Difference between revisions of "RobustStatisticsSegmentation"

From NAMIC Wiki
Jump to: navigation, search
 
(19 intermediate revisions by the same user not shown)
Line 2: Line 2:
 
= Robust Statistics Based Segmentation =
 
= Robust Statistics Based Segmentation =
  
−
== Description ==
+
 
 +
{|
 +
|[[Image:RssPanel.png|thumb|280px|RSS module panel]]
 +
|[[Image:RssVentricle.png|thumb|400px|RSS result for Brain MR image]]
 +
|}
 +
 
 +
== General Information ==
 +
=== Description ===
  
 
Given a few initial seeds, we use the robust statistics to construct a feature image out of the original image, and the image segmentation is carried out in the feature image. To do so, we first compute the certain robust statistics, such as the median absolute deviation(MAD) and the interquartile range (IQ), from the seed points. Further, a multi-dimentional probability distribution function is computed from the feature seeds. At each point in the image, the active contour is driven by the probabilities of this pixel belonging to each category indicated by the seeds.
 
Given a few initial seeds, we use the robust statistics to construct a feature image out of the original image, and the image segmentation is carried out in the feature image. To do so, we first compute the certain robust statistics, such as the median absolute deviation(MAD) and the interquartile range (IQ), from the seed points. Further, a multi-dimentional probability distribution function is computed from the feature seeds. At each point in the image, the active contour is driven by the probabilities of this pixel belonging to each category indicated by the seeds.
  
−
==How to get the module==
+
===Module Type & Category===
−
===Slicer 3.6===
+
 
−
RSS is part of the download, listed under the category "Segmentation"
+
CLI module in 3D Slicer, in the category of Segmentation
  
−
===Slicer 3.5===
+
===Authors, Collaborators & Contact===
−
# start nightly built Slicer 3.5 (Most recently tested is Slicer3-3.5-alpha-2010-01-11-linux-x86_64)
+
* Yi Gao (Author): Georgia Tech
−
# open View->Extension Manager
+
* Allen Tannenbaum (Author): Georgia Tech
−
# check "Find & Install", click "Next"
+
* Ron Kikinis (Author): BWH
−
# in the list, select "RobustStatisticsSegmentor", then "Download & Install", and "Next"
+
* Contact: Yi Gao, yi.gao@gatech.edu
−
# Slicer will ask to restart, confirm.
 
−
# after restarts, the module is in Module category: "Segmentation-->Robust Statistics Segmentation"
 
  
 
== Usage ==
 
== Usage ==
−
 
 
[[Image:RobustStatisticsSegmentation_usage1.png | Module parameters | 300px]]  --->   
 
[[Image:RobustStatisticsSegmentation_usage1.png | Module parameters | 300px]]  --->   
 
[[Image:RobustStatisticsSegmentation_usage2.png | Adv Module parameters | 300px]]  --->   
 
[[Image:RobustStatisticsSegmentation_usage2.png | Adv Module parameters | 300px]]  --->   
 
[[Image:RobustStatisticsSegmentation_usage3.png | Adv Module parameters | 300px]]
 
[[Image:RobustStatisticsSegmentation_usage3.png | Adv Module parameters | 300px]]
  
