PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 330723

Estimating false positive and false negative error rates in cervical cell classification.

L H Oliver·R S Poulsen·G T Toussaint
View original source ↗Export metadata ↓

Abstract

The performance of a cell recognition system on unknown data is often estimated in terms of its error rates on a test set. This paper investigates methods for producing estimates of error rates in cervical cell classification. Classification performance curves calculated using these methods are given for several classification schemes used to classify 1500 cervical cells.

Explore related subjects

AutoanalysisCell NucleusCervix UteriCytological TechniquesCytoplasmFalse Negative ReactionsFalse Positive ReactionsFemaleHumansUterine Cervical NeoplasmsJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

L H Oliver, R S Poulsen, G T Toussaint. 1977. Estimating false positive and false negative error rates in cervical cell classification.. https://doi.org/10.1177/25.7.330723

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2007

Use of the Agilent 2100 bioanalyzer for rapid and reproducible molecular typing of Streptococcus pneumoniae.

Restriction fragment length polymorphism (RFLP) analysis is an economic and fast technique for molecular typing but has the drawback of difficulties in accurately sizing DNA fragments and comparing banding patterns on agarose gels. We aimed to improve RFLP for typing of the important human pathogen Streptococcus pneumoniae and to compare the results with the commonly used typing techniques of pulsed-field gel electrophoresis and multilocus sequence typing. We designed primers to amplify a noncoding region adjacent to the pneumolysin gene. The PCR product was digested separately with six restriction endonucleases, and the DNA fragments were analyzed using an Agilent 2100 bioanalyzer for accurate sizing. The combined RFLP results for all enzymes allowed us to assign each of the 47 clinical isolates of S. pneumoniae tested to one of 33 RFLP types. RFLP analyzed using the bioanalyzer allowed discrimination between strains similar to that obtained by the more commonly used techniques of pulsed-field gel electrophoresis, which discriminated between 34 types, and multilocus sequence typing, which discriminated between 35 types, but more quickly and with less expense. RFLP of a noncoding region using the Agilent 2100 bioanalyzer could be a useful addition to the molecular typing techniques in current use for S. pneumoniae, especially as a first screen of a local population.

Autoanalysis↗
PubMed2007

A three-armed trial of the ThinPrep Imaging System.

We compared the performance of the ThinPrep (TP) Imaging System (TIS) with manual reading of TP slides (TPM) and with manual reading of the paired conventional Pap smear (PS) in terms of sensitivity for and positive predictive value (PPV) of high-grade disease and productivity. The study consisted of 11,416 routine PS and paired TP slides as well as 103 confirmed abnormal TP slides. In terms of sensitivity for the detection of biopsy-confirmed high-grade disease, overall there was no statistically significant difference between TIS-screened TP (TPI) and TPM (81.1% vs. 86.8%). For the routine cases, TPI was significantly more sensitive than PS (73.4% vs. 57.8%). In terms of PPVs for the cytologic prediction of high-grade disease, there was no statistically, significant difference among TPI, TPM, and PS (75.6%, 73.9%, and 84.6%). For cytologic reports of possible high-grade disease, the PPVs were equivalent for TPI (45.0%) and TPM (37.0%) and there was no significant difference in PPVs between TPI and PS (61.3%). For TP slides, TIS screening showed a 27% productivity gain when compared with manual screening and a 54% productivity gain when compared with manual screening of PS slides. Use of TIS showed productivity benefits when compared with TPM and both productivity and sensitivity benefits over use of PS.

Autoanalysis↗