Digitalizing the Art & Science of Cytology

Change is inevitable. Humans naturally resist change, and it is a leader’s responsibility to help inspire their followers to embrace the continuous improvement mindset and become change agents. Clearly, this topic is dear to me as it fueled my dissertation. In laboratory medicine, we’re privy to some remarkable technologies, yet our fear of being replaced by technology often outweighs our curiosity. I can promise you, at least in our lifetime, laboratory professionals will not and cannot be replaced by technology. I remember back in my cytology program (yes, it’s been a decade, and no, we don’t need to discuss that further), seasoned cytologists were panicking about molecular testing. They’d say, “we’re done. They’re not going to need us anymore! Look for another career because you’re wasting your time here.” I’m not sure why I never believed them, or maybe I’ve just never been able to fully trust AI and that gave me some sense of security. Regardless, I persevered and fell in love with our field. In school, I learned about telepathology. Users were able to attach a camera to a microscope and capture static and dynamic images of pathology slides for digital archiving and for performing consultations rapid onsite evaluations (ROSE). Institutions were piloting novel yet comprehensive systems for ROSE so that cytologists could attend the biopsy procedures to prepare and maneuver the slides, allowing the pathologist to perform an adequacy assessment remotely. I thought it was fascinating and such a phenomenal use of resources.

Cytology is both an art and a science, as Dr. Richard DeMay so eloquently described. From a scientific standpoint, there has been substantial effort in developing imaging algorithms that systematically capture the unique features of dysplastic and malignant cells and differentiate them from benign-appearing cells. My first experience with this was the ThinPrep Imaging System (TIS), which helps reduce the rate of false-negative pap smears by providing 22 fields of view that may prompt a full manual review. I can comfortably say that, even after more than a decade, this system has not replaced the human eye, but assisted in our primary screening. Now with AI advancement, the Genius Digital Diagnostics System was designed to expand upon computer-assisted screening. The Genius system not only identifies features of dysplasia, but suggests benign components that are fundamental to the overall diagnosis, such has glandular cells and microorganisms. Presenting these fields in a gallery view along with the whole slide digitally imaged enables the user to better classify the whole picture and review cells surrounding the gallery-selected objects of interest. For someone who has missed a single pseudohyphae of candida lurking between a few squamous cells (hi, it’s me), this technology is a game-changer. Again, this is not a substitute for the cytologist because the cytologist is still responsible for primary screening and rendering a diagnosis. With that said, technology is not perfect. To err is to human, to fault is to… technology. But in a world where cytologists are afraid that HPV primary testing will replace the need for cytology, the continuous development of digital cytology for gynecologic specimens is ever in our favor. Just like there will always be HPV-negative dysplasias, there will always be cells that technology won’t capture, the skilled art that we practice will always have a critical value in patient care.

On the educational and logistics front, imagine being able to digitally archive study sets from the most unique cases your institution has seen. Whole slide imaging (WSI) and Z-stacking technology enables a user to create and access an expansive digital reference library with the ability to zoom in and focus on fields of view from a computer screen. Yes, I prefer using a microscope just as I prefer reading a paperback book, but the world has adopted e-books, so I’m fairly certain we can adapt to e-slides as well. Regardless, publishing and sharing a digital reference set is beneficial to the field of anatomic pathology, whether you’re studying for a board exam or rendering a diagnosis on an unknown by comparing to previously diagnosed cases elsewhere. Logistically, the same principle applies for pathology consultations. The idea of never losing a patient’s original slide in the mail is titillating. Digitalizing pathology slides for consultation and sending the file to another institution through a secure server is more efficient yet just as diagnostic as a traditional consultation through the mail.

But wait! There’s more. AI technology can help us interpret ancillary tests, such as FISH/UroVysion. It is still up to us to agree with the machine’s classifications, but tools like this can ideally reduce turnaround time. Pathologists are already familiarizing themselves with algorithms for IHC and predictive biomarkers. Larger academic medical centers are securing grants left and right to develop and train both diagnostic and prognostic algorithms. Digital pathology can improve efficiency, reallocate our resources, and serve as an aid.

Granted, these technologies are costly and require ample digital storage space. I’m talking petabytes of data here, and of course, there’s a need for impenetrable security in the cloud. WSI can be tedious and time-consuming, and may even warrant the need for an additional full-time employee (or two). Additionally, many of these technologies require extensive validation. Other than cost and storage, one of the more significant challenges in digital cytology is adequately capturing a variety of cytopreparations. Unlike smooth histology sections, cytopreparations including smears, cytospins, and even liquid-based preparations, preserve (to some extent) the three-dimensional nature of cell groups. While it’s beautiful to be able to manually fine-tune our focus throughout a cluster of cells under the microscope, AI may struggle with capturing and systematically categorizing cells within these groups or clusters. Current technologies that work well for histology slides might be insufficient for cytology slides, which serves as a hurdle for research and development teams and a barrier to users embracing a technology that has not yet been “perfected” for a field.

Digital cytology is relatively young, and like any early technology, there are going to be bumps and hiccups. With that said, the benefits of digital pathology overall far outweigh any possible negatives, and we must continue to move forward. We, as laboratory professionals, cannot slow down and resist the future of our field. We must serve as change agents and reassure our future colleagues that there is a secure place for them pathology and laboratory medicine. This is my call to all of you to let your curiosity take over. Take the plunge and let technology be your ally, your diagnostic companion.

Note: I have no financial interests or relationships to disclose. Opinions are purely my own and are not representative of my employer or ASCP.

-Taryn Waraksa-Deutsch, MS, SCT(ASCP)^CM, CT(IAC), has worked as a cytotechnologist at Fox Chase Cancer Center, in Philadelphia, Pennsylvania, since earning her master’s degree from Thomas Jefferson University in 2014. She is an ASCP board-certified Specialist in Cytotechnology with an additional certification by the International Academy of Cytology (IAC). She is also a 2020 ASCP 40 Under Forty Honoree.