−
=== General guideline ===
+
===Use Cases, Examples===
 +
 
 +
* This module is a general purpose segmenter.
 +
* Link to examples of the module's use
 +
 +
<gallery widths="400px" heights="300px" perrow="2">
 +
Image:RSSkidneyL.png|Left kidney, CT image(512*512*204). IO time: 10sec, module running time: 12sec (Intel 3.0GHz), <br>Approximate volume: 200 ml, <br>Intensity homogeneity: 0.1, <br>Boundary smoothness: 0.5
 +
Image:RSS_rkidney.png|Right kidney, CT image(512*512*204). IO time: 10sec, module running time: 15sec (Intel 3.0GHz), <br>Approximate volume: 200 ml, <br>Intensity homogeneity: 0.1, <br>Boundary smoothness: 0.5
 +
Image:RSS_tumor.png|Brain tumor, MR image(256*256*123). IO time: 3sec, Module running time: 2.5sec (Intel 3.0GHz), <br>Approximate volume: 50 ml, <br>Intensity homogeneity: 0.1,  <br>Boundary smoothness: 0.2
 +
Image:RssVentricle.png|Ventricle, MR image(256*256*124). IO time: 3sec, Module running time: 2.5sec (Intel 3.0GHz), <br>Approximate volume: 30 ml, <br>Intensity homogeneity: 0.02,  <br>Boundary smoothness: 0.0
 +
Image:RSS-aorta.png|Aorta, CT image(512*512*204). IO time: 10sec, module running time: 12sec (Intel 3.0GHz), <br>Approximate volume: 60 ml, <br>Intensity homogeneity: 1.0,  <br>Boundary smoothness: 0.0
 +
Image:RSSMandible.png|Head, CT image(512*512*460) from http://pubimage.hcuge.ch:8080/ MANIX data set. IO time: 16sec, module running time: 160sec (Intel 3.0GHz), <br>Approximate volume: 100 ml, <br>Intensity homogeneity: 0.5,  <br>Boundary smoothness: 0.0
 +
</gallery>
 +
 
 +
===Tutorials===
 +
 
 
* First run:
 
* First run:
−
# Give a rough estimate of the object volume and put fiducial points (preferred to be 2 or more) in the object.
+
# Give a rough estimate of the object volume and use the editing module to paint several non-zero labels, called seeds in the following, in the object.
 
# Run the module using the default parameters.
 
# Run the module using the default parameters.
  
 
* Note:
 
* Note:
 
# The Approximate volume is just a rough upper limit for the volume. It should be at least the size of the object. This is because when the volume reaches that, the program must stop. However, other criteria may stop the algorithm before the volume reaches this value.
 
# The Approximate volume is just a rough upper limit for the volume. It should be at least the size of the object. This is because when the volume reaches that, the program must stop. However, other criteria may stop the algorithm before the volume reaches this value.
−
# The fiducial points can be thrown into the object. What I do is I just add two fiducial points and move them into the object within one slice.
+
# The positions of the seeds have to be in the object, preferably close to center.
  
 
* Troubleshooting
 
* Troubleshooting
Line 46: Line 65:
 
** '''Some regions are missed, at the same time leakages to some other regions'''. Try (either one)
 
** '''Some regions are missed, at the same time leakages to some other regions'''. Try (either one)
 
*** Increase "Intensity homogeneity"
 
*** Increase "Intensity homogeneity"
−
*** Add another fiducial point
+
*** Add some other seeds
  
  
−
=== Multiple-value label map handeling ===
 
  
−
The parameter "Output Label Value"(OLV) is for user to assign the output label value. Moreover, it is also used in cases where the user provided label map contains more than one label value. In general, the module handles three cases:
+
* Data Set http://www.spl.harvard.edu/publications/item/view/1180 Tumorbase.zip at page bottom, in the zip file, case3/grayscale.nrrd
  
−
# user provided label map contains only one label value, L. In this case, the output label value will be set to OLV, not matter what value L takes.
+
1. Draw label map in Editing module
−
# user provided label map contains multiple label values, one of which matches OLV. Then only that label will be effective and all the others are discarded. The output will have label value OLV too.
+
In the editing module, select any drawing tool, for example the "Paint" tool circled in red. Draw some strokes in the target in one of the 2D views. In this example, we only freely draw the "S" shaped label circled in green. The label image is available at: [[File:RSS_Grayscale-label.nrrd|Label image]]
−
# user provided label map contains multiple label values, but none matches OLV. Then all the non-zero labels will be considered as a single label value and then come back to the case 1 above.
 
  
−
== Testing ==
+
[[Image:RSS_editStep.png | Drawing label map for RSS| 700px]]
 +
2. Run RSS.
 +
To run the RSS module. The parameters which may affect the segmentation results include: ''Approximate volume'', ''Intensity homogeneity'', ''Boundary smoothness'', ''Max running time''. The particular setting of the parameters for this example case is shown in the screenshot, in the red box. Some guidelines for adjusting them are given in the '''Troubleshooting''' section above, whereas their general roles in influencing the results are given in the '''Quick Tour of Features and Use''' section below. The sensitivity of the parameters are not quantitatively evaluated. In some easy cases, the algorithm is rather robust to the parameters. However for some objects with inhomogeneous intensity as well as irregular shape, the parameters may need to be carefully tuned.
  
