Digitale Pathologie

The use of digital pathology is becoming increasingly important in the field of clinical pathology and molecular pathology. However, in contrast to digital imaging, as in radiology or photography, the digital image is created following the complete cytological or histopathological processing of a smear, biopsy or surgical specimen, at the end of which a histological section is prepared. In contrast to conventional diagnostics, the histological section is not only examined under a microscope, but also digitized and then viewed, used for diagnosis, or used as input in further processing.

The digitization of histological sections allows the development of algorithms in scientific tasks - e.g. through automated recognition of certain structures or patterns - using artificial intelligence (AI) or "machine learning", as well as rapid and flexible location-independent telepathological diagnostics in the clinical field, e.g. in the context of clinical-pathological case discussions or conferences.

However, there are still major challenges that need to be solved in the future, such as the quality control of the generated sectional images, as only high quality digitized histological sections allow for high quality computational pathology. In addition, questions regarding the necessary storage capacities need to be resolved, as the high-quality scanning of histological sections is labor-intensive and time-consuming and therefore personnel-intensive. At high resolution, e.g. to be able to recognize details at subcellular level, images with a size of over 1 GB are often required. Storing this amount of data in the long term requires either compression with loss of detail or deletion of data, which is ultimately undesirable.

Bigpicture, a pan-European network involving the Medical Universities of Vienna and Graz and funded by the EU's Innovative Health Initiative as part of the 2018 call "Central repository of digital pathology slides to support the development of artificial intelligence tools", was founded to set standards in this area. The aim of the project is to develop an archive of 3 million digital sectional images that meet the highest quality standards, comply with data protection regulations and are sustainably available for developments in computational pathology.