navigating a project in the pipeline program:
- Index: not a specific index.
- TaskIndex: Selected task of a defined order in the pipeline. First task is 1
- MethodIndex: Selected method of available methods in a task: First method is 1
- ModalityIndex: Selected modality of available modalities: First modality is 1
- ImageIndex: Selected image of available images: First image is 1 (default)
- FunctionIndex: Selected function of available functions: First function is 1
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 2.2.3 The project field The project has four fields each containing a structure, which are explained in thefollowing sections.
Fields of the project:
- pipeline: Contain both settings for the used pipeline and status information of the
pipeline process.
- taskDone{TaskIndex}: Information of a finished tasks in the pipeline process.
- Handles: Handles/pointers to used figures in the pipeline program.
- sysInfo: General information of used directories and files.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 2.2.4 The pipeline field and sub-fields The pipeline field contain all information regarding set-up and process status andcontrol of a pipeline. The field taskSetup{TaskIndex,MethodIndex} contain a sub-fieldof settings for the given method of a task. This task{,}-field sets up a given pipelineand is therefore refereed as a set-up file. The set-up file is an input parameter whenstarting the pipeline program.
The other fields in the pipeline field are mainly process status and control of apipeline.
The pipeline field is shown in Figure 7 and a short description of each field is given inthe following.
Fields of pipeline:
• LastTaskIndex(Index): Vector containing the index to the last succeeded task in
an increasing order.
• StatusTask(TaskIndex): Vector containing a status-flag of each task: 0= Ready,
1=Active/runing and 2=Done (if succeed) • DefaultPipeline(TaskIndex):Vector containing index to the method which is by
default chosen when PVElab is started.
• UserPipeline(TaskIndex): Vector containing index to the methods used in the
pipeline of a PVE correction process. A user could have changed the defaultmethod indices.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab • taskSetup{TaskIndex,MethodIndex}: Cell-array containing a sub-entry to a
setup structure of a given method in a given task.
− task: Task that method belongs to. The task name will appear in the pipeline
− method: Name of the method (e.g short-term) to appear in the pipeline
− method_name: Full name of method.
− function_name: Program/function realising the method.
− function_wrapper: Wrapper that interface method and the pipeline program.
− description: Information of method, will appear in the pipeline program
− requireTaskIndex: Index of which tasks has to be finished before current
− prefix: Added to output files of actual task.
− documentation: Text or link how to get further documentation of given
method. Will appear in ‘about menu' of the pipeline program.
− who_did: Name of group and year. Will appear in ‘about menu' of the
pipeline program.
− configuratorfile: (if exist) text file with settings for actual method
o filename: Name of file
o path: path to file
− configurator_wrapper: (if exist) Wrapper to interface a configurator and the
pipeline program.
− versionlabel: Text label identifying version number of given method in a
matlab file.
− fileexist: Ready flag if all needed functions in the task field are found by the
pipieline program. 1=exist , 0=do not exist and method can not be executed.
− filestatus{functionIndex}: File status of functions for a given method in the
current field of tasks{,}, which should be available to the pipeline program.
o path: Where function is found
o indexname: Field name
o filename: Name of file for the actual function.
o version: Found version label.
o exist: If file exist = 1 else = 0.
o os: Actual operation system
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 2.2.5 The taskDone field and its sub-fields Information for each finished task in the pipeline is stored in this field. This regardsfrom selected method, settings and program files. If an error is detected the field iscleaned up and deletes all generated files, except input and program files.
In the field ‘userdata' additional user information can be written either into a variableor a sub structure. All other fields are read-only.
Figure 8:
Fields of taskDone{TaskIndex}:
• Username: username of logged in the operation system.
• time: Evaluation time for given method.
− start: Start time(yy.dd.hh.mm.ss)
− finish: End time (yy.dd.hh.mm.ss)
• task: Name of task given in: project.pipeline.{,}.task
• method: Tag of selected method given in: project.pipeline.{,}.method
• inputfiles{ImageIndex,ModalityIndex} (filenames are automatic made before
the method is executed)
− filename: header (*.hdr) or image (*.img) filename of input image, given a
modality. Expected image format: Analyze.
− Path: path of input image file of given modality.
− Info: (optional user information) Additional text information of image.
• outputfiles{ImageIndex,ModalityIndex}:(file names are automatic made)
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab − filename filename: header (*.hdr) or image (*.img) filename of output image,
given a modality. Expected image format: Analyze − path: path of input image file of given modality.
− Info: (optional user information) Additional text information of image.
• error{index}: cell-array of detected errors. Detected errors will clean up the field
and delete generated files. For a finished task the error field is empty.