−
Several tests are conducted and shown here along with the parameters used to get the results.
+
[[Image:RSS_run.png | Set up RSS module| 700px]]
  
−
== Testing case: left kidney ==
+
* Multiple-value label map handling
  
−
* Data set: http://wiki.na-mic.org/Wiki/images/8/8d/Patient1.tar.gz
+
The parameter "Output Label Value"(OLV) is for user to assign the output label value. Moreover, when user provided label map contains several labels, the target corresponding to the OLV is the one get segmented. More specifically, there are three difference situations:
−
* Approximate volume: 200 mL
 
−
* Intensity homogeneity: 0.1
 
−
* Boundary smoothness: 0.5
 
  
−
[[Image:RobustStatisticsSegmentation_LeftKidney.png | Segmentation of left kidney | 800px]]
+
# user provided label map contains only one label value, L. In this case, the output label value will be set to OLV, not matter what value L takes.
 +
# user provided label map contains multiple label values, one of which matches OLV. Then only that label will be effective and all the others are discarded. The output will have label value OLV too.
 +
# user provided label map contains multiple label values, but none matches OLV. Then all the non-zero labels will be considered as a single label value and then come back to the case 1 above.
  
 +
In the current version, regardless of the number of different labels appearing in the label map, only one object is extracted from the image. The on-going work extends this to extracting multiple objects simultaneously.
  
−
== Testing case: right kidney ==
+
===Quick Tour of Features and Use===
  
−
* Data set: http://wiki.na-mic.org/Wiki/images/8/8d/Patient1.tar.gz
+
A list panels in the interface, their features, what they mean, and how to use them. For instance:
−
* Approximate volume: 200 mL
 
−
* Intensity homogeneity: 0.1
 
−
* Boundary smoothness: 0.5
 
  
−
In the figure below we show the position of the two seeds. They are both in the cortex region. And it can be seen in the segmentation result, the renal cortex got segmented and the pelvis is not.  
+
{|||
 +
* '''Parameters panel:'''
 +
** '''Approximate volume:''' The estimated upper limit of the target volume. The resulting volume will be less or equal than this value.
 +
** '''Intensity homogeneity:''' If the target contains homogeneous intensity, then give a close-to-1 value here.
 +
** '''Boundary smoothness:''' Larger value will result in smoother boundary and a more spherical looking result.
 +
** '''Output Label Value:''' Defined the label value of the output. Also refer to the "Multiple-value label map handling" above.
 +
** '''Max running time:''' The upper limit for program running time.
 +
* '''IO panel:'''
 +
** '''Input Image:''' The image to be segmented.
 +
** '''Label Image:''' The label map providing initial seeds.
 +
* '''Output Volume:''' The output volumetric image.
 +
|[[Image:RssPanel.png|thumb|280px|User Interface]]
 +
|}
  
−
[[Image:RobustStatisticsSegmentation_RightKidney.png | Segmentation of right kidney | 800px]]
+
== Development ==
  
 +
===Source code & documentation===
  
−
== Testing case: spleen ==
+
The source code of the algorithm is in 3D Slicer source directory tree. Under Slicer3/Applications/CLI/RobustStatisticsSegmenter/
  
−
* Data set: http://wiki.na-mic.org/Wiki/images/8/8d/Patient1.tar.gz
+
Moreover, the code above is a self-contained implementation for the 3D Sparse Filed Level Set method. The level set evolution is implemented in the base class. Specific flow, such as Chan-Vese, Geodesic Active Contour, or the Robust Statistics based flow, can be implemented in the derived class. Therefore, it can be reused for other types of flow with not much work.
−
* Approximate volume: 300 mL
+
−
* Intensity homogeneity: 1
 
−
* Boundary smoothness: 0.3
 
  
−
In the figure below we show the position of the two seeds.  
+
= Multi-Object RSS (MORSS)=
 +
{|
 +
|[[Image:RSS_MultiObjSeg1.png|thumb|400px|RSS result for abdominal CT image]]
 +
|}
  
−
[[Image:RobustStatisticsSegmentation_Spleen.png | Segmentation of spleen | 800px]]
 
  
 +
== Description ==
  
−
== Testing case: liver ==
+
In contrast to the above single object RSS, in MORSS, the user can use different label (color) to draw the seed label image. After that, the algorithm detect the number of objects and evolve multiple contours simultaneously.
  
