Making Cytological Diagnoses on Digital Images Using the iPath Network

Background: The iPath telemedicine platform Basel is mainly used for histological and cytological consultations, but also serves as a valuable learning tool. Aim: To study the level of accuracy in making diagnoses based on still images achieved by experienced cytopathologists, to identify limiting factors, and to provide a cytological image series as a learning set. Method: Images from 167 consecutive cytological specimens of different origin were uploaded on the iPath platform and evaluated by four cytopathologists. Only wet-fixed and well-stained specimens were used. The consultants made specific diagnoses and categorized each as benign, suspicious or malignant. Results: For all consultants, specificity and sensitivity regarding categorized diagnoses were 83-92 and 85-93%, respectively; the overall accuracy was 88-90%. The interobserver agreement was substantial (κ = 0.791). The lowest rate of concordance was achieved in urine and bladder washings and in the identification of benign lesions. Conclusion: Using a digital image set for diagnostic purposes implies that even under optimal conditions the accuracy rate will not exceed to 80-90%, mainly because of lacking supportive immunocytochemical or molecular tests. This limitation does not disqualify digital images for teleconsulting or as a learning aid. The series of images used for the study are open to the public at http://pathorama.wordpress.com/extragenital-cytology-2013/.