• command: A commands given to the method
• configuration: Used configuration parameters: NOT READY/IMPLEMENTED.
• show{ImageIndex,ModalityIndex}:(file names are automatic made) Show
snapshot of the output data.
− filename: Images (*.bmp,*jpg or like) that gives a snapshot of the output
result from given method.
Note: The images are shown in the Result window of the pipeline program.
These images are not made by the pipeline program, but by the programmerfor the given method.
• filestatus{functionIndex}: File status of functions for a given method in the
current field of tasks{,}, which should be available to the pipeline program.
− path: Where function is found
− indexname: Field name
− filename: Name of file for the actual function.
− version: Found version label.
− exist: If file exist = 1 else = 0.
− os: Actual operation system: Linux, Windows and ect.
• userdata: A user defined entry for the actual method. Here additional information
can be stored as sub-entries the user needs or like to handle to the following tasksin the pipeline.
2.2.6 The handles field Handles to axes that give the possibility for a method to present the user forprogressing data or show images in the user interface of the pipeline program. If thepipeline program is executed without the user-interface the handles will not beavailable but empty.
Figure 9:
Fields of handles:
• h_logwin: Handle to the log window where log information are presented to the
• h_sitewin: Handle to the site window, where progress data and images of results
can be presented to the user.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab • h_mainfig: handle to the main figure of the user-interface, called mainGUI.
2.2.7 The info field In this field system information for the loaded pipeline and for the pipeline programare found.
Figure 10:
Fields of sysInfo:
• workdir: Directory where all files generated in a pipeline process are saved.
Note: A workdir is automatically created for each new project when the first taskin the pipeline is finished.
• systemdir: Directory where system files for the pipeline program are placed.
• Prj_filename: filename of project with extension ‘*.prj', saved in workdir.
• version: version number of the loaded pipeline set-up file.
• logfile: A log file exist for each project, and saved in the workdir with extension
‘*.log'. Log information are transferred to the log bar via project.handles.h_log if
the user-interface is used.
− tmp{index}: Temporary cell-array of log data is saved (added) to the log
filename when a method has finish.
− filename: name of log file with extension ‘*.log', saved in workdir.
3 Realising a pipeline
Before continuing into deeper levels of the pipeline program where the process, and
set-up of a pipeline in the pipeline program are explained, a summarize is given.
Preparation must be done to set-up a pipeline in the pipeline program. First definename and order of the tasks in a pipeline. Then within each task, name and program ofavailable methods is considered and methods tested. In section XX a preparationguideline can be found After preparation the next step is implementation of:- A set-up file to initialise a pipeline in the pipeline program.
- Wrappers that interface the pipeline program and a given method.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab - Wrappers that interface the pipeline program and a configurator for a given 3.1 Matlab and the pipeline program
Varargin, varargout 3.2 Connection diagram of the pipeline program
In the following the connectivity between user-interface, wrappers and files in thepipeline program are explained and all are referenced to Figure 11.
Robustness, transparency and generality are key words of the pipeline program.
- Robustness: The pipeline program is not affected by a crashing method.
- Transparency: All necessary information regards set-up, used settings and log information are automatic collected and are fully available within the pipelineprogram.
- Generality: The pipeline program is easy to use, independent of operation systems, and different type of executable realising a method can be used.
A wrapper is a function, which interface to/from the pipeline program. Basically twotype of wrappers exist in the pipeline program: the main_wrapper and thefunction_wrapper / configurator_wrapper.
The user-interface is handled by a function called mainGUI, section 2.1. Here handlesto the site and the log window are given, in this way progressing data, results and loginformation can be presented for the user.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab Figure 11:
3.3 A function_wrapper
The function wrapper is responsible to make the pipeline program general in a way sothat different methods can be implemented without confliction. It is typically calledby the main_wrapper, but also called by another function-wrapper. (Figure 11) Somemethods are very alike in this case the same function-wrapper can be used.
The only restrictions making a function wrapper are those for reading/writing theproject (see section 2.2.1), and that a number of input- output arguments are fixed: Where project is the data structure and the indices are refereed to the field of theproject project.pipeline.taskSetup{TaskIndex,MethodIndex} (section 2.2.4). Varargin holds anarbitrary number of user input arguments. These are only understood by the givenfunction.
The syntax for programming a function wrapper: 1) Retrieve information from the project such as input/output names which are automatic made, given project.taskDone{TaskIndex} 2) Prepare configuration of the method either using  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab project.pipeline.taskSetup{TaskIndex,MethodIndex}.configuratorfile or another field in the data structure.
