Diagnostic agreement and usability issues for a web-based proficiency testing system in cervical screening

Federico Viel 1, Vincenzo Della 1, Francesca Demichelis 2, Jonni Santi 2, Paolo Dalla Palma 2, Carlo Alberto Beltrami 1
1 University of Udine, 2 City Hospital of Trento, Italy

Introduction: Quality control is crucial in cervical screening programmes. This can bring the form of integrated continuing education and proficiency testing initiatives for the personnel involved (cytotechnicians, biologists and pathologists). In 1996, the European Federation of Cytology Societies has established a Committee For Quality Assurance, Training And Education (QUATE), which developed a method for proficiency testing, according to which a subject should answer a written test, screen 10 full PAP smears in two hours and diagnose on 20 fixed microscope fields. As the glass slides should be the same for all the subjects, the test is usually carried out during congresses, by using 20 microscopes each of which hosting a specific glass slide; the subjects pass from one microscope to the other, in sequence. Their organisation details make it difficult to carry out many tests. A solution is given by digital slides (1), i.e., digital copies of glass slides. They can be copied and effectively distributed, differently from what happens with glass slides, which are fragile and, in case of cytology, definitely unique. This may allow for a more effective implementation of quality control programmes, either in cytology as well as in histology (2). Among the disadvantages, digital slides are really large, being composed by thousands of images. The present paper describes a preliminary trial on the use of digital slides for proficiency testing.

Methods: 6 PAP smear glass slides and 6 thin layer slides have been digitally acquired at the institute of Pathology, Udine; 4 per type were chosen for the trial, while the others were used for training. PAP smears have an area of 21x45mm; thin layer samples are spread on a circle of 19 mm diameter. Due to the different physical sizes of the samples, the average number of images per digital slide varied between 6000 and 30000. For simplicity, the cases were then burnt on a DVD, together with a specific web-based viewer. The viewer was also able to send data about interface usage to a server on the Web (3). The viewer can be used with a recent Web browser, as it is written in HTML and Javascript. Six cytotechnicians of the City Hospital of Trento have been trained in advance to use the digital slide viewer, by means of a lecture illustrating the features of the software and of the trial scope, and by giving access to the training cases for two weeks. During the trial, the computers on the internal network of the Institute of Pathology of Trento have been used, and their features (processor, memory, monitor size) were registered. cases were available on a local server. The diagnoses given by the cytotechnicians have been gathered using the QUATE forms. At the end, each subject has been interviewed about the system features. The diagnoses were compared with those given on glass slides at the Institute of Pathology of Udine. The time needed for diagnosis was evaluated too.

Results: Diagnostic agreement was obtained in all but one case, for which a minor discordance has been found. The time needed to complete a diagnosis on a digital slide is much higher than on glass slide (20 minutes for thin layer slides, about 28 for PAP smears, versus 5-10 for a glass slide). This is partially due to the subjects habits, much more acquainted with microscopes than computers, partially on the network performance, and partially on the interface features (which will be modified after the experiment).