Download CHENG-CHANG LU - Computer Science

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Amitosis wikipedia , lookup

Mitosis wikipedia , lookup

Cytokinesis wikipedia , lookup

Cell growth wikipedia , lookup

Organ-on-a-chip wikipedia , lookup

List of types of proteins wikipedia , lookup

Transcript
KENT STATE UNIVERSITY
DEPARTMENT OF
COMPUTER SCIENCE
DIGITAL IMAGE PROCESSING
PROJECT WORK
Neuron Classification
Presented to:
CHENG-CHANG LU
Professor and Assistant Chair.
Presented by:
TEAM-18
Sasrusha Vasireddy([email protected] 810732880)
Kushal Babu([email protected] 810732862)
Shashank Rao([email protected] 810710582)
REPORT
AIM:




To Analyze segmented cells
Identify cell body and cell location
Number of projections (length, size)
Polarity( Uni Vs Multi)
Implementation Method:
The main part of the Neuron Classification is to segment the neurons from the 3D
image.
 In the process of automating the segmentation, first we make the image as
binary which converts an image to black and white. It will set an automatic
threshold level to create the binary image. This is simply done by running
inbuilt plugin named “Binary”.
 And after that we use fill holes which will helps to fill holes in objects by
filling the background.
 The next step we performed is the watershed method. Watershed
segmentation is a way of automatically separating or cutting apart particles
that touch.
 The above watershed algorithm is simply implemented by calling the
functions “watershed”.
 After segmenting the cells what we did is finding centroid to the each and
every segmented cell which helps us to create seed points.
 Then we applied the level set method which is one of the segmentation
techniques based on differential equations that is Progressive evaluation of
the differences among the neighboring pixels to find the object boundaries
and then we used filters. Lever set Segmentation is also implemented by the
code below.
 After completion of level sets we have used the grouped Z which helps to
perform a z-projection on a stack of images. The stack is segregated into
some number of groups (by specifying a group size) and a z-projection is
performed on each group using Z Projector. The result is a new stack of
images, one slice for each group. Here actually the Z projector will use the
average intensity method and then we use the threshold adjuster which helps
to adjust the lower and upper threshold levels of the active image.
 In the Level Set algorithm the seed point we have used is the Centroid of the
neurons which we obtain from the Particle analyzer.
Procedure we followed
A brief explanation of the process we followed
 Storing all the images in a stack
 Filter is applied to decrease the noise in the image
 The image obtained from the above step is converted into 8 bit image.
 Then we apply the watershed algorithm to find the cell location.
 By using particle analyzer we will find the table with cell location
 Then we will convert image to binary and level set algorithm is used for
seed growing.
 A grouped Z projector is used to combine all the images to a single 3D
image.
 Then skeleton analyzer is used to find the branch length and the branching
patterns.
Result:
We have successfully identified the location of the cell body by segmenting the
neurons and then identified the number of branches and their lengths.
Tools Used: Java, ImageJ.
References:





http://www.csee.usf.edu/~manohar/Papers/Segmentation/Seeded%20Region%20growing
.pdf
http://textofvideo.nptel.iitm.ac.in/117105079/lec31.pdf
http://rsbweb.nih.gov/ij/docs/guide
https://rsb.info.nih.gov/ij/plugins
http://textofvideo.nptel.iitm.ac.in/117105079/lec31.pdf

http://microscopy.berkeley.edu/courses/dib/sections/04IPIII/particles.html