Skip to content

sumitb/nuclei-analysis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

77 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

###nuclei-analysis

a hadoop-gis project for data analytics.

#####Problem Statement: Given set of Si of polygons (nuclei), compute feature vectors Fi from Pathology images. Be creative about choice of features. Examples:

  • Mask area, perimeter, shape
  • Within-mask texture
  • Padded region texture

Develop an interactive program that supports the following:

  1. Select subset of features
  2. Select number of clusters
  3. Choose initial cluster centroids by random choice of data points
  4. Invoke Kmeans clustering algorithm to cluster feature vectors
  5. Display clusters generated by clustering algorithm, cluster centroids along with within-cluster sum of squares
  6. Store clusters from Si : assignment of data points to clusters, initial cluster centroids, within-cluster sum of squares, feature subset in database and all other metadata needed to re-run computation

#####TODO after Phase 1:

  1. UI Features
  • Move to proper cluster section from nav menu
  • Pick a single query dropodown and place it properly
  • Remove dummy images from work and put workflow images
  • Display polygons in a info section from cluster
  • Custom query option redesign and work with actual data
  1. Back-end
  • Run whole slide images
  • Figure out how to use cluster power
  • Use mongo-db to prepopulate tables
  • Fix bug to populate clusters from predefined queries

####Phase 2 #####Requirements

  • Logon to eagle bmi cluster and add anaconda module.
  • Clone this repository and set root path of dataNode.
  • Feature Extraction: python analysis.py
  • Run flask server: python hackrpi/run.py
  • Access web interface from http://localhost:5000

About

a hadoop-gis project for data analytics.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •