ANALYSIS OF A LOCALLY VARYING INTENSITY TEMPLATE FOR SEGMENTATION OF KIDNEYS IN CT IMAGES

1
ANALYSIS OF A LOCALLY VARYING INTENSITY TEMPLATE
FOR SEGMENTATION OF KIDNEYS IN CT IMAGES

• MANJARI I RAO
• UNIVERSITY OF NORTH CAROLINA AT CHAPEL HILL

2
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

3
External Beam Radiation Therapy

• (Image Types CT, MRI, Ultrasound, PET etc.)
• Simulation
• Treatment Planning

4
Segmentation

• Segmentation is a process of extraction of
  information from an image in such a way that the
  output image contains much less information than
  the original one, but the little information that
  it contains is much more relevant to the purpose
  of the task .
• Medical Image Segmentation Extraction of
  anatomical structures such as the kidney.

5
Segmentation in Treatment Planning

• Manual Segmentation
• Computer-based models
• Combination techniques

6
Virtual Simulation

• Combination of simulation and treatment planning
• Detect tumor sites from different viewpoints
• Design irradiation beam and orientation
• Calculate dose distribution
• Presence of patient not required

7
Manual Segmentation - Limitations

• Dependent on the experience of the operator
• Inter-operator and Intra-operator variability
• Difficult to identify 3D structures on a 2D slice
• Complexity is increased due to presence of
  infiltrating tumor and disease

8
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

9
Shape Representation Using Medial Models

• A medial axis is the locus of centers of spheres
  that are bitangent to the surface of the object
  being represented
• The center point of each such sphere is a point
  on the medial surface
• M-reps are a class of medial models

10
Object Representation via M-reps

• An m-rep consists of a grid of medial atoms, each
  of which is made of a hub and a pair of spokes
• The atoms in an m-rep define the following
• A medial position x in the mesh
• Two vectors of length r, which is the radius of
  the bitangent sphere inscribed inside the object.
  The vectors are the spokes that point to these
  points of tangency.
• The angle T formed between each of the spokes and
  the angular bisector b.
• A frame F (n,b,b-), where n is normal to both b
  and b-, which defines the tangent plane to the
  medial sheet at x.
• The curvature ? of the object about a point.

11
Object Representation via M-reps

• M-reps define an object-based intrinsic
  coordinate system represented by
• (u, v, t, ? ), where u and v represent the row
  and column corresponding to the position of the
  atom in the medial mesh
• t indicates which side of the medial locus the
  point lies on t -1 or 1 for internal medial
  points and varies around the crest
• from -1 through 0 at the boundary to 1
• ? measures the distance along the spokes from
  the boundary, with ? gt 0 outside and ? lt 0 inside
  the boundary and ? -1 at the medial locus

?
?
u
t
v
12
Object Representation via M-reps

• The medial locus is a curve for objects in 2D and
  a sheet for objects in 3D
• It is sparsely sampled to produce an approximate
  surface of the m-rep or the implied boundary.

13
Segmentation using M-reps

• Transformation of the whole kidney model based on
  both the similarity transform and Principal
  Geodesic Analysis or PGA -translation, rotation,
  scaling of the entire model and shape variation
  according to the principal modes of variation of
  a mean model.

14
Segmentation using M-reps

• Deformation of each medial atom in the
  model-translation, rotation and scaling of each
  individual atom of the medial mesh.

15
Segmentation using M-reps

• Atom deformation in this step is based on
• Geometric Typicality - Calculated by relating an
  atoms coordinates predicted by the stage
  immediately previous to the current deformation
  stage, and by its neighbors at the current stage.
  
• Geometry to Image Match Calculated based on a
  template defined in a collar region about the
  boundary.

16
Segmentation using M-reps

• Fine-scale surface refinement, i.e., each surface
  tile of the implied surface is shifted along its
  normal to optimize the objective function. This
  scale was not used in the present study.

17
Kidney Segmentation

• Surrounded by crowded soft tissue environment
  most soft tissue does not contrast well against
  its background.
• High risk due to radiation exposure during
  treatment of abdominal and pelvic tumors. Hence,
  important in segmentation during treatment
  planning.
• Simple organ, can be represented by a
  single-figure m-rep

18
Kidney as an object of interest for this study

• Variation of intensities along the boundary of
  the kidney
• Examples

19

• Axial View Coronal View

20

• Axial View Coronal View

21
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

22
Training and Target Images

• Training Images
• To generate a mean kidney m-rep model
• To generate intensity profiles for the locally
  varying template

23
Training Images

• CT Images Siemens Somatom Plus 4 scanner
• Raster resolution (number of pixels per slice)
  512 x 512
• Pixel size ranged from 0.098mm2 to 0.156mm2

24
Criteria for selection of Training Images

• Presence of whole kidney(s)
• Position of patient during scan

25
Criteria for selection of Training Images

• Absence of contrast agent
• Absence of tumor, disease or kidney stones

26
Criteria for selection of Training Images

• No more than moderate motion artifacts

27
Criteria for selection of Training Images

• Margin of at least 2cm on the superior and
  inferior edges of the kidney.
• Slice thickness less than or equal to 5mm per
  slice.

