SOX2 IHC Pipeline

In the manuscript entitled “Whole-brain spectroscopic MRI biomarkers identify infiltrating margins in glioblastoma patients” (Neuro-oncology, 2016), we describe a pipeline for automatically computed several features related to SOX2 immunohistochemistry, including the SOX2 density metric used in the statistical analyses.


This pipeline is based on the HistomicsTK library being developed by our collaborator, Dr. Lee Cooper, which can be found on the Digital Slide Archive’s Github Repository.

We used two scripts based on this library, as described below. The scripts are written in Matlab and designed to run in a Linux environment. The code for these two additional scripts can be found here.

  • IHCProcessSlidesArea.m – This script computes the standard features from Dr. Cooper’s pipeline (location of nuclei, centroids of nuclei, separation of background from foreground) with the addition of calculating the area of each detected nucleus, based on the number of pixels which make it up divided by the pixel resolution of the slide.
  • TissueArea.m – This script computes the total area of a slide by separating out the foreground from the slide background.

The SOX2 density metric is taken to be the sum of areas of all SOX2+ nuclei divided by the tissue area in the slide, using the two scripts above.


Support for this work comes from two National Cancer Institute grants, U01CA172027 and U24CA194362.