Using Dynamic Features for Automatic Cervical Precancer Detection

Abstract

Cervical cancer remains a major public health concern in developing countries due to financial and human resource constraints. Visual inspection with acetic acid (VIA) of the cervix was widely promoted and routinely used as a low-cost primary screening test in low- and middle-income countries. It can be performed by a variety of health workers and the result is immediate. VIA provides a transient whitening effect which appears and disappears differently in precancerous and cancerous lesions, as compared to benign conditions. Colposcopes are often used during VIA to magnify the view of the cervix and allow clinicians to visually assess it. However, this assessment is generally subjective and unreliable even for experienced clinicians. Computer-aided techniques may improve the accuracy of VIA diagnosis and be an important determinant in the promotion of cervical cancer screening. This work proposes a smartphone-based solution that automatically detects cervical precancer from the dynamic features extracted from videos taken during VIA. The proposed solution achieves a sensitivity and specificity of 0.9 and 0.87 respectively, and could be a solution for screening in countries that suffer from the lack of expensive tools such as colposcopes and well-trained clinicians.

Study protocol for a two-site clinical trial to validate a smartphone-based artificial intelligence classifier identifying cervical precancer and cancer in HPV-positive women in Cameroon

“…Furthermore, we expect the described trial to confirm that the mean accuracy, precision, sensitivity and specificity of the AVC test outperforms human assessment of VIA, as shown in our preliminary results [ 15 ]. Evidence should be found that the AVC test allows a reliable automation of cervical precancer and cancer detection.…”

“…Then, the contours of the lesions are obtained, delimitating all predicted lesions. The size of the lesions is compared to a predefined threshold (450 pixels) to decide whether the patient is negative or positive [ 15 ].…”

“…Preliminary results indicate that the accuracy of the algorithm was 0.89. The precision, sensitivity, and specificity achieved were 0.93, 0.90, and 0.87, respectively [ 15 ]. These results were obtained based on videos from 44 patients.…”

Smartphone-Based Visual Inspection with Acetic Acid: An Innovative Tool to Improve Cervical Cancer Screening in Low-Resource Setting

“…Bae et al developed a smartphone-based endoscope that acquires and classifies images before and after application of the acetic acid [ 45 ]. Finally, Viñals et al developed an AI algorithm based on smartphone images without any additional component [ 46 ]. In the near future, smartphone mHealth and automated visual evaluations may allow for the automated and accurate detections of CIN and become key elements in cervical cancer screenings.…”

Segmentation of the cervical lesion region in colposcopic images based on deep learning

“…In 2021, Roser et al. ( 15 ) proposed using PCA to reduce the dimensionality of the RGB vector and used an ANN to generate the probability map of the precancerous lesion for each pixel. Then, seed point region growth was used to connect the points exceeding the threshold value to the segment region, and whether HSIL+ was determined according to the size of the lesion region, but HSIL+ had nothing to do with the size of the region.…”