Image Processing Presentation-1

1
Image Processing Presentation-1

• Nucleus classification
• By
• 1.Murali Kirshna
• 2.Rami Reddy
• 3.Sai Sandeep

2
Contents

• Analyzing the 3d image
• Problem
• Idea of implementation
• Work done so far

3
Analyzing the 3d Image

• 1.Basically the 3d image is stack of the images
  arranged frame by frame
• 2.So we can analyze the 3d image structure by
  studying the layer by layer which can give a
  better view of segmentation with at most
  accuracy.

4
Problem

• Segmentation of the nuclei from 3d image.
• Segment the nucleolus and heterochromatin from
  the extracted nucleus.
• Calculating the volume, area and density of
  nucleolus and heterochromatin

5
Idea of Segmentation-Manual

• The segmentation of the image is based on frame
  by frame.
• Each frame is loaded and then segmentation of the
  each layer is extracted using the wand tool and
  then stored as array
• When all the layer are merged together and then
  it would result in a well structured 3d segmented
  nucleus.

6
Segmented Image Layer By Layer
7
Segmented Image In 3D
8
Idea of Segmentation-Automated

• Select the seed point and then based on that seed
  point the region is grown .
• The region is achieved based on the threshold
  value .
• Once the region is achieved then its separated
  and same process is repeated for the remaining
  layers.

9
Idea of Segmentation-Automated

• All the separated regions together form the well
  defined 3d segmented stack.

10
Algorithm and steps followed

• So after every iteration the output is moved to
  stack where all the segmented nucleus from each
  layer is present

11
Work done so far

• We have done the work manually where we have
  segmented the layer by layer and then we are
  trying to automate the same.
• We have selected a selected a seed point and then
  able to grow the region based on the threshold
  value for one layer
• Next task is automate this process for the
  remaining layers of the image

12
(No Transcript)
13
(No Transcript)
14
Modules and sample code

• Selecting the seed point
• Add the segmented layers into a stack
• Once the stack is formed then calculate the
  volume and density

SAMPLE CODE int nSeeds (int)IJ.getNumber("Numbe
r of Regions/Seeds", 1) SeedGenerator sg new
SeedGenerator(ip) sg.selectSeedsManually(imp,
nSeeds)
15
References

• Seeded Region Growing Features Extraction
  Algorithm by N. A. Mat-Isa, M. Y. Mashor, and N.
  H. Othman
• R. Adams, and L. Bischof, Seeded region
  growing, IEEE Trans. Pattern Anal. Machine
  Intell., vol. 16, no. 6, pp. 641-647, June, 1994.
  
• Seeded Region Growing (ImageJ Plugin)

16

• Questions

17
Thank you