Documentation/4.1/Modules/TrainModel - Revision history

UpdateBot: Prepend documentation/versioncheck template. See http://na-mic.org/Mantis/view.php?id=2887

2013-06-14T07:28:33Z

Prepend documentation/versioncheck template. See http://na-mic.org/Mantis/view.php?id=2887

Mscully: Expanded use cases.

2012-06-22T22:23:19Z

Expanded use cases.

Mscully at 19:55, 22 June 2012

2012-06-22T19:55:27Z

Mscully: Added panel info and screenshot

2012-06-22T19:45:39Z

Added panel info and screenshot

Mscully: Initial version

2012-06-21T19:53:46Z

Initial version

New page

{{documentation/{{documentation/version}}/module-header}}



{{documentation/{{documentation/version}}/module-section|Introduction and Acknowledgements}}
{{documentation/{{documentation/version}}/module-introduction-start|{{documentation/modulename}}}}
{{documentation/{{documentation/version}}/module-introduction-row}}
Extension: [[Documentation/{{documentation/version}}/Extensions/LesionSegmentation|LesionSegmentation]]<br>
Acknowledgments:
This work is part of the National Alliance for Medical Image Computing (NAMIC), funded by the National Institutes of Health through the NIH Roadmap for Medical Research, Grant U54 EB005149.<br>
Author: Mark Scully ({{collaborator|name|Biomedical Mining}})<br>
Contact: Mark Scully, <email>mark@biomedicalmining.com</email><br>
{{documentation/{{documentation/version}}/module-introduction-row}}
{{documentation/{{documentation/version}}/module-introduction-logo-gallery
|{{collaborator|logo|namic}}|{{collaborator|longname|namic}}
}}

{{documentation/{{documentation/version}}/module-introduction-end}}


{{documentation/{{documentation/version}}/module-section|Module Description}}
This module is used to train new segmentation models for white matter lesion segmentation. In order to use this tool your data must include a T1, T2, FLAIR, brain mask, and expert lesion segmentation for each subject. All data must be preprocessed including intra-subject co-registration, AC-PC alignment, bias correction, consistent spacing between sequences, and brain mask creation.


{{documentation/{{documentation/version}}/module-section|Use Cases}}
# Training a new model.
In order to train a new model you must first have preprocessed data on a number of subjects. The data required includes T1, T2, FLAIR, brain masks, and lesion masks. All data must be preprocessed including intra-subject co-registration, AC-PC alignment, bias correction, consistent spacing between sequences, and brain mask creation. Subjects do not need to be registered to each other. A model can be created on a single subject, but greater than 6 is recommended and between 10 and 15 is best. The more subjets included in the model the slower model creation will be and the slower segmentation using that model will be. However, models using more subjects will almost always be more accurate.

Navigate to Modules->Segmentation->LesionSegmentation->TrainModel. The TrainModel panel looks like:
(Image of TrainModel panel to go here.)

The required inputs are


{{documentation/{{documentation/version}}/module-section|Tutorials}}
Coming soon!

{{documentation/{{documentation/version}}/module-section|Panels and their use}}


{{documentation/{{documentation/version}}/module-section|Similar Modules}}
*[[Documentation/{{documentation/version}}/Modules/PredictLesions| LesionSegmentation->PredictLesions]]

{{documentation/{{documentation/version}}/module-section|References}}
* Scully M, Anderson B, Lane T, Gasparovic C, Magnotta V, Sibbitt W, Roldan C, Kikinis R and Bockholt HJ (2010) An automated method for segmenting white matter lesions through multi-level morphometric feature classification with application to lupus. Front. Hum. Neurosci. doi:10.3389/fnhum.2010.00027
http://frontiersin.org/neuroscience/humanneuroscience/paper/10.3389/fnhum.2010.00027/


{{documentation/{{documentation/version}}/module-section|Information for Developers}}
{{documentation/{{documentation/version}}/module-developerinfo}}


{{documentation/{{documentation/version}}/module-footer}}
[[Category:Documentation/{{documentation/version}}/Modules/Segmentation]]

← Older revision		Revision as of 07:28, 14 June 2013
Line 1:		Line 1:
		+	<noinclude>{{documentation/versioncheck}}</noinclude>
	<!-- ---------------------------- -->		<!-- ---------------------------- -->
	{{documentation/{{documentation/version}}/module-header}}		{{documentation/{{documentation/version}}/module-header}}

