Development of Multigene Expression Signature Maps at the Protein Level from Digitized Immunohistochemistry Slides

Abstract

Molecular classification of diseases based on multigene expression signatures is increasingly used for diagnosis, prognosis, and prediction of response to therapy. Immunohistochemistry (IHC) is an optimal method for validating expression signatures obtained using high-throughput genomics techniques since IHC allows a pathologist to examine gene expression at the protein level within the context of histologically interpretable tissue sections. Additionally, validated IHC assays may be readily implemented as clinical tests since IHC is performed on routinely processed clinical tissue samples. However, methods have not been available for automated n-gene expression profiling at the protein level using IHC data. We have developed methods to compute expression level maps (signature maps) of multiple genes from IHC data digitized on a commercial whole slide imaging system. Areas of cancer for these expression level maps are defined by a pathologist on adjacent, co-registered H&E slides, allowing assessment of IHC statistics and heterogeneity within the diseased tissue. This novel way of representing multiple IHC assays as signature maps will allow the development of n-gene expression profiling databases in three dimensions throughout virtual whole organ reconstructions.

Registration of segmented histological images using thin plate splines and belief propagation

“…We show the results of registering adjacent histological slices of prostate 1 and rat kidney (Figure 1). The images of a typical size 1400 × 2050 were very quickly semi-automatically segmented by Ilastik * into four classes.…”

“…The slices need to be registered to fuse information from the different stains. 1 The misalignment is due to the cutting and acquisition process. The different stainings results not only in different colour but also in different elements of the tissue being visible.…”

Registration of multiple stained histological sections

“…This superpixel-based simplified representation of the image is fed to an Expectation-Maximization (EM) algorithm that estimates a four class Gaussian Mixture Model (GMM). From the probability model, a compact 6-level multi-class segmentation pyramid (see Figure 3 (b) and (c)) is computed using Graph Cuts segmentation [5] with smoothness coefficients logarithmically distributed in the range [1,100000].…”

“…However, due to cost and time constraints, immunohistochemistry (IHC) is the method of choice in clinical settings. Metzger et al [1] recently developed a platform that computes multiple-gene expression signature maps from consecutive IHC sections digitized using a commercial whole-slide imaging system. The critical step in this process is the correct assignment of spatial correspondence between differently stained sections.…”

“…The critical step in this process is the correct assignment of spatial correspondence between differently stained sections. In their platform, the IHC images were aligned to a reference Hematoxylin and Eosin (H&E) image through a two-step process involving coarse manual alignment followed by an automatic fine rigid registration using TurboReg 1 . Nevertheless, histological images often suffer from non-linear…”