PAPNET. The human and other dimensions

To study the effect of specimen preparation and observer variation on the accuracy of interpretation of PAPNET images.

The PAPNET scanner makes use of conventional Papanicolaou-stained smears and presents the cytologist with pictures of selected "abnormal" cells. This minitrial was designed to investigate the human variability in different grades of personnel and to consider any problems caused by the use of conventional smears.

The personnel involved consisted of a consultant cytopathologist, a senior chief medical laboratory scientific officer and two senior cytotechnologists, one from another laboratory. Experience with the system ranged from six months to a few years. The best interobserver agreement was found between the senior chief and the senior technologist, who worked in the same laboratory. Majority agreement with the "gold standard" (at least three out of four) was highest for the extremes of "negative" and "high grade" smears. Three high grade smears were "missed" due to the quality of the smears and scantiness of "abnormal" cells.

Although fatigue in screening is reduced by the interactive mode, it does not overcome the subjectivity of human examination both in review of the dat tape and in microscopic examination of selected cells on the slides. Experience with the method is important, and different background experiences will have an effect. The technical quality of the smear is important since the automated scanner cannot be expected to select abnormal cells if they are absent or obscured for any reason. A monolayer may be the solution, although at greater cost.