Pathology workflow and the integration of image analysis

Abstract

In today’s pathology lab, understanding and optimizing the workfl ow becomes more and more important. Th e integration of image analysis into the workfl ow requires special considerations for generating correct results. Currently, image analysis in pathology is used primarily for the quantifi cation of IHC slides like Her2. Pathology laboratories can choose from a number of image analysis systems which scan complete slides at different magnifi cations and generate virtual slides for review by pathologists. Pathologists can mark regions of interest on virtual slides for quantifi cation. Specifi c algorithms can then be applied to each region, generating results/scores that are calculated for these regions. Th e resulting data is then, together with pre-selected images, compiled into a customizable report. Th ere are four key components for image analysis: • Staining quality • Image quality • Algorithm • Region selection Staining quality can be improved through the use of automated staining with defi ned staining protocols. Additionally, linking specifi c pre-treatment, antibody, and staining protocols to an image algorithm is key for consistent results. Having consistent staining is necessary for successful image analysis because diff erences in staining lead to diff erences in acquired digital images thereby aff ecting image analysis. Image quality can be achieved through automation and calibration. Algorithms need to be specifi c to the antibody used to generate the appropriate results. Region selection relies on the pathologist’s expertise. Th us, image analysis is used as an aid to the pathologist. A key component to image analysis is the integration of the image analysis system into the overall lab workfl ow. For example, Dako Link is used to connect diff erent instruments in the IHC laboratory and to capture data from diff erent steps in the slide preparation process, which are stored and can be retrieved. In this workfl ow, the fi rst step is accessioning, where case information is captured together with the requested staining protocol. Th e automated staining instrument uses this information (including the antibody and the staining protocol) to stain the slide. Th e actual information for a specifi c slide is then stored. Once the slides are stained and ready, they are loaded onto the ACIS® III, which accesses the Dako Link database and retrieves information about the slide. Th e information is then used to select the appropriate algorithm, which will be used to process regions on the slide. By linking automated staining systems with image analysis systems, the user is not required to re-enter information and therefore the opportunity for errors is lowered, while at the same time it is guaranteed that the correct algorithm is used for the slide. Th is concept could be used to connect other image analysis systems to laboratory workfl ow, improving overall workfl ow and reducing the number of opportunities for transcription and other errors. In the future, digital pathology will include digitization of the complete workfl ow. Information will be generated in the lab, including staining information and digitization of the slide and the process will be completed by image analysis.