Virtual microscopy: an evaluation of its validity and diagnostic performance in routine histologic diagnosis of skin tumors

Summary

Digitization of histologic slides is associated with many advantages, and its use in routine diagnosis holds great promise. Nevertheless, few articles evaluate virtual microscopy in routine settings. This study is an evaluation of the validity and diagnostic performance of virtual microscopy in routine histologic diagnosis of skin tumors. Our aim is to investigate whether conventional microscopy of skin tumors can be replaced by virtual microscopy. Ninety-six skin tumors and skin-tumor–like changes were consecutively gathered over a 1-week period. Specimens were routinely processed, and digital slides were captured on Mirax Scan (Carl Zeiss MicroImaging, Göttingen, Germany). Four pathologists evaluated the 96 virtual slides and the associated 96 conventional slides twice with intermediate time intervals of at least 3 weeks. Virtual slides that caused difficulties were reevaluated to identify possible reasons for this. The accuracy was 89.2% for virtual microscopy and 92.7% for conventional microscopy. All κ coefficients expressed very good intra- and interobserver agreement. The sensitivities were 85.7% (78.0%-91.0%) and 92.0% (85.5%-95.7%) for virtual and conventional microscopy, respectively. The difference between the sensitivities was 6.3% (0.8%-12.6%). The subsequent reevaluation showed that virtual slides were as useful as conventional slides when rendering a diagnosis. Differences seen are presumed to be due to the pathologists' lack of experience using the virtual microscope. We conclude that it is feasible to make histologic diagnosis on the skin tumor types represented in this study using virtual microscopy after pathologists have completed a period of training. Larger studies should be conducted to verify whether virtual microscopy can replace conventional microscopy in routine practice.

Introduction

Whole slide imaging (WSI) is the process in which entire histologic or cytologic slides are digitalized at high resolution. In relation to WSI, virtual microscopy enables review of all regions of interest within the digitalized slide at different magnifications. This is done by using a personal computer with relevant software [1], [2].

When making histologic diagnoses on virtual slides in routine settings, the advantages are numerous. The advantages rely on the ability to send and store slides digitally that leads to convenient access without regard to the location of the pathologist. This enables easy access to second opinions, digital conferences, and decentralized primary diagnostic review. Digital storage allows integration of digital slides into the patient's electronic journal as well as easy access to archived slides.

The use of digital slides has also been proven beneficial in quality assurance and teaching [3], [4], [5], [6].

WSI systems on the market are becoming increasingly fast, capable, and reliable. Recent studies indicate that virtual slides are equivalent to conventional slides when making histologic diagnoses [2], [6], [7], [8]. Currently, the application could be limited by factors such as large file sizes, for example, affecting image transmission speeds [1]. However, the rapid development of computer and internet technologies is expected to eliminate such barriers, making virtual microscopy feasible in near future.

Still, there are very few articles evaluating virtual microscopy in routine settings; and to our knowledge, there are no studies gathering slides consecutively, reflecting the case selection in routine practice. It is therefore unanswered whether virtual microscopy is ready for routine application. To address this question, controlled validation studies in routine settings are needed. In addition, such studies are important to identify limitations of the technology that can guide further development and define areas of future focus.

The present study is an evaluation of the validity and diagnostic performance of virtual microscopy when making routine histologic diagnoses on skin tumors. The evaluation seeks to clarify whether image quality and image formation are sufficient when using virtual microscopy in a routine setting.

The aim of the study is to investigate whether the conventional microscopy of skin tumors in routine practice can be replaced by virtual microscopy.