At our lab we are interested in developing software tools for fast and automated analysis of images. Please see below for a variety of software tools developed here.
Feature Extraction

Software analysis of cells is divided into two aspects, Feature Extraction and Object Classification
Feature extraction involves extracting relevant features from the fluorescence micrograph through the use of image segmentation. A variety of features are extracted for each channel. They are broadly classified as "Morphology", "Intensity", "Texture" and "Colocalization". Details of the features can be found here. The script is developed using Acapella(Perkin Elmer) script.

Script & Procedures

NEW: Texture analysis suite Procedures to calculate Haralick's texture features, threshold adjacency statistics, and radial moments
Colocalisation Coefficients' procedure suite Calculates Pearson's, Mander's, ICQ and Overlap coefficients for two user defined images using the area defined by a user selected objects list.
'Background correction' procedure Subtracts the mean intensity of the background from an image. The background is assumed to be the area outside the user-defined object list. By Karsten Kottig, Evotec.
NEW: Also includes a Rolling Ball like algorithm for background correction.
'Cell segmentation' example script Illustrates how to:
        - generate stencils of different subcellular           compartments
        - calculate mean intensity of different compartments
        - calculate the ratio of two compartments   
Nearest Neighbour' Deblur Performs a crude nearest neighbour 'deblur' based on a top, middle and bottom slice.
'Script Bits' Useful for combining tables, outputting object-level data to text files, retrieving experimental information.
«  Back to Top

Classification and Exploration - CAFE MiCells

Schematic illustration of flowchart for CAFE MiCells

We have developed a software capable of classification of the data (which is an output of the feature extractor). The software is called CAFE MiCells . The flow chart is displayed in figure. The input format is very generic and can be used with any feature extractor. The software is written in MATLAB and involves the use of Statistics, Neural network and bioinformatics toolboxes. The code can be downloaded here. The manual for the software is available here.

Download the Following files
1. miClassify_ver0.7.1.zip
2. miControls_ver0.6.3.zip
3. miFeatures_v0_15.script
4. miClassify_manuals.pdf
5. miFeatures_manual.pdf
«  Back to Top

Imagej is an image processing and analysis open source software developed by NIH. We were able to extend the features of imagej to analyze FLIM and FRET data . Please download our version of imagej here
«  Back to Top