Interobserver Variability in Pulmonary Cytopathology Diagnostics by Whole-slide Digital Cytology-A multi-expert International Study

Why clinicians should know about it

• Picked for Pathology and Forensic Medicine (paper of the day, 23 October 2026): Pulmonary cytology agreement, diagnostic accuracy, and digital workflow

Abstract

BACKGROUND: Digital imaging in cytopathology is transforming information exchange and remote diagnostics. Internet-based platforms and algorithms are increasingly expected to support cyto‑ and histopathological reporting workflows. This study evaluated interobserver concordance, diagnostic accuracy, causes of misdiagnosis, scanning time, and review time among ten expert cytologists using whole slide imaging (WSI). The primary aim was to identify challenges in applying WSI to cytopathology and to propose strategies to address them. METHODS: This retrospective study included 61 intraoperative pulmonary cytology samples (30 women, 31 men; mean age 66.1 years) from the LungenClinic Grosshansdorf acquired between January 2020 and December 2021. A total of 61 Giemsa‑stained preparations (54 imprint/touch preparations, five fine‑needle aspirates, and two sediment smears of pleural effusions) were scanned using a Leica Aperio AT2 scanner and uploaded to an internet platform. Histology served as the reference standard and showed 19 benign and 42 malignant lesions. RESULTS: Sixty‑one cytological cases (19 benign, 42 malignant) were assessed by ten observers. Interobserver agreement for categorical diagnoses was slight (Fleiss' kappa = 0.099), whereas ranking concordance was moderate (Kendall's tau = 0.314). Evaluation times ranged from 20 to 667.8 seconds per case, indicating substantial variability in digital navigation speed. CONCLUSIONS: WSI enables international collaboration and remote review of pulmonary cytology, but substantial variability persists in observer interpretations. WSI is currently feasible mainly for consultation, education, and quality assurance, but it is not yet equivalent to conventional microscopy for primary diagnostics, underlining the need for workflow optimization, cytology‑specific validation, and structured digital training.

Abstract as published, via PubMed.