The assessment of an endocervical component in cervicovaginal smears with the PAPNET system

The use of neural network systems is on the rise. Among these systems is PAPNET which was recently approved by the FDA to re-evaluate negative cervicovaginal smears previously diagnosed by manual evaluation. The neural network of this system is used to locate representative samples for review and triage by cytotechnologists. The Bethesda System (TBS), requires the presence of an endocervical component (ECC) for smear adequacy, consisting of either endocervical cells and/or squamous metaplastic cells. The purpose of this study is to evaluate the reliability of this system in selecting and displaying an endocervical component. In addition, the evaluation of this study will provide an endorsement to previous findings regarding the efficiency of the PAPNET system. This study consisted of 158 cervicovaginal smears from patients with cytologic diagnoses of ASCUS by light microscopy triaged with the PAPNET system for detection of an endocervical component. The racial distribution of this group was as follows: Afro-American (132 cs.), Caucasian (26 cs.). The patients' age ranged from 14 years to 64 years, with a mean of 26 years (n=26.1). Colposcopic biopsy diagnoses confirmed 32 as negative, 100 as low-grade squamous lesions, 26 as high-grade squamous lesions, and none with invasive carcinomas. Cases evaluated by the PAPNET system revealed that an endocervical component was on at least one of the two review screens. The results were as follows: no component-4 cases (2.53%), endocervical cells only-1 case (0.63%), metaplastic cells only-66 cases (41.77%), both metaplastic and endocervical cells-87 cases (55.06%). At least one of the necessary components required for specimen adequacy was found in 97% of the cases reviewed with the PAPNET system. This could be vital in reassuring smear adequacy and prove useful as part of an efficient quality assurance program in cytopathology.