3) Execute the method found in: project.pipeline.taskSetup{TaskIndex,MethodIndex}.function_nameORInsert code instead of a function. This require in the taskSetup{,} field in theproject that function_name=function_wrapper for the given method.
4) User data can be added to the project: project.taskDone{TaskIndex}.userdata, or the available standard fields in the project as given in section 2.2.5.
An example on how programming a function wrapper is shown in section 4.1.
3.4 A configurator_wrapper
Some methods can be configured with different settings. A configurator user-interfacecan be set-up, so the user has the possibility to change the settings, an example isshown section 2.1. This user-interface is not a part of the pipeline program, but can bestarted by the configurator_wrapper. New settings will be used next time the methodis executed if they are stored in the data structure of the projectproject.pipeline.taskSetup{TaskIndex,MethodIndex}.configuratorfile. Note: It shouldalways be possible to restore default settings through the configurator user-interface.
The syntax for making a configurator wrapper: 1) jk2) fg3) ghj4) fg An example on how to make a configurator wrapper is shown in section 4.2.
3.5 The main_wrapper
The robustness of the pipeline program is insured by the main_wrapper. The mainwrapper is executing programs by refereed to in the taskSetup by the indicesTaskIndex and MethodIndex. field found inproject.pipeline.taskSetup{TaskIndex,MethodIndex}, see section 2.2.4 for possiblefunctions and programs.
In a pipeline it is important to control process, so two methods can not be executed atthe time, the program is available, the input arguments exist etc. etc.
The processTo control the process in a pipeline  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 3.6 The mainGUI
3.7 Preparation: A task guideline
Tasks defining tasks and order of a pipeline…Task named ‘Others': typical functions linked to a menu…Task ‘end': Always last task: Information of actual pipeline…  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 3.8 Preparation: A method guideline
In the following a guideline is given when preparation which to full file the set-up fileis given NOTE: By default it is expected:
That a method reads and save input/output MR- PET-images in the Analyze That the orientation between MR and PET is the same.
That the function realising a method can be executed on Linux-, Unix- and window platforms.
Method: Tag name of method name to appear in the pipeline program.
Method_name: Full name of method.
Task: Name of the task the given method belongs to/ is realising.
Version label: A text string to identify the version number of the method in a Matlab
file.
Description: Information given via the information window to the user about selected
method and its special features.
Documentation: Article or where (e.g. www) the method is published.
Who did: Who did the method (group), city and year.
Require task index: Tasks to be done before method may run.
Function name: Name of the function that is realising the method. The function is
executed within the Matlab environment. The same name for the function and method
can be used.
Function syntax: Syntax for the function.
Input arguments: Description of the input parameters.
Output arguments: Description of the output parameters.
Code type: Matlab, executable or like to be executed from Matlab.
Configuration: (NOT READY…under construction) If exist, the configurator will be
used to set-up/initialise the method before execution. Further the configurator gives
the user the possibility to change the set-up through a user-interface.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab 4 How to…
4.1 How to write a function wrapper
In following are given an example of a function_wrapper called registrate_wrapper.
The function_wrapper is used in the co-registration task, and the same function isused for both the IIO and the ISP methods realising a co-registration task.
Note that in this example the filename and path of an AIR-file, containing the rigidtransformation matrix after co-registration, is saved in the sub-entry userdata of theproject. All other information to set up the IIO or the ISP is given in the structure ofthe project.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab function project=registrate_wrapper(project,TaskIndex,MethodIndex,varargin)% function_wrapper for the two registration methods: IIO_method and IPS_method%% Input:% project: Structure for actual PVE correction process% TaskIndex: Index in the project-structure for actual task% MethodIndex: Index in the project-structure for actual method% varargin: Extra input arguments, not used%% Outout:% project: structure for actual PVE correction process% % T. Dyrby, 170303, NRU%%SW version: 170303TD % _ Check if registration function is returning data.
if( isstruct(project)) % No project structure is given!! returnend% % _ Add to project: AIR file where to store registration matrixname=project.taskDone{TaskIndex}.inputfiles{1,1}.name;[tmp,AIRfile,extAIR,tmp]=fileparts(name); % _ Init matrix to Coreg function. Defined in header of 'Coreg.m'name=project.taskDone{TaskIndex}.inputfiles{1,1}.name;path=project.taskDone{TaskIndex}.inputfiles{1,1}.path;files.STD=fullfile('',path,name); % _ Possible to load a default registration matrixfiles.A{1}=eye(4,4);% Initial registration matrix % _ Get handle to registration method only used for UIWAITh_registration=feval(project.pipeline.taskSetup{TaskIndex,MethodIndex}.function_name);set(h_registration,'Visible','off'); % _ Call registration methodparent=project.handles.h_mainfig;% Parent handleReturnFcn=project.pipeline.taskSetup{TaskIndex,MethodIndex}.function_wrapper;% Where to return afterwardsvisualizer=project.pipeline.taskSetup{7,3}.function_name;% Which method to browse:Could be in a configurator.