−
* Data set: http://wiki.na-mic.org/Wiki/images/8/8d/Patient1.tar.gz
+
==Module Type & Category==
−
* Approximate volume: 2000 mL
 
−
* Intensity homogeneity: 0.8
 
−
* Boundary smoothness: 0.8
 
−
* Max running time (min): 20
 
  
−
In the figure below we show the position of the three seeds.
+
CLI module in 3D Slicer Extension, in the category of Segmentation. Need to install from Extension Manager.
−
 
 
−
[[Image:RobustStatisticsSegmentation_Liver.png | Segmentation of spleen | 800px]]
 
−
 
 
−
== Testing case: kidneys, spleen, liver ==
 
−
 
 
−
* Data set: http://wiki.na-mic.org/Wiki/images/8/8d/Patient1.tar.gz
 
−
 
 
−
User can chose output label value. If different labels are used for different organs, then user can use the "Modules->Filtering->Image Label Combine" to combine the different label maps into one. This is shown in the following figure.
 
−
 
 
−
[[Image:RobustStatisticsSegmentation_LiverSpleenKidney.png | Segmentation of kidneys, spleen, and liver | 800px]]
 
−
 
 
−
== Testing case: tumor ==
 
−
 
 
−
* Data set: http://wiki.na-mic.org/Wiki/images/0/0f/MayExperiments.zip (DiffusionEditorBaselineNode.nrrd)
 
−
* Approximate volume: 30 mL
 
−
* Intensity homogeneity: 0.1
 
−
 
 
−
[[Image:RobustStatisticsSegmentation_MayExp_diffusionEditorBaseline.png | Segmentation of brain tumor | 800px]]
 
−
 
 
−
 
 
−
== Testing case: tumor ==
 
−
* Data set: http://www.spl.harvard.edu/publications/bitstream/download/4217 (case3/grayscale.nrrd)
 
−
* Approximate volume: 30 mL
 
−
* Intensity homogeneity: 0.4
 
−
 
 
−
[[Image:RobustStatisticsSegmentation_TumerBase_3.png | Segmentation of brain tumor | 800px]]
 
−
 
 
−
== Testing case: tumor ==
 
−
* Data set: http://www.spl.harvard.edu/publications/bitstream/download/4217 (case5/grayscale.nrrd)
 
−
* Approximate volume: 3 mL
 
−
* Intensity homogeneity: 0.5
 
−
* Boundary smoothness: 0.4
 
−
 
 
−
This is a difficult case. All the other methods tested either captures only the middle dark spot, or leaks out of the tumor. But the method here nicely capture both the dark core and the bright shell, without leaking out to regions with intensities in the middle.
 
−
 
 
−
[[Image:RobustStatisticsSegmentation_TumerBase_5.png | Segmentation of brain tumor | 800px]]
 
  
 +
== Usage ==
 +
The basic steps of MORSS is the same as RSS. In fact, if only one type of seed is given (one non-zero label), the MORSS is in fact the single object RSS above.
  
