Diagnosis of dermatopathology patients using whole-slide images on a digital microscope demonstrated noninferiority to diagnosis with glass slides from traditional microscopy.
The retrospective study included 499 previously diagnosed patients accrued from an independent, national, university-affiliated dermatopathology laboratory. Parameters for diagnosis included image resolution, with a focus on eosinophils in inflammatory cases and mitotic figures in melanomas.
Three dermatopathologists established an interobserver ground-truth consensus for each of the 499 patients. Three other dermatopathologists evaluated the patients with both whole-slide images and traditional microscopy, with at least a 30-day washout period between the methodologies. The researchers calculated intraobserver concordance between whole-slide images and traditional microscopy, and interobserver concordance.
Results indicated that the mean intraobserver concordance between the two imaging methodologies was 94%. Similarly, the concordance was 94% between whole-slide images and ground-truth consensus, and between traditional microscopy and ground-truth consensus.
The mean interobserver concordance between the two imaging methodologies and the ground-truth consensus was 91%.
Whole-slide image diagnoses demonstrated noninferiority to those made by traditional microscopy. Whole-slide image read rates corresponded with experience with that imaging methodology. The most experienced user of whole-slide images could reach parity with those done by traditional microscopy.
“This study supports the viability of [whole-slide images] for primary diagnosis in the clinical setting,” the researchers concluded. – by Rob Volansky
Disclosure: One author reports being minority owner of Clearpath, the Laboratory Information System (LIS) used to view and report the scanned whole-slide images.