@@ Line 49: / Line 49: @@
 ** <span style="color:green">'''Highest Quality Images Index'''</span> [<span style="color:orange">----inputIndexOfBestImages</span>] : The index in the list of images that represents the best T1, T2, and FLAIR images.  These images will be used as the standard that the other images are intensity standardized to.  It defaults to the first image in the list. (This number is 1-indexed) ''Default value: 1''
 * <span style="color:blue">'''''Advanced Options'''''</span>: Advanced input parameters
-** <span style="color:green">'''Percent NonLesion'''</span> [<span style="color:orange">----inputPercentNonLesion</span>] : What percent of the nonlesion voxels should be used for training.  Higher numbers results in larger model files and potentially slower runtimes.  Default value is 5. ''Default value: 5''
+** <span style="color:green">'''Percent NonLesion'''</span> [<span style="color:orange">----inputPercentNonLesion</span>] : What percent of the nonlesion voxels should be used for training.  Higher numbers results in larger model files and potentially slower runtimes.  ''Default value: 5''
 * <span style="color:blue">'''''Output Options'''''</span>: Output Options
 ** <span style="color:green">'''Output Model Filename'''</span> [<span style="color:orange">----outputModel</span>] : Required: Filename to save the generated model to.

@@ Line 25: / Line 25: @@
 <!-- ---------------------------- -->
 {{documentation/{{documentation/version}}/module-section|Use Cases}}
-# Training a new model.
+*Use Case: Segmenting white matter lesions in a disorder for which there is no current lesion segmentation model
-In order to train a new model you must first have preprocessed data on a number of subjects.  The data required includes T1, T2, FLAIR, brain masks, and lesion masks.  All data must be preprocessed including intra-subject co-registration, AC-PC alignment, bias correction, consistent spacing between sequences, and brain mask creation.  Subjects do not need to be registered to each other.  A model can be created on a single subject, but greater than 6 is recommended and between 10 and 15 is best.  The more subjets included in the model the slower model creation will be and the slower segmentation using that model will be.  However, models using more subjects will almost always be more accurate.
+If a lesion segmentation model already exists for the disorder of interest you most likely do not need to train a new one.  However, if a model does not exist or the existing model(s) give poor performance on your data you can use this module to train a new model, and then use that model to automatically segment white matter lesions.
-Navigate to Modules->Segmentation->LesionSegmentation->TrainModel.  The TrainModel panel looks like:
+In order to train a new model you must first have preprocessed data on a number of subjects.  The data required includes T1, T2, FLAIR, brain masks, and lesion masks.  All data must be preprocessed including intra-subject co-registration, AC-PC alignment, bias correction, consistent spacing between sequences (i.e all 1mm isotropic), and brain mask creation.  Subjects do not need to be registered to each other.  A model can be created on a single subject, but greater than 6 is recommended and between 10 and 15 is best.  The more subjets included in the model the slower model creation will be and the slower segmentation using that model will be.  However, models using more subjects will almost always be more accurate. When selecting subjects to use for the model try to pick the largest variety that you can.  This means, if appropriate for your dataset, multiple ages, multiple brain sizes, multiple lesion loads, a balance between genders, multiple stages of disease progression, etc.  Then identify the set of images for a single subject that has the cleanest and highest contrast scans and use that subject's scans as the first set of inputs.  They will be used as the reference for intensity standardization.  You will need to use those images as the reference scans when predicting as well.
 <!-- ---------------------------- -->

@@ Line 22: / Line 22: @@
 {{documentation/{{documentation/version}}/module-section|Module Description}}
 This module is used to train new segmentation models for white matter lesion segmentation.  In order to use this tool your data must include a T1, T2, FLAIR, brain mask, and expert lesion segmentation for each subject.  All data must be preprocessed including intra-subject co-registration, AC-PC alignment, bias correction, consistent spacing between sequences, and brain mask creation.
 <!-- ---------------------------- -->
@@ Line 39: / Line 38: @@
 <!-- ---------------------------- -->
 {{documentation/{{documentation/version}}/module-section|Panels and their use}}
 <!-- ---------------------------- -->