Robotic Interactive Telepathology in proficiency testing / quality assurance schemes

Article type: Research Article

Authors: Leong, Franz Joel W-M | McGee, James OD

Affiliations: University of Oxford, Nuffield Department of Clinical Laboratory Sciences, Level 4, Academic Block, John Radcliffe Hospital, Oxford OX3 9DU, United Kingdom |

Note: [] Address for correspondence: Professor J OD McGee Nuffield Department of Medicine Level 5 John Radcliffe Hospital Oxford OX3 9DU United Kingdom Phone: +44 1865 741166 Fax: +44 1865 220078 Email: james.mcgee@ndm.ox.ac.uk http://www.ndcls.ox.ac.uk/telepathology

Abstract: The U.K. National Breast Screening Programme operates a National External Quality Assurance Scheme (NEQAS) for breast screening pathology. Twelve microscopic glass slides are circulated twice yearly among histopathologists (>400). Confidential answers are submitted centrally, and diagnostic performance of each pathologist is compared with expert breast pathologists diagnosis, ie. consensus diagnosis (CD). Kappa scores for diagnostic accuracy (benign, in situ and invasive cancer) are calculated and transmitted to each participating pathologist. NEQAS has raised kappa scores, eg. cancer grading improved from <0.1 to >0.65 after 6 years. Conventional circulation of glass slides each year to >400 pathologists is logistically cumbersome. Similarly both participating pathologists submitted diagnoses and performance feedback are still sent by conventional mail. Telepathology theoretically represents an alternative modality to slide and paper circulation in histopathology NEQAS. We tested this hypothesis, assessing the technical/clinical effectiveness of robotic interactive telepathology (RITPath) in two trials. The initial study involved a single pathologist examining 87 breast cases chosen randomly from a series of 192 from the U.K. Breast Pathology NEQAS files. The average time taken for telepathology (TP) diagnosis was 3.9 minutes. The diagnostic accuracy of TP compared with light microscopy (LM) was 100%, for all cases benign and malignant. The accuracy of TP versus LM and CD was 91.3% for carcinoma in situ grading (21/23 cases). TP grading of breast cancer achieved accuracy of 97.7% (43/44) compared with LM and CD. In tumour typing, TP accuracy was 97% (65/67) using LM and CD as comparators. The second study of similar design involved 7 pathologists of varying experience, each assessing 20-33 cases. Average diagnostic times per slide were 4.5-6.5 minutes. Concordance with CD was 81-100%. These data demonstrate that telepathology of RITPath type is a clinically and technically effective modality for histopathology quality assurance schemes. Additionally, all RITPath data can be transmitted for analysis without a paper trail. The implications of these data for global quality assurance/professional proficiency testing schemes are discussed.

Keywords: robotic telepathology, breast cancer screening, digital microscopy, quality assurance

Journal: Electronic Journal of Pathology and Histology, vol. 7, no. 2, pp. 01-01, 2001