−
== Testing case: vervet brain ==
+
Moreover, since the number of labels are determined after the algorithm starts, on the user interface, it is difficult to adjust the parameters for each of the target object. Therefore in the current version, the same parameters set on the GUI apply to all the contours. This, however, may not be desirable in all cases. Similar situation is for the expected volume. So on the GUI, we didn't ask for the expected volume. Instead, we used a more engineering term "number of iterations", which also applies to all the contours.
−
* Data set: http://www.na-mic.org/Wiki/index.php/Vervet_MRI_registration (http://www.bsl.ece.vt.edu/data/vervet_atlas/VPA-10-1.1.zip  Vervet_T1_Template_WholeHead.nii)
 
−
* Approximate volume: 15 mL
 
−
* Intensity homogeneity: 0.5
 
  
−
Previous tests all run with two fiducial points, and running time are all less than 10 seconds. This case has 6 points and runs for 15min to get the result shown.
+
Looking forward, if the algorithm is implemented as a loadable module, then we can have more flexible control over the GUI. In particular, we can create a several-step-module. In the first step the user provide the input and label image. By clicking "Process label image", the number of different labels are learned from the label image. In addition, the GUI of the second step is adaptively adjusted.
  
−
[[Image:RobustStatisticsSegmentation_Vervet.png | Segmentation of vervet brain | 800px]]
+
===Use Cases, Examples===
 +
====Meningioma====
 +
The testing image of this example can be downloaded at: [[File:RSS_Grayscale.nrrd|example image]]
 +
The label image is at: [[File:MORSS_Grayscale-multilabel.nrrd|multiple label seed image]]
  
−
== Key Investigators ==
+
====Cardix====
 +
The testing image of this example is sub-sampled from the CARDIX data set from Orisix data set and can be downloaded at: [[File:RSS_CARDIX-sub2.nrrd|example image]]
 +
The label image is at: [[File:RSS_sub2_cardix_LVlabel.nrrd|left ventricle label seed image]]
 +
Parameters are: Approx vol limit 300mL, IH = 0.01, BS=0.1, Threshold= default(= 2)
  
−
Georgia Tech: Yi Gao and Allen Tannenbaum
+
The multi label image is at: [[File:MORSS_CARDIX-sub2-multilabel.nrrd|multi-label seed image]]
 +
Parameters are: IH = 0.5, BS=0.6, number of iteration 600.
  
−
BWH: Katie Hayes, Andriy Fedorov, and Ron Kikinis
+
=Acknowledgment=
 +
This  work  was  supported  in  part  by  grants  from  NSF, AFOSR, ARO, as well as by a grant from
 +
NIH (NAC P41 RR-13218) through Brigham and Women’s
 +
Hospital. An NSF Fellowship supported part of the work.
  
 
== Publications ==
 
== Publications ==
  
 
The method is based on:  
 
The method is based on:  
−
* Y. Gao, A. Tannenbaum, R. Kikinis, ''Simultaneous Multi-Object Segmentation using Local Robust Statistics and Contour Interaction'', MICCAI 2010, Medical Computer Vision, Workshop on [[File:Rss.pdf|pdf]]
+
* Yi Gao, Ron Kikinis, Sylvain Bouix, Martha Shenton, Allen Tannenbaum, ''A 3D Interactive Multi-object Segmentation Tool using Local Robust Statistics Driven Active Contours'', Medical Image Analysis, 2012, http://dx.doi.org/10.1016/j.media.2012.06.002
  
 
==Rons==
 
==Rons==
 
[[File:RobustSeg.zip|example case]]
 
[[File:RobustSeg.zip|example case]]

Latest revision as of 16:25, 24 October 2012

Home < RobustStatisticsSegmentation

Robust Statistics Based Segmentation

RSS module panel
RSS result for Brain MR image

General Information

Description

Given a few initial seeds, we use the robust statistics to construct a feature image out of the original image, and the image segmentation is carried out in the feature image. To do so, we first compute the certain robust statistics, such as the median absolute deviation(MAD) and the interquartile range (IQ), from the seed points. Further, a multi-dimentional probability distribution function is computed from the feature seeds. At each point in the image, the active contour is driven by the probabilities of this pixel belonging to each category indicated by the seeds.

