Tissue Information as a Specific Instance

A specific example of the conversion of biological information to digital information occurs at the tissue level. Until recently, it was felt that only a pathologist could interpret the meaning of patterns of cells and other structural components of tissue. This meaning was summed up in the diagnosis that was applied to the tissue and used to guide healthcare decision-making. Over the past two decades, digital imaging of tissues on slides has created the basis for management of tissue information at the image level for ease of data sharing between pathologists and researchers. However, this did not convert the data completely into digital form, because human interpretation and diagnostic assignment of the overall image were still required. This limited the ability to correlate tissue data with other biological information to the level of resolution that diagnosis provided.

Recently, it has become possible to use automated machine vision analysis systems to measure all of the components that can be made visible within a tissue (both structural and functional) with reference to one another. This is known as Hyperquantitative Analysis of Tissue (Fig. C.6).

ï Cells

x,y; #

Mean±SD

ï Molecules

x,y; #

Mean±SD

ï Matrix

x,y; #

Mean±SD

ï Layers

x,y; #

Mean±SD

ï Structures

x,y; #

Digital Automated Digital Imaging Tissue Analysis

Computation

Pathologist Knowledge Incorporated Here

Data Available For Rich Digital Correlation With Other Datasets, Including Genomic, Proteomic, Etc.

Figure C.6. Capture of tissue information in hyperquantitative fashion. All components of the tissue that can be made visible are located simultaneously after robotic capture of slide-based images. This step automates the analysis of tissue, putting it immediately into a form that enables sharing of images and derived data.

Preparation of tissue information in this way requires two steps:

a. automated imaging that enables location of tissue on a microscope slide and the capture of a composite image of the entire tissue — or tissues — on the slide b. the application of image analytic software that has been designed to automatically segregate and co-localize in Cartesian space the visible components of tissue (including molecular probes, if applied)

Tissue information captured in this way enables very precise mathematical comparison of tissues to detect change (as in toxicology testing or, ultimately, clinical diagnostics). In each case, substantial work must first be done to collect normative reference data from tissue populations of interest.

More importantly, when tissue information is reduced to this level of scale, the data is made available for more precise correlation with other data sets in the continuum of bioinformatics in the following applications:

• Backward correlation: "Sorter" of genomic and proteomic data

Rationale: When gene or protein expression data are culled from a tissue that has undergone hyperquantitative analysis, tighter correlations are possible between molecular expression patterns and tissue features whose known biological roles help to explain the mechanisms of disease — and therefore may help to identify drug targets more sharply.

• Forward correlation: Stratifier of diagnosis with respect to prognosis

Rationale: When tissue information is collected along with highly detailed clinical descriptions and outcome data, subtle changes in tissue feature patterns within a diagnostic group may help to further stratify prognoses associated with a diagnosis and may prompt more refined diagnostic classifications.

Pan Correlation: Tighten linkage of prognosis with molecular diagnostics

Rationale: Since tissue is the classical "site of diagnosis," the use of tissue information to correlate with molecular expression data and clinical outcome data validates those molecular expression patterns with reference to their associated diseases, enabling their confident application as molecular diagnostics.

Nanotechnology developments applicable to imaging and computational science will aid and abet these discoveries.

0 0

Post a comment