Accuracy of optical spectroscopy for the detection of cervical intraepithelial neoplasia without colposcopic tissue information; a step toward automation for low resource settings

Abstract

Optical spectroscopy has been proposed as an accurate and low-cost alternative for detection of cervical intraepithelial neoplasia. We previously published an algorithm using optical spectroscopy as an adjunct to colposcopy and found good accuracy (sensitivity=1.00 [95% confidence interval (CI)=0.92 to 1.00], specificity=0.71 [95% CI=0.62 to 0.79]). Those results used measurements taken by expert colposcopists as well as the colposcopy diagnosis. In this study, we trained and tested an algorithm for the detection of cervical intraepithelial neoplasia (i.e., identifying those patients who had histology reading CIN 2 or worse) that did not include the colposcopic diagnosis. Furthermore, we explored the interaction between spectroscopy and colposcopy, examining the importance of probe placement expertise. The colposcopic diagnosis-independent spectroscopy algorithm had a sensitivity of 0.98 (95% CI=0.89 to 1.00) and a specificity of 0.62 (95% CI=0.52 to 0.71). The difference in the partial area under the ROC curves between spectroscopy with and without the colposcopic diagnosis was statistically significant at the patient level (p=0.05) but not the site level (p=0.13). The results suggest that the device has high accuracy over a wide range of provider accuracy and hence could plausibly be implemented by providers with limited training.

Fig. 1

Current and proposed pathways for screening and diagnosis of cervical cancer.

Fig. 2

By-patient boxplot of spectroscopy scores on the test set by histologic diagnosis for (a) combined screening and diagnostic populations (second-generation device); (b) screening population (first- and second-generation device); and (c) diagnostic population (first- and second-generation device).

Fig. 3

By-patient receiver operating characteristic curve analysis of logistic regression algorithm using spectroscopy with and without the colposcopic diagnosis on the second-generation device test set. The partial AUCs for sensitivity between 0.8 and 1.0 were statistically significantly different ($p=0.05$).

Fig. 4

By-site receiver operating characteristic curve analysis of logistic regression algorithm using spectroscopy with and without the colposcopic diagnosis on the second-generation device test set. The partial AUCs for sensitivity between 0.8 and 1.0 were not statistically significantly different ($p=0.13$).

Fig. 5

(a) By-site and (b) by-patient receiver operating characteristic curve analysis of logistic regression algorithm using spectroscopy without the colposcopic diagnosis, using both first- and second-generation devices, on test set by study population.

Fig. 6

By-patient receiver operating characteristic curve analysis of logistic regression algorithm using spectroscopy without the colposcopic diagnosis, with and without the biographical variables on the second-generation device test set.

Fig. 7

Scatter plot of the accuracy of colposcopy (using the whole data set) compared with the accuracy of colpscopically directed but colposcopic diagnosis-independent spectroscopy (using the second-generation device test set). The crosshairs on each point give the 95% confidence interval—the horizontal line is the confidence interval for the accuracy of colposcopy and the vertical line is for spectroscopy.

Fig. 8

Evolution of cancer imaging devices for high-resource settings: point-probe spectroscopy (FastEEM) and multispectral digital colposcope (MDC). The FastEEM2 and FastEEM3 are the first- and second-generation devices, respectively.

Fig. 9

Proposed cancer-imaging devices for low-resource settings.