Format

Send to

Choose Destination
Exp Toxicol Pathol. 2007 Aug;58(6):411-8. Epub 2007 Apr 30.

Cell proliferation measurement in cecum and colon of rats using scanned images and fully automated image analysis: validation of method.

Author information

1
Safety Profiling & Assessment, Pathology, Novartis Pharma AG, Basel, Switzerland. elke.persohn@novartis.com

Abstract

The purpose of this study was to establish and validate fully automatic measurement of cell proliferation on scanned images of rat cecum and colon. Tissue slides were taken from a 4-week mechanistic study and processed for BrdU immunohistochemistry. Four sections of the cecum and colon per slide were scanned with the Zeiss MIRAX SCAN and transferred to the Definiens eCognition Analyst LS5.0 system for evaluation. Two rule sets for automatic counting of BrdU-positive and negative nuclei from mucosal cells on the image tiles were created by Definiens, one for cecum, one for colon. For validation, manual counting of 16 randomly selected tiles from five different slides of colon and cecum was performed. Negative and positive cell nuclei were counted in each image tile by four different people. Comparison of results from manual counting with the automatic counting showed that the sum as well as single tile data and labeling index (LI) from automatic counting were within the range of manual counting results +/-10%. Automatic counting included only cell nuclei within the mucosa whereas muscularis and lymphoid tissue as well as wrinkles from tissue preparation were excluded. In addition, two data sets from automatic counting of the same image tile were compared: (1) data where image tiles with incorrect detection of mucosa were excluded from further calculation of LI and area, and (2) data where no visual check was performed and all measurements were included. Results were very similar for both data sets. The necessity of the manual correction may therefore be doubted.

PMID:
17467963
DOI:
10.1016/j.etp.2007.01.005
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center