Skip to Content

College of Pharmacy

Microscopy Core

Completed Projects


Histochemistry: Automatic Detection of Nuclei

Image of cells

 

 

 

 

 

 

 

Project Sample Method
Total Cell Number Counting Mice Breast Tumor Tissue Hematoxylin Counterstaining

PI: Professor Kiaris Hippokratis, Ph.D.
Project leaders: Youwen Zhang, Elena Farmaki, Ph.D.
ImageJ macros: Vitali Sikirzhytski, Ph.D.

Custom Imagej macros allows user to count number of nuclei (cells) on hematoxylin counterstained IHC slides.  The user selects the best threshold for every IHC image or automatically applies the same threshold to all images in the folder. The smallest size of nucleus to be counted (the default value is 50 pixels) and initial threshold can be set within the “Parameters “ section of the macros. The obtained statistics are saved as a results.csv file (see the table below). Almost 100% of nuclei were detected correctly within the test images. However, some big cells can be misinterpreted as two or more cells. In some cases, non-cellular components with a high contrast can be misassigned to “nuclei” class.

# File Name Number of Nuclei
1 Image_1.tif xxx
2 Image_2.tif yyy
3 Image_3.tif zzz

Histochemistry: Quantification of Tumor Area

Image of cells

 

 

 

 

 

 

 

Project Sample Method
Estimation of Tumor Area HER2 (+) xenograft Hematoxylin and Eosin Staining (H & E)

PI: Associate Professor Eugenia Broude, Ph.D.
Project leaders: Amanda Sharko, Ph.D., Gary Schools, Ph.D.
Image and Image Processing: Vitali Sikirzhytski, Ph.D.

Instrumentation: Leica DMIRE2 microscope, Zeiss AxioCam MRc color camera, automation by Inscoper cube and custom software, 20x C PLAN NA 0.3 PH1 Leica objective, automatic focusing using reference points. 

Tissue sections were imaged using automatic XY scanning and autofocus. The section shown above (left panel) required about 15 min of imaging time. It consists of 676 single images  (26 x 26; see the right panel for the example of a single image) stitched using custom ImageJ script. Transitions between single images are not noticeable due to custom flat field correction algorithm. We use scaled down mosaics to estimate size of tumor areas and original high-quality images to evaluate staining, sectioning, and assess morphological changes.

 


Challenge the conventional. Create the exceptional. No Limits.

©