BACKGROUND: To assess the interobserver and intraobserver variation of Papanicolaou (Pap) smear screening with the computer-assisted (neural network based) PAPNET Testing System in diagnosing cervical smear abnormalities, results of agreement were compared with the interobserver and intraobserver variation of conventional smear analysis. METHODS: Cervical smears obtained from women in 1996 were reevaluated both by conventional light microscopy and with use of the PAPNET Testing System by the same four investigators, and results were compared with the original screening diagnoses obtained by both methods. RESULTS: The interobserver results for epithelial abnormalities (the degree of agreement between the cytologists), characterized by weighted kappa statistics, were 0.71 (95% CI: 0. 68-0.73) for PAPNET screening and 0.69 (95% CI: 0.66-0.72) for conventional screening. No significant differences were found among the individual results obtained by the four cytotechnologists (intraobserver variation) with conventional screening versus PAPNET reviewing. CONCLUSIONS: Pap smear grading with the PAPNET Testing System has interobserver and intraobserver variation similar to that of conventional screening of Pap smears in routine use. Cancer (Cancer Cytopathol) Copyright 1999 American Cancer Society.
BACKGROUND: To assess the interobserver and intraobserver variation of Papanicolaou (Pap) smear screening with the computer-assisted (neural network based) PAPNET Testing System in diagnosing cervical smear abnormalities, results of agreement were compared with the interobserver and intraobserver variation of conventional smear analysis. METHODS: Cervical smears obtained from women in 1996 were reevaluated both by conventional light microscopy and with use of the PAPNET Testing System by the same four investigators, and results were compared with the original screening diagnoses obtained by both methods. RESULTS: The interobserver results for epithelial abnormalities (the degree of agreement between the cytologists), characterized by weighted kappa statistics, were 0.71 (95% CI: 0. 68-0.73) for PAPNET screening and 0.69 (95% CI: 0.66-0.72) for conventional screening. No significant differences were found among the individual results obtained by the four cytotechnologists (intraobserver variation) with conventional screening versus PAPNET reviewing. CONCLUSIONS: Pap smear grading with the PAPNET Testing System has interobserver and intraobserver variation similar to that of conventional screening of Pap smears in routine use. Cancer (Cancer Cytopathol) Copyright 1999 American Cancer Society.