Digital cytopathology: a constant evolution
转自 Digital cytopathology ， https://doi.org/10.1111/cyt.12554
Abstract: Capitanio et al provide a succinct review of digital cytopathology technology with particularly good overviews of scanner focusing methodologies as well as explanations of deep focusing techniques. The aim of this letter is to briefly comment and expand on some of the information provided by these authors.
The field of digital cytopathology is in constant evolution, the main focus being how to accommodate the z-axis cell distribution. Capitanio et al provide a succinct review of digital cytopathology technology with particularly good overviews of scanner focusing methodologies as well as explanations of deep focusing techniques. The aim of this letter is to briefly comment and expand on some of the information provided by these authors.
In an ideal world spacing between z-stacks should be tailored to each individual specimen to account for the unpredictable and varied distribution of cells in the z-axis. However, this is impractical and clear evidence-based guidelines for appropriate z-axis parameters are delayed. From a practical perspective, topography software certainly assists in the modification and selection of such scanning parameters. The Royal College of Pathologists of Australasia Quality Assurance Programs (RCPAQAP) is increasingly z-stacking their cytopathology slides in entirety2 , utilising Sectra Uniview to view these z-stacked images through smart streaming. Trials are underway to determine the efficiency of using this z-axis technology for screening and reporting cytopathology.
Capitanio et al discuss optimizing whole slide image quality through manipulation of software. However the challenges of cytopathology WSI (whole slide imaging) can certainly be partially circumvented by modifying cytology specimens themselves so that they are better suited to scanning and image acquisition. Simple solutions such as rapid on-site evaluation of cytology specimens to remove tissue fragments for embedding or use of cell block preparation which presents a more suitable cell distribution topography, is a straightforward way to improve the quality of final WSI.
Focus-fusion or extended focus imaging (EFI) discussed by Capitanio et al – where multiple z-axis planes are assimilated into a single plane – is tempting technology.1 The positive is that the final file for the EFI uses the memory and transfer requirements of a single plane scan. Negatives include the scanning time (which is still the same as that required for creating a multi-layered z-stack) as well as data loss (once all information is combined into EFI, data from the separate z-axis planes is discarded permanently). There can also be problems with background and resolution. Cells that were blurry in the multilayer stack are brought into focus with EFI, but the EFI algorithm is also applied to the extracellular background debris throughout the thickness of the slide potentially resulting in a grainier appearance. This is why sematic focus point analysis, as described in the review by Capitanio et al, may hold promise. Here software integrates cells in a 3-dimensional capacity that are sharply in focus, avoiding artefact and out-of-focus objects. Capitanio et al then takes this thought-process one step further suggesting compression methods that “discard the redundant material that is replicated between z-stack layers, saving only the differences between them”.
Capitanio et al are correct in stating that there is a lot of negative perception concerning digital cytopathology. However, digital pathology, especially digital cytopathology, has the potential to reverse the “absence” of pathologist input in medical education as well as to improve the quality and equity of medical educational resource availability. Capitanio et al cite the numerous advantages afforded by digital cytopathology for medical education. Additionally, WSI can be easily incorporated into effective and efficient E-learning platforms. Interactivity when teaching pathology, a feature easily promoted through use of WSI, has been shown to improve engagement, performance in examinations, as well as increase cytopathology diagnostic accuracy skills compared to learning with glass slides and a microscope. Capitanio et al tabulate a helpful range of digital cytopathology educational resources that are available online. More recently, The Royal College of Pathologists of Australasia have also developed a Cytopathology eCase library containing hundreds of cytopathology WSI. Many of the slides have been scanned entirely in the z-axis and can be viewed seamlessly in the z-axis with Sectra Uniview. Access to this educational library is restricted to RCPA trainees and pathologists.
In response to discussions on proficiency testing raised by Capitanio et al, we would also like to respond from the Australian perspective. The Royal College of Pathologists of Australasia Quality Assurance Programs (RCPAQAP) has been offering proficiency programs globally in gynaecological and non-gynaecological cytopathology for more than 20 years, with WSI cases being introduced as a mandatory part of the non-gynaecological cytopathology modules since 2011.
Negative perception towards digital cytopathology can hopefully be partially addressed by advancing z-axis technology. Complex computational filtering of cellular material from multiple focal planes with incorporation into WSI with a single focal plane, may hold promise. We certainly still need a convenient way of annotating z-stacked WSI. Improvements in viewing technology as well as workstation ergonomics will address some of the current barriers to efficient screening of cytopathology WSI. Modifying pre- and post-scanning workflow may also be helpful. Such improvement are already making a difference to education and proficiency testing.