Cervical lesion assessment using real-time microendoscopy image analysis in Brazil: The CLARA study
Brady Hunt, José Humberto Tavares Guerreiro Fregnani, David Brenes, Richard A Schwarz, Mila P Salcedo, Júlio César Possati-Resende, Márcio Antoniazzi, Bruno de Oliveira Fonseca, Iara Viana Vidigal Santana, Graziela de Macêdo Matsushita, Philip E Castle, Kathleen M Schmeler, Rebecca Richards-Kortum, Brady Hunt, José Humberto Tavares Guerreiro Fregnani, David Brenes, Richard A Schwarz, Mila P Salcedo, Júlio César Possati-Resende, Márcio Antoniazzi, Bruno de Oliveira Fonseca, Iara Viana Vidigal Santana, Graziela de Macêdo Matsushita, Philip E Castle, Kathleen M Schmeler, Rebecca Richards-Kortum
Abstract
We conducted a prospective evaluation of the diagnostic performance of high-resolution microendoscopy (HRME) to detect cervical intraepithelial neoplasia (CIN) in women with abnormal screening tests. Study participants underwent colposcopy, HRME and cervical biopsy. The prospective diagnostic performance of HRME using an automated morphologic image analysis algorithm was compared to that of colposcopy using histopathologic detection of CIN as the gold standard. To assess the potential to further improve performance of HRME image analysis, we also conducted a retrospective analysis assessing performance of a multi-task convolutional neural network to segment and classify HRME images. One thousand four hundred eighty-six subjects completed the study; 435 (29%) subjects had CIN Grade 2 or more severe (CIN2+) diagnosis. HRME with morphologic image analysis for detection of CIN Grade 3 or more severe diagnoses (CIN3+) was similarly sensitive (95.6% vs 96.2%, P = .81) and specific (56.6% vs 58.7%, P = .18) as colposcopy. HRME with morphologic image analysis for detection of CIN2+ was slightly less sensitive (91.7% vs 95.6%, P < .01) and specific (59.7% vs 63.4%, P = .02) than colposcopy. Images from 870 subjects were used to train a multi-task convolutional neural network-based algorithm and images from the remaining 616 were used to validate its performance. There were no significant differences in the sensitivity and specificity of HRME with neural network analysis vs colposcopy for detection of CIN2+ or CIN3+. Using a neural network-based algorithm, HRME has comparable sensitivity and specificity to colposcopy for detection of CIN2+. HRME could provide a low-cost, point-of-care alternative to colposcopy and biopsy in the prevention of cervical cancer.
Keywords: cervical cancer prevention; deep learning; diagnostic imaging; high-resolution microendoscopy; point-of-care.
Conflict of interest statement
Conflict of interest
R. Richards-Kortum is an inventor on patents owned by the University of Texas licensed to Remicalm LLC. P. E. Castle has received HPV tests and assays for research at a reduced or no cost from Roche, Becton Dickinson, Cepheid, and Arbor Vita Corporation. No potential conflicts of interest were disclosed by the other authors.
© 2021 UICC.
Figures
Source: PubMed