Year: 2022
Type: article
Abstract: within the laboratory information system to a vendor whole slide image viewer.Remote review utilized consumer-grade computers and monitors connecting to an institution clinical workstation via secure virtual private network.A third lymphoma pathologist conducted digital diagnostic review of whole slide images, who was blinded to the diagnoses rendered by the primary pathologists.Sequential steps for digital diagnosis by the third pathologist (a) H&E whole slide images reviewed for each case (b) recording the choice of immunohistochemistry markers by third pathologist (c) evaluation of the digitally scanned immunohistochemistry markers performed (d) comparison of utilization and interpretation of ancillary studies for each case between the primary pathologists and the third pathologist in relation to digital pathology diagnosis (e) correlation of final diagnosis (f) follow-up light microscopic evaluation was performed by the third pathologist on all the cases after two months.The glass slides were de-identified for light microscopy, with the pathology resident holding key to connect the cases.Results: Digital images were of high quality to enable diagnostic evaluation in all the cases, with 54/60 cases showing digital to glass slide intraobserver concordance.Final diagnosis rendered in 6/60 cases using digital evaluation required further clarification by light microscopy, with concordance.Interestingly, all the 6 cases were of reactive etiology.Utilization of ancillary studies by primary pathologists using light microscopy and the third pathologist using digital platform showed (1) immunohistochemistry: 11.3 vs 9.3 average (2) Flow cytometry studies performed in 34/60 versus 0/60 cases (3) no discordance in utilization of molecular studies and fluorescent in situ hybridization.Conclusion: Our experience shows high intraobserver diagnostic concordance between digital and glass slides, without compromising optimal patient care.Digital evaluation of reactive lymphoid proliferations requires longer learning curve than lymphomas.Utilization of ancillary studies was consistent with the diagnostic algorithms followed by individual pathologists in routine practice and was not influenced adversely by digital pathology.
Source: Journal of Pathology Informatics
Authors: Ahsan Siddiqi, Martin Matzuk, John Joseph, Amy Zingalis, Sunil Singhal, Ramya Masand, Carolina Gutierrez, Bettye Cox, Chen Chen, Shima Mousavi, Hafeez Diwan, Harold Unruh, Johnny Lee, Maria Singhal, Lee Leiber, Chad Howard, Joe Doty, Damian Stutz, Stephanie Cravens, Mark Frushone, Francis Gannon
Institution: Baylor College of Medicine
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,400
OpenAlex ID: W4293217091