PERFORMANCE OF SMARTPHONE-BASED DIGITAL IMAGES FOR CERVICAL CANCER SCREENING IN A LOW-RESOURCE CONTEXT

Abstract

Colposcopes are expensive, heavy, and need specialized technical service, which may outreach the capacity of low-resource settings. Our aim was to assess the performance of smartphone-based digital images for the detection of cervical intraepithelial neoplasia grade 2 or worse (CIN2+). Human papillomavirus (HPV)-positive women recruited through a cervical cancer screening campaign had VIA/VILI assessment (visual inspection after application of acetic acid/lugol's iodine). Cervical digital images were captured with a smartphone camera, randomly coded with no prior selection and distributed on an online database (Google Forms) for evaluation. Healthcare providers were invited to evaluate the images and identify CIN2+. The gold standard was the histopathological diagnosis. The sensitivity and specificity for the detection of CIN2+ was assessed for each reader and reported with the 95 percent confidence interval (Clopper-Pearson method). One hundred twenty-five consecutive HPV-positive women were included, with 19 CIN2+ (15.2 percent). Forty-five gynecologists completed the assessment, one-third were considered as experts (>50 colposcopies) and two-thirds as novices (<50 colposcopies). The sensitivity and specificity for CIN 2+ detection was 71.3 percent (67.0-75.7 percent) and 62.4 percent (57.5-67.4 percent), respectively. The performance of novices and experts was similar. The readers assessed 73.1 percent of images as acceptable for diagnostic. Smartphone-based digital images, with its high portability, have a great potential for the diagnosis of CIN2+ in low-resource context.