Module Type & Category

CLI module in 3D Slicer, in the category of Segmentation

Authors, Collaborators & Contact

  • Yi Gao (Author): Georgia Tech
  • Allen Tannenbaum (Author): Georgia Tech
  • Ron Kikinis (Author): BWH
  • Contact: Yi Gao, yi.gao@gatech.edu

Usage

Module parameters ---> Adv Module parameters ---> Adv Module parameters

Use Cases, Examples

  • This module is a general purpose segmenter.
  • Link to examples of the module's use

  • Left kidney, CT image(512*512*204). IO time: 10sec, module running time: 12sec (Intel 3.0GHz),
    Approximate volume: 200 ml,
    Intensity homogeneity: 0.1,
    Boundary smoothness: 0.5

  • Right kidney, CT image(512*512*204). IO time: 10sec, module running time: 15sec (Intel 3.0GHz),
    Approximate volume: 200 ml,
    Intensity homogeneity: 0.1,
    Boundary smoothness: 0.5

  • Brain tumor, MR image(256*256*123). IO time: 3sec, Module running time: 2.5sec (Intel 3.0GHz),
    Approximate volume: 50 ml,
    Intensity homogeneity: 0.1,
    Boundary smoothness: 0.2

  • Ventricle, MR image(256*256*124). IO time: 3sec, Module running time: 2.5sec (Intel 3.0GHz),
    Approximate volume: 30 ml,
    Intensity homogeneity: 0.02,
    Boundary smoothness: 0.0

  • Aorta, CT image(512*512*204). IO time: 10sec, module running time: 12sec (Intel 3.0GHz),
    Approximate volume: 60 ml,
    Intensity homogeneity: 1.0,
    Boundary smoothness: 0.0

  • Head, CT image(512*512*460) from http://pubimage.hcuge.ch:8080/ MANIX data set. IO time: 16sec, module running time: 160sec (Intel 3.0GHz),
    Approximate volume: 100 ml,
    Intensity homogeneity: 0.5,
    Boundary smoothness: 0.0

Tutorials

  • First run:
  1. Give a rough estimate of the object volume and use the editing module to paint several non-zero labels, called seeds in the following, in the object.
  2. Run the module using the default parameters.
  • Note:
  1. The Approximate volume is just a rough upper limit for the volume. It should be at least the size of the object. This is because when the volume reaches that, the program must stop. However, other criteria may stop the algorithm before the volume reaches this value.
  2. The positions of the seeds have to be in the object, preferably close to center.
  • Troubleshooting
    • Surface is too rough. Try:
      • Increase "Boundary smoothness"
    • Leakage into thin/narrow regions. Try:
      • Increase "Boundary smoothness"
    • leakage into similar (but still different) intensity regions (which is not necessarily thin), Try:
      • Increase "Intensity homogeneity"
    • Some regions are missed: Try (either one):
      • Increase "Max volume"
      • Decrease"Intensity homogeneity"
      • Decrease "Boundary smoothness"
    • Some regions are missed, at the same time leakages to some other regions. Try (either one)
      • Increase "Intensity homogeneity"
      • Add some other seeds


1. Draw label map in Editing module

In the editing module, select any drawing tool, for example the "Paint" tool circled in red. Draw some strokes in the target in one of the 2D views. In this example, we only freely draw the "S" shaped label circled in green. The label image is available at: File:RSS Grayscale-label.nrrd

Drawing label map for RSS 

2. Run RSS.

To run the RSS module. The parameters which may affect the segmentation results include: Approximate volume, Intensity homogeneity, Boundary smoothness, Max running time. The particular setting of the parameters for this example case is shown in the screenshot, in the red box. Some guidelines for adjusting them are given in the Troubleshooting section above, whereas their general roles in influencing the results are given in the Quick Tour of Features and Use section below. The sensitivity of the parameters are not quantitatively evaluated. In some easy cases, the algorithm is rather robust to the parameters. However for some objects with inhomogeneous intensity as well as irregular shape, the parameters may need to be carefully tuned.

Set up RSS module

  • Multiple-value label map handling

The parameter "Output Label Value"(OLV) is for user to assign the output label value. Moreover, when user provided label map contains several labels, the target corresponding to the OLV is the one get segmented. More specifically, there are three difference situations:

  1. user provided label map contains only one label value, L. In this case, the output label value will be set to OLV, not matter what value L takes.
  2. user provided label map contains multiple label values, one of which matches OLV. Then only that label will be effective and all the others are discarded. The output will have label value OLV too.
  3. user provided label map contains multiple label values, but none matches OLV. Then all the non-zero labels will be considered as a single label value and then come back to the case 1 above.

