Diagnostic Accuracy in Intraoperative Consultations with Robotic Microscopy and Whole Slide Imaging
Russell Silowash ; University of Pittsburgh;
Content:
Intraoperative consultation is an important component of anatomic pathology practice. Digital pathology is be useful when a local pathologist is not available. Tissues analyzed during intraoperative consultation are often thicker than the permanent tissue preparations; therefore, these tissue samples may present challenges to current digital slide technologies in providing pathologists with diagnostic grade images. Researchers at the University of Pittsburgh Medical Center are conducted a study evaluating the capabilities of robotic microscopy and whole slide imaging in the analysis of intraoperative consultation tissue samples.
Technology:
Whole Slide Images were created using a Trestle/Zeiss 50-slide loading robotic slide virtualizer paired with a Olympus BX51 microscope, equipped with a Jai 3CCD RGB camera. These digital slides were served from a machine equipped with dual 3 GHz Xeon processor with 4 GB of RAM and Microsoft Windows 2000 Server. Air Force and VA pathologists used an Olympus BX-41 microsocope equipped with a Jai M7 camera. Digital slides viewed with robotic microscopy and whole slide imaging used Trestles MedMicro software. Data was collected using a Microsoft Access database.
Design:
Forty cases randomly selected from University of Pittsburgh Medical Center hospitals were reviewed by the principle investigator, and 20 diagnostic intraoperative slides were identified . Pathologists from multiple sites analyzed these 20 cases using each technology and derived a diagnosis. Data was collected for case complexity, diagnostic confidence, and perceived time-to-complete. Diagnostic accuracy was rated by the principal investigator and statistically analyzed.
Results:
Two University of Pittsburgh Medical Center pathologists, four United States Air Force pathologists, and one pathologist from a veteran affairs hospital completed 109 total cases. These results are preliminary, and diagnostic accuracy results across digital technologies will be reported as well as data regarding diagnostic confidence, case complexity, and perceived time-to-complete.
Conclusion:
Findings for both technologies are promising. However, there may be differences in diagnostic accuracy when robotic microscopy is compared with whole slide imaging. More validation work is required, and the University of Pittsburgh Medical Center is conducting studies involving digital slide interpretation.