28
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

29
Image Match based on a Template

• A target pattern at different locations in an
  image - maximum response when the intensity
  values of the pixels in the image correlate to
  the values at the same locations specified by the
  template
• Determined by training the model on a population
  of user-approved segmentations or the answers

30
Image match based on a Template

• For an m-rep, the template is defined in the
  collar region about the boundary
• The width of this region ranged from 0.3r to
  0.3r with 0 at the boundary in this study
• The template thus corresponds to a Gaussian
  centered about the boundary with a standard
  deviation of 0.15

31
Locally varying intensity template

• Function of figural positions along the boundary
  of an m-rep
• Generated from a population of training images
• Several profile types as compared to the
  Gaussian which had only one

32
Training

• Manual segmentation of kidneys from the training
  data set
• Conversion of resulting contours into blurred
  binary images
• Deformation of an m-rep model into the blurred
  binary images to generate a mean model with
  principal geometric modes of variation via PGA
• Segmentation of the training kidneys by deforming
  the mean model into the gray-scale training
  images

33
Training

• Generation of intensity profiles at many points
  on the surface of the segmented kidneys
• Classification of the each of the intensity
  profiles into one of three categories, based on
  the best match among three analytic filter types
• Final Step Segmentation of kidneys in the target
  images using the locally varying intensity
  template to compute image match

34
Manual segmentation

• Software used for radiation treatment planning
  similar to drawing tools
• Contours were traced on each cross-sectional
  slice
• Intensity windowing to enhance the quality of the
  displayed image

35
Manual Slice-by-slice contouring using
Anastruct-Editor
36
Generation of Training Binary Images

• Output of hand-segmentation series of contour
  stacks
• Converted to binary images (have two intensity
  values, 0 for black and 1 for white)
• Gaussian smoothing operator was used to blur the
  images. The ? of the operator was approximately
  1mm

37
Generation of the mean kidney m-rep

• Single figure mesh of 15 atoms (5 rows and 3
  columns) was fit into each of the training
  blurred binary images
• The mean model and the corresponding principal
  modes of variation were obtained from this
  population of m-reps
• Initial m-rep was replaced by the mean and the
  process was iterated until convergence

38
Generation of the locally varying Kidney Template

• Initial Analytic Filters
• Light-to-dark - higher intensities in the kidney
  as compared to surrounding structures
• Dark-to-light - lower intensities in the kidney
  as compared to surrounding structures
• Notch - similar intensities in the kidney as well
  as surrounding structures with a narrow dark
  region in between

39
Initial Analytic Filters
Intensity
Points along the normal
40

• Image corresponding to the light-to-dark filter

41

• Image corresponding to the dark-to-light filter

42

• Image corresponding to the notch filter

43
Generation of Profiles for the Template

• End points of the spokes of the outer atoms in
  the medial lattice of an m-rep are linked
  together to form a quadrangular mesh surface
• Surface was subdivided to provided a natural
  framework for defining the boundary at 2562
  points on the surface of the m-rep

44
Generation of a Profiles for the Template

• At each of the 2562 boundary points, normals were
  drawn (length from
• -0.3r to 0.3r with 0 at the boundary as defined
  by the collar)
• Each normal was sampled at 11 points for each of
  the 2562 boundary points

45
Generation of Profiles for the Template

• Segmentation of the training blurred binary
  images with the mean kidney m-rep
• Generation of profiles for each m-rep in this
  population with intensity information from
  corresponding gray-scale images

46
Generation of the locally varying Kidney Template

• Responses of the profiles to each of the three
  analytic filters, given by the dot product,
• X.Y ?xiyi , i 1,2,, n,
• X -vector representing the profile
• Y -vector representing the filter
• The highest response among the three filters for
  each point determined the filter type
  classification for that point

47
Classification of the profiles according to
responses
40.66
16.82
-34.74
48
Converged Mean Filters
Intensity
Points along the normal
49
Filter distribution on the surface of the left
and right kidneys in the templates
Left Kidney
Right Kidney
50
Segmentation of Target Images
51
Segmentation of Target Images
Axial View
Coronal View
52
Segmentation of Target Images
Axial View
Coronal View
53
Accuracy Evaluation

• VALMET Software package involving voxel-based
  comparision
• Volume overlap - intersection of the two volumes
  divided by their union
• Maximum Surface Distance or Hausdorff Distance -
  the largest distance between two surfaces (Not
  symmetric)
• Mean absolute surface distance - how much on
  average the two surfaces differ (Also not
  symmetric)

Curve 1
a
b
Curve 2
54
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

55
Comparison using VALMET Gaussian Vs Locally
Varying Template

• 24 Target Cases (12 left and 12 right kidneys)
• Segmentations were compared to human
  segmentations performed by two experts
• Average increase in volume overlap over all the
  cases - 1.3
• Mean surface separation between human
  segmentations and M-rep segmentation -0.33cm for
  the Gaussian template and 0.32cm for the locally
  varying template

56
Comparison using VALMET Gaussian Vs Locally
Varying Template

• Mean Surface Separation

Rater B Vs M-rep
Rater A Vs M-rep
57
Results (Contd.)
Axial View
Coronal View
58
Results (Contd.)
Axial View
Coronal View
59
Results (Contd.) Error in Human Segmentation
Coronal View
Sagittal View
60
Results (Contd.)
Coronal View
Sagittal View
61
OVERVIEW

• Introduction
• Background
• Materials and Methods
• Results
• Conclusions and Discussion

62
Analysis of the Results

• Locally varying template based segmentation
  showed improvement for 65 of the cases
• Achieved a greater degree of automation in the
  entire segmentation process

63
Analysis of the Results
Coronal View
Axial View
64
Analysis of the Results
Axial View
Coronal View
65
Instances of Human Intervention
Axial View
Coronal View
66
Instances of Human Intervention
Axial View
Coronal View
67
Future Directions

• More than three filter types to define the
  template
• Statistical evaluation of variation of profiles
  along the boundary
• Templates based on weighted profiles, responses
  which are neighbor dependent
• Higher density m-rep model