Telepathology diagnosis of prostate needle biopsies

We conducted a prospective analysis of the diagnostic accuracy of a static-image telepathology system (Roche RIAS, Elon College, NC) in the interpretation of needle biopsies of the prostate (NBx). Two hundred consecutive cases received in consultation were included. Each case was examined by one of the researchers (MHW), and images were captured either according to the areas of concern designated by the referring pathologist (set A; 100 cases) or according to the judgment of MHW (set B; 100 cases). The other researcher (JIE) daily rendered diagnoses first on the video images and then by direct microscopy. Accuracy of video diagnosis was categorized as 0 (correct), t (minor error), 2 (major error), or 3 (deferred). An average of 5.49 images were captured per case in set A, and 5.28 for set B. Seventy-seven, 9, 9, and 5 cases were categorized as 0, 1, 2, and 3, respectively, for set A, and 78, 17, 1, and 4 cases, respectively, for set B. Video versus direct diagnoses for the type 2 errors were five carcinoma versus markedly atypical, two carcinoma versus atypical, one carcinoma versus nonspecific granulomatous prostatitis, and two benign versus atypical. In these difficult NBx, telepathology allowed an essentially correct diagnosis in almost all of the cases. The number of images required was reasonable, and the images were of excellent quality. However, the accuracy varied from set A to set B, with the fractions of nondeferred cases that were given an essentially correct video diagnosis totaling 91% and 99%, respectively (P < .01). Accuracy of telepathology diagnosis using static images may depend on the person capturing the images, even in the case of small biopsies.