Computer-assisted assessment of the Human Epidermal Growth Factor Receptor 2 immunohistochemical assay in imaged histologic sections using a membrane isolation algorithm and quantitative analysis of positive controls

doi:10.1186/1471-2342-8-11

BMC Medical Imaging

Volume 8

Viewing options:

Abstract
Full text
PDF (2.5MB)
Additional files

Associated material:

Related literature:

Articles citing this article
on Google Scholar
on PubMed Central
Other articles by authors
on Google Scholar

Hall BH
Ianosi-Irimie M
Javidian P
Chen W
Ganesan S
Foran DJ

on PubMed

Hall BH
Ianosi-Irimie M
Javidian P
Chen W
Ganesan S
Foran DJ
Related articles/pages
on Google

on Google Scholar

on PubMed

Tools:

Post to:

Research article

Computer-assisted assessment of the Human Epidermal Growth Factor Receptor 2 immunohistochemical assay in imaged histologic sections using a membrane isolation algorithm and quantitative analysis of positive controls

Bonnie H Hall¹^,2^,3^,4 , Monica Ianosi-Irimie⁴^,5 , Parisa Javidian⁴^,5 , Wenjin Chen²^,3^,4 , Shridar Ganesan³^,4^,6 and David J Foran²^,3^,4^,5

¹Graduate School for the Biomedical Sciences, UMDNJ, 675 Hoes Lane, Piscataway, New Jersey, USA

²Center for Biomedical Imaging and Informatics, 675 Hoes Lane, Piscataway, New Jersey, USA

³The Cancer Institute of New Jersey, 195 Little Albany Street, New Brunswick, New Jersey, USA

⁴Robert Wood Johnson Medical School, UMDNJ, 675 Hoes Lane, Piscataway, New Jersey, USA

⁵Department of Pathology and Laboratory Medicine, Robert Wood Johnson Medical School, UMDNJ, 1 Robert Wood Johnson Place, New Brunswick , New Jersey, USA

⁶Department of Medicine and Pharmacology, Robert Wood Johnson Medical School, UMDNJ, 1 Robert Wood Johnson Place, New Brunswick , New Jersey, USA

author email corresponding author email

BMC Medical Imaging 2008, 8:11doi:10.1186/1471-2342-8-11


Published:	5 June 2008

Additional files

Additional file 1:

Selecting the optimal threshold for membrane pixels. Although lower thresholds can produce visually more satisfying membrane isolation results for weaker staining cases (especially IHC = 1+), responses greater than 15 produced the greatest area under the ROC curve using the M_nfeature, and thus the k = 15 threshold was selected to produce the results in these experiments.

Format: TIFF Size: 690KB Download file

Additional file 2:

Membrane Isolation Algorithm results using both high and low thresholds. These are the results from the membrane isolation algorithm using 2 different thresholds k = 15, and k = 5. These are specimens which stained less intensely.

Format: PDF Size: 244KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 3:

ROC curves using Membrane Isolation Algorithm based on k = 15 and a combination (k = 15, k = 5). This is a comparison of ROC curves based on different Membrane Isolation Algorithm (MIA) variations. Since lighter cases showed enhanced membrane isolation detection when lower thresholds were used, a MIA using k = 5 threshold for lighter images and k = 15 for darker images was evaluated (green line). However, this did not improve AUC in comparison to one universal threshold (black line), and consequently, only one threshold (k = 15) was used in the results of this manuscript. It is interesting to note that the combination threshold used had very similar results to manual scoring (blue line).

Format: PDF Size: 10KB Download file

This file can be viewed with: Adobe Acrobat Reader