In the current version, regardless of the number of different labels appearing in the label map, only one object is extracted from the image. The on-going work extends this to extracting multiple objects simultaneously.

Quick Tour of Features and Use

A list panels in the interface, their features, what they mean, and how to use them. For instance:

  • Parameters panel:
    • Approximate volume: The estimated upper limit of the target volume. The resulting volume will be less or equal than this value.
    • Intensity homogeneity: If the target contains homogeneous intensity, then give a close-to-1 value here.
    • Boundary smoothness: Larger value will result in smoother boundary and a more spherical looking result.
    • Output Label Value: Defined the label value of the output. Also refer to the "Multiple-value label map handling" above.
    • Max running time: The upper limit for program running time.
  • IO panel:
    • Input Image: The image to be segmented.
    • Label Image: The label map providing initial seeds.
  • Output Volume: The output volumetric image.
User Interface

Development

Source code & documentation

The source code of the algorithm is in 3D Slicer source directory tree. Under Slicer3/Applications/CLI/RobustStatisticsSegmenter/

Moreover, the code above is a self-contained implementation for the 3D Sparse Filed Level Set method. The level set evolution is implemented in the base class. Specific flow, such as Chan-Vese, Geodesic Active Contour, or the Robust Statistics based flow, can be implemented in the derived class. Therefore, it can be reused for other types of flow with not much work.


Multi-Object RSS (MORSS)

RSS result for abdominal CT image


Description

In contrast to the above single object RSS, in MORSS, the user can use different label (color) to draw the seed label image. After that, the algorithm detect the number of objects and evolve multiple contours simultaneously.

Module Type & Category

CLI module in 3D Slicer Extension, in the category of Segmentation. Need to install from Extension Manager.

Usage

The basic steps of MORSS is the same as RSS. In fact, if only one type of seed is given (one non-zero label), the MORSS is in fact the single object RSS above.

Moreover, since the number of labels are determined after the algorithm starts, on the user interface, it is difficult to adjust the parameters for each of the target object. Therefore in the current version, the same parameters set on the GUI apply to all the contours. This, however, may not be desirable in all cases. Similar situation is for the expected volume. So on the GUI, we didn't ask for the expected volume. Instead, we used a more engineering term "number of iterations", which also applies to all the contours.

Looking forward, if the algorithm is implemented as a loadable module, then we can have more flexible control over the GUI. In particular, we can create a several-step-module. In the first step the user provide the input and label image. By clicking "Process label image", the number of different labels are learned from the label image. In addition, the GUI of the second step is adaptively adjusted.

Use Cases, Examples

Meningioma

The testing image of this example can be downloaded at: File:RSS Grayscale.nrrd The label image is at: File:MORSS Grayscale-multilabel.nrrd

Cardix

The testing image of this example is sub-sampled from the CARDIX data set from Orisix data set and can be downloaded at: File:RSS CARDIX-sub2.nrrd The label image is at: File:RSS sub2 cardix LVlabel.nrrd Parameters are: Approx vol limit 300mL, IH = 0.01, BS=0.1, Threshold= default(= 2)

The multi label image is at: File:MORSS CARDIX-sub2-multilabel.nrrd Parameters are: IH = 0.5, BS=0.6, number of iteration 600.

Acknowledgment

This work was supported in part by grants from NSF, AFOSR, ARO, as well as by a grant from NIH (NAC P41 RR-13218) through Brigham and Women’s Hospital. An NSF Fellowship supported part of the work.

Publications

The method is based on:

  • Yi Gao, Ron Kikinis, Sylvain Bouix, Martha Shenton, Allen Tannenbaum, A 3D Interactive Multi-object Segmentation Tool using Local Robust Statistics Driven Active Contours, Medical Image Analysis, 2012, http://dx.doi.org/10.1016/j.media.2012.06.002

Rons

File:RobustSeg.zip

Retrieved from "https://www.na-mic.org/w/index.php?title=RobustStatisticsSegmentation&oldid=77791"