A prospective randomized study on limits of colposcopy and histology: the skill of colposcopist and colposcopy-guided biopsy in diagnosis of cervical intraepithelial lesions

Development and validation of artificial intelligence-based analysis software to support screening system of cervical intraepithelial neoplasia

Cervical cancer, the fourth most common cancer among women worldwide, often proves fatal and stems from precursor lesions caused by high-risk human papillomavirus (HR-HPV) infection. Accurate and early diagnosis is crucial for effective treatment. Current screening methods, such as the Pap test, liquid-based cytology (LBC), visual inspection with acetic acid (VIA), and HPV DNA testing, have limitations, requiring confirmation through colposcopy. This study introduces CerviCARE AI, an artificial intelligence (AI) analysis software, to address colposcopy challenges. It automatically analyzes Tele-cervicography images, distinguishing between low-grade and high-grade lesions. In a multicenter retrospective study, CerviCARE AI achieved a remarkable sensitivity of 98% for high-risk groups (P2, P3, HSIL or higher, CIN2 or higher) and a specificity of 95.5%. These findings underscore CerviCARE AI's potential as a valuable diagnostic tool for highly accurate identification of cervical precancerous lesions. While further prospective research is needed to validate its clinical utility, this AI system holds promise for improving cervical cancer screening and lessening the burden of this deadly disease.

Colposcopy Accuracy and Diagnostic Performance: A Quality Control and Quality Assurance Survey in Italian Tertiary-Level Teaching and Academic Institutions-The Italian Society of Colposcopy and Cervico-Vaginal Pathology (SICPCV)

Quality Control (QC) and Quality Assurance (QA) principles are essential for effective cervical cancer prevention. Being a crucial diagnostic step, colposcopy's sensitivity and specificity improvements are strongly advocated worldwide since inter- and intra-observer differences are the main limiting factors. The objective of the present study was the evaluation of colposcopy accuracy through the results of a QC/QA assessment from a survey in Italian tertiary-level academic and teaching hospitals. A web-based, user-friendly platform based on 100 colposcopic digital images was forwarded to colposcopists with different levels of experience. Seventy-three participants were asked to identify colposcopic patterns, provide personal impressions, and indicate the correct clinical practice. The data were correlated with a panel of experts' evaluation and with the clinical/pathological data of the cases. Overall sensitivity and specificity with the threshold of CIN2+ accounted for 73.7% and 87.7%, respectively, with minor differences between senior and junior candidates. Identification and interpretation of colposcopic patterns showed full agreement with the experts' panel, ranging from 50% to 82%, in some instances with better results from junior colposcopists. Colposcopic impressions correlated with a 20% underestimation of CIN2+ lesions, with no differences linked to level of experience. Our results demonstrate the good diagnostic performance of colposcopy and the need for improving accuracy through QC assessments and adhesion to standard requirements and recommendations.

Accuracy of colposcopy in the Swedish screening program

INTRODUCTION

Sensitivity and specificity of colposcopy vary greatly between studies and efficacy in clinical studies seldom corresponds with effectiveness in a real-life setting. It is unclear whether colposcopists' experience affects assessment; studies show divergent results. The study's objective was to investigate the accuracy of colposcopies in the Swedish screening program, the variability in colposcopists' assessments and whether degree of experience affects accuracy in a routine setting.

MATERIAL AND METHODS

Cross-sectional register study. All colposcopic assessments with a concomitant histopathological sample from women aged at least 18 years, performed between 1999 and September 2020 in Sweden. The main outcome measure was accuracy. The accuracy of colposcopic assessments was calculated as overall agreement with