Cytodiagnosis through use of a z-axis video by volunteer observers: a promising tool for external quality assessment

A small-scale experimental study of breast FNA consultation on the internet using Panoptiq

INTRODUCTION

To test the potential for cytopathology consultation using Panoptiq (ViewsIQ, Richmond, BC, Canada; this is a new type of whole-slide image that is made manually and incorporates video content), we investigated its application in the cytopathological diagnosis of cases that were difficult to diagnose by breast fine-needle aspiration (FNA).

MATERIALS AND METHODS

Panoptiq files were created from liquid-based cytology slides prepared by the BD CytoRich Red (BD, Franklin Lakes, NJ) method. The slides were prepared from 23 consecutive samples of breast FNA that had been diagnosed as atypical or suspicious by the Hokkaido Cancer Center, Hokkaido, Japan. Nine volunteer reviewers, who were provided with the URL of the Panoptiq file, the original cytopathological diagnosis, and the clinical information, were asked to classify the cytopathological diagnosis of each case into 4 diagnostic categories (benign, atypical, suspicious, or malignant). We examined the consultation benefit (CB)-how much closer the reviewer's cytopathology diagnosis came to the final histopathological diagnosis than the original cytodiagnosis. The CB scoring system was decided in advance.

RESULTS

All 9 reviewers showed a positive total CB score and 2 reviewers showed a significantly higher CB score (Wilcoxon's signed rank test). The representative diagnosis (ie, the most frequently rendered diagnosis in each case) also showed a significant CB.

CONCLUSIONS

Our small-scale experimental study, in which Panoptiq was used in the diagnosis of cases that were difficult to diagnose definitively by breast FNA, revealed a positive CB score by every reviewer and the representative diagnosis showed a significant CB. The study suggests that Panoptiq could be used for cytopathology consultation.

Comparison of glass slides and various digital-slide modalities for cytopathology screening and interpretation

BACKGROUND

Whole-slide imaging in cytology is limited when glass slides are digitized without z-stacks for focusing. Different vendors have started to provide z-stacking solutions to overcome this limitation. The Panoptiq imaging system allows users to create digital files combining low-magnification panoramic images with regions of interest (ROIs) that are imaged with high-magnification z-stacks. The aim of this study was to compare such panoramic images with conventional whole-slide images and glass slides for the tasks of screening and interpretation in cytopathology.

METHODS

Thirty glass slides, including 10 ThinPrep Papanicolaou tests and 20 nongynecologic cytology cases, were digitized with an Olympus BX45 integrated microscope with an attached Prosilica GT camera. ViewsIQ software was used for image acquisition and viewing. These glass slides were also scanned on an Aperio ScanScope XT at ×40 (0.25 μm/pixel) with 1 z-plane and were viewed with ImageScope software. Digital and glass sides were screened and dotted/annotated by a cytotechnologist and were subsequently reviewed by 3 cytopathologists. For panoramic images, the cytotechnologist manually created digital maps and selected representative ROIs to generate z-stacks at a higher magnification. After 3-week washout periods, panoramic images were compared with Aperio digital slides and glass slides.

RESULTS

The Panoptiq system permitted fine focusing of thick smears and cell clusters. In comparison with glass slides, the average screening times were 5.5 and 1.8 times longer with Panoptiq and Aperio images, respectively, but this improved with user experience. There was no statistical difference in diagnostic concordance between all 3 modalities. Users' diagnostic confidence was also similar for all modalities.

CONCLUSIONS

The Aperio whole-slide scanner with 1 z-plane scanning and the Panoptiq imaging system with z-stacking are both suitable for cytopathology screening and interpretation. However, ROI z-stacks do offer a superior mechanism for overcoming focusing problems commonly encountered with digital cytology slides. Unlike whole-slide imaging, the acquisition of representative z-stack images with the Panoptiq system requires a trained cytologist to create digital files. Cancer Cytopathol 2017;125:701-9. © 2017 American Cancer Society.

Digital imaging in cytopathology

Rapid advances are occurring in the field of cytopathology, particularly in the field of digital imaging. Today, digital images are used in a variety of settings including education (E-education), as a substitute to multiheaded sessions, multisite conferences, publications, cytopathology web pages, cytology proficiency testing, telecytology, consultation through telecytology, and automated screening of Pap test slides. The accessibility provided by digital imaging in cytopathology can improve the quality and efficiency of cytopathology services, primarily by getting the expert cytopathologist to remotely look at the slide. This improved accessibility saves time and alleviates the need to ship slides, wait for glass slides, or transport pathologists. Whole slide imaging (WSI) is a digital imaging modality that uses computerized technology to scan and convert pathology and cytology glass slides into digital images (digital slides) that can be viewed remotely on a workstation using viewing software. In spite of the many advances, challenges remain such as the expensive initial set-up costs, workflow interruption, length of time to scan whole slides, large storage size for WSI, bandwidth restrictions, undefined legal implications, professional reluctance, and lack of standardization in the imaging process.

Z-axis video for cytology database is a useful tool for the case presentation prior to the cytology training workshop

We have built a database on the Internet managing z-axis video for cytology (Zavic), and report on a new style of case discussion supported by the Zavic database. Z-axis video for cytology is a movie file derived from the video recording of a microscopic field with changes in the focus. We used it for the case presentation of EUS-FNA of pancreatic lesions on the Internet prior to a training workshop. The attendees were asked to observe the Zavic and to make diagnoses of 20 cases before the workshop. Fourteen attendees also observed lesions under a microscope on that day, and the results were compared with those of Zavic observation. The evaluation of the Zavic database (DB) was surveyed by a questionnaire. The average number of accurate diagnoses by 46 Zavic observers was 10.8. These accuracies for those who observed both the Zavic and glass slides were 11.57 and 11.43, respectively, for the videos and slides. Compared with Zavic observation alone, the diagnoses with glass slide observation were shifted to a correct diagnosis in two cases, but were shifted to an incorrect diagnosis in two cases. Approximately 60% of Zavic observers replied in the questionnaire that the movies on Zavic DB started to play within 3-4 seconds after clicking the play button. We successfully carried out the new style of case discussion supported by the Zavic DB. It was evaluated favorably by many attendees, who were psychologically still dependent on the glass slide observation.