Since the days of Rudolf Virchow, microscopy is an integral part of diagnostics and a foundation of pathology since over 100 years. Technical progress has increased considerably in recent years, particularly in the field of instrumental analysis methods. But unlike in laboratory medicine, diagnostics in pathology is still largely based on the pathologist’s personal assessment of the morphology and interpretation in the overall context of the patient. In addition, there is the complexity of the examination material, which ranges from body fluids and small biopsy samples to complex surgical specimens. However, the performance and quality of digital image processing has increased considerably in recent years and, above all, the transmission of huge quantities of image data has only recently become possible. What began in university teaching as a microscope course with virtual (scanned) sections for medical students has now the potential to largely replace the microscope at the workplace in routine diagnostics as well. Through the use of computer-based technology, digital pathology utilizes virtual microscopy. Glass slides are converted into digital slides that can be viewed, managed, shared and analyzed on a computer monitor. Saved on a server as digital data, these scans can then be called up as part of the diagnosis process and examined at different magnifications like in a microscope.

Advantages of digital pathology

This system of digital pathology offers numerous advantages. Traditionally, tissue slides that may be stained to highlight cellular structures, are observed by the pathologist under a microscope. If slides are digitized, they can be shared with specialists or students via telepathology and analyzed and correlated with patient data using computer algorithms. These algorithms can be used to automate the manual counting of structures, or for classifying the condition of tissue such as is used in grading and staging of tumors. This has the potential to reduce human error and improve the accuracy of diagnosis. Digitized slides can be shared, increasing the potential for data use in education and in consultations between pathology experts. With this system, glass slides are no longer unique and cases can be accessed from anywhere on earth. Digital image analysis can also be applied and archived images can be easily accessed. In a sense, digital pathology also allows a look back in time, as the slides of referred cases are still easily accessible even after the slides have been stored in an archive.

.

Cancer under the microscope of the Digital Pathology at CBmed in Graz, using multispectral picture technologies with new analysis tools.

New technologies at CBmed

All in all, the role of pathologists is changing as cancer therapy advances, and personalized treatment options require a more precise diagnosis. In order to achieve this goal, cooperation in the procurement and integration of the relevant information from various disciplines is necessary. Precise diagnostics for personalized treatment and research require the involvement of multiple disciplines and locations. As a research center, CBmed serves as an interface to various institutions. In close cooperation with the Institute for Pathology at the Medical University Vienna as well as the Institute for Pathology and the Clinical Department for Oncology at the Medical University of Graz and together with the Biobank Graz, CBmed provides solutions such as Cohort design and assembly, pathological analyses, IHC (immunohistochemistry) and fluorescence in-situ hybridization (FISH), digital evaluation of slides as well as other techniques. Besides, the expertise ranges from the development of assays, multiplex immunofluorescence (IF) imaging to prevalence studies and patient stratification.

One focus of digital pathology research at CBmed is immunohistochemical analysis, and its workflow has become well established over the years. Usually we analyze surgical specimens or biopsies, e.g. from tumors of patients or from mouse models. After fixation in formalin, this tumor is dehydrated and embedded in paraffin. 2-4 micrometer sections are made from this FFPE tissue block (Formalin-Fixed Paraffin-Embedded). The FFPE tissue block is subjected to immunohistochemical staining, which allows us to draw conclusions about the concentration, distribution and activation of protein molecules in the fine tissue structures of the tumor cells in situ. Usually only 1 protein/section could be detected. In the past, this technique was performed individually for each marker of interest. The multiplexing method allows the analysis of up to 6 protein molecules on a single section. This allows, among other things, interaction studies of proteins at cellular structures and their spatial distribution in the cells. Digital imaging now enables further data analysis in the field of AI, such as machine or deep learning, which, combined with the other advantages of digital pathology already mentioned, brings far-reaching innovations in pathology also to this field. Multispectral imaging technology (MSI) has the potential to completely change the fields of digital and computer-assisted pathology and open up new avenues in the histological analysis of FFPE tissue specimens, such as digital staining or the reduction of inter-laboratory variability of digitized samples.

Mouse pancreas at the Digital Pathology of CBmed in Graz.

Focus on multiplexing workflow at CBmed

Multiplexing offers an unmatched insight into spatial proteomics in a variety of cell populations and is bridging the gap between high throughput and single cell technologies. Various approaches to multiplexing exist, each relying on a different technology including stain removal, image cycling and mass cytometry. What makes the digital pathology at CBmed special is the ability to create several panels with up to 6 biomarkers, for human and mouse tissues for comparative pathology. This workflow begins with a multiplex staining on FFPE tissue, using same species antibodies. With the staining of up to 6 biomarkers in a single section, the next step is usually a multispectral imaging, using a PerkinElmer Vectra imaging system. It allows a whole slide scanning (4x, 10x), TMA (tissue microarrays) core detection and selecting of ROIs (regions of interest). Pictures with a higher resolution (20x, 40x) can be made available for further image analysis. Using inForm Software, the image analysis allows a deeper look in the tissue segmentation, cell segmentation and cell phenotyping as well as the composite image. With an unmix multispectral image, it is possible to remove AF and assign false colors. Data science finally goes beyond the visualisation of the generated data – with the help of more specific algorithms, and as already mentioned AI and machine learning, the results are being interpreted profoundly and in great detail. This method enables analysis such as the spatial and cell distribution.

We have already established 6 individual panels and we will be standardising more tailored panels in the near future. We are currently working on several international research projects as well as commissioned work from industry partners. Several scientific papers are in the works as well. Just recently, a scientific team of CBmed investigated the morphological properties of images containing a large number of similar objects like biological cells and if they can be exploited for image classification. The paper „C2G-Net: Exploiting Morphological Properties for Image Classification” can be downloaded here.

The Authors