% Wait until function (figure object) return (is finished)uiwait(h_registration)  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab implemented
registrate_wrapper, which belong to the co-registration task. The same
function_wrapper is used for the two methods realising the co-registration:
Coreg(…) and the Registrate(.).
4.2 How to write a configurator wrapper
4.3 How to make a set-up file
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab function tasks=setupPVELab;% Setting up the user defined information about% the different methods and which task it belongs.
%% The information is stored in the project structure:% project.pipeline.tasks{TaskIndex,MethodIndex}%% %SW version: 170303TD, T. Dyrby, 120303, NRU % % _TASK: FILELOADTaskIndex=1;% FileLoadMethodIndex=1;% FileLoadtmpTask.task='FileLoad'; % % TASK: RegistrationTaskIndex=2;% RegistrationMethodIndex=1;% IIOtmpTask.task='Registration';  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab Figure 13. Example of setting up a method for a given task. Here is shown: the
first task:'Fileload' with one metod:'fileload'. And the second task:'Registration'
with two methods:'IIO' and ‘IPS'.
4.4 How to make a method guideline
Method: IIO
Method name: Interactive Image Overlay
Version label: version
Description: Interactive Image Overlay (IIO) a manual method for co-registration (6-
DOF) high resolutions MR scan and a low resolution PET/SPECT brain scan.
Who did: NRU, Copenhagen, 2000-2003
Require task index: 1 (Files load)
Function name: coreg
Filename: structure containing a high resolution MR scan and one or more lowresolutions PET/SPECT given in a cell array. All in the Analyze image format.
Example of Filename structure:Filename.RES: Name and path for high resolution MR scanFilename.STD{}: Name and path of one or more low resolution PET/SEPCTscansFilename.A: If exist, a given co-registration matrix is used asinitiated co-registrationFilename.AIR: If filename exist, co-registration matrix is saved inair-format.
ParentHandle: Figure handle to parent/main figure so it can be found whenreturning fra Coreg.
Example: ParentHandle=h_mainfig ReturnFucn: Function to call exiting the Coreg program.
Example when exiting from Coreg in Matlab: feval(ReturnFucn,'ReturningData',ParentHandle,Aall); where- ReturnFucn='checkOut'- ‘ReturningData': Parameter telling that co-registration matrix is returned by Coreg.
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab - Aall cell array contain co-registration matrix.
Output parameters: None, is using the ‘ReturnFucn'
Code type: Matlab code, (figure object)
4.5 File formats and directories
The file name of the first loaded modality given in project.pipeline.modality{1}…In general: Use prefix given in the set-up added to the filebName of project.pipeline.inputfile.name{1,1} or/and with an extension.
Files are only made if workdir exist, which is generated when the first task is done oran existing project file is loaded.
• Workdir: unique project directory
o Path: sub directory to project.pipeline.inputfile.path{1,1} with prefix ‘WORK'. If directory exist an index is added to the prefix,‘WORK_index'.
• Log file: Text file w. log information
o Filename: project.pipeline.inputfile.filename{1,1} with extension o Path: workdir • Project file: Matlab data file.
o Filename: project.pipeline.inputfile.filename{1,1} with extension o Path: workdir • Set-up file: A Matlab m-file w. a function setting up a pipeline
o Filename: project.pipeline.inputfile.filename{1,1} with extension o Path: Within the Matlab path • Macro file: Mathlab data file, w. user defined set-up of a pipeline
o Filename: project.pipeline.inputfile.filename{1,1} with extension o Path: user defined • Output files: Generated image files in Analyze format
o Filename: project.pipeline.inputfile.filename{1,1} add prefix given in the set-up file.
o Path: workdir • Configurator file: Text file containing in parameters for a method
o Filename: user definedo Path: Within the Matlab path.
• Function_wrappers, configurator_wrapper: Interfacing the pipeline
o Filename: Given in the set-up file.
o Path: Within the Matlab path • Method program files: Realising a task
 The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03 The pipeline program for PVElab o Filename: Given in the set-up file.
o Path: Within the Matlab path  The PVEout project within the European 5th framework, NRU T. Dyrby, 06-07-03

Source: http://pveout.ibb.cnr.it/internal/download/AppendixB.pdf