Preliminary slide scanner throughput evaluation in a intensive digitization facility setting

Background Whole Slide Imaging (WSI), called also Digital Microscopy, is the most current approach to digitization of histological information [1,2]. It allows for transferring a whole histological slide into digital form, thus enabling any kind of digital treatment from storage and transmission, to telediagnosis, to automatic image analysis. WSI technologies developed only recently, and thus most uses described in literature are coming from research and teaching applications [3]. However, one acknowledged potential use of this technology is also aimed at dematerializing slide archives, by bringing them in digital form inside a so-called PACS (Picture Archiving and Communication System) [4]. This application would provide a major boost in the adoption of WSI in the routine work of a clinical pathology laboratory. Two main differences can be recognised between academic applications like research and teaching, and routine application in the pathology laboratory: slide volume, and diagnostic reliability. The former difference is related to the number of involved glass slides and the time span on which scanning occurs. Teaching in particular, but also research applications, foresee the acquisition of a limited number of slides, in terms of either total number or scanning needs per time unit. In fact, teaching with digital slides usually involves slides accumulating into a teaching archive that may slowly grow, in years and years, but with no massive amounts of slides involved. Research trials instead might involve the scanning of a large number of slides but in a limited time frame, related to the life span of the research project, and that can be archived offline at the end of the project. Both cases differ from the routine acquisition in a clinical pathology laboratory, where there is the need for a sustained acquisition of a fraction (or all) the glass slides daily produced by the laboratory, to be made available to pathologists when needed. This means that the scanning procedure should be as efficient as possible, and in particular able to perform a sustained scanning as quick as glass slide production is in the specific laboratory. This means also that there is need for personnel (e.g., laboratory technicians) that feed the scanner with glass slides, start the scanning procedure, check associated patient data, verify results, unload slides, etc. Any technical hitch occuring in those phases (e.g., software bugs, slide loading difficulties, etc) is likely to decrease the overall thoughput of the scanner. The latter difference is related to how digital slides are used. Scanning is not a process without errors: loss of information is always present, and derives in part from the process itself, in part from specific features and pitfalls of the scanning device, in part from the preparation technical quality of the source glass slide. In teaching applications, slides are selected for their educational meaning, so there is a selection of the acquired material, that allows for recognising scanning errors or simply missing information. Such selection is also made possible by the low number of slides acquired at each scanning session. The same can be considered true for any research usage, since it is done as part of the research project. On the other side, routine scanning is aimed at providing slides for the diagnostic work of the pathologist, either for primary diagnosis or for giving access to previous slides of the same patient when diagnosing a new histologic exam. Thus, diagnostic reliability of the digital slide should be guaranteed, and this means that acquired slides should be as good as possible as they come out from the scanner. Since every device may fail in acquisition of some slide, the least they fail, the better is. * Correspondence: [email protected] † Contributed equally Dept. of Mathematics and Computer Science, University of Udine, Italy Full list of author information is available at the end of the article Della Mea et al. Diagnostic Pathology 2013, 8(Suppl 1):S45 http://www.diagnosticpathology.org/content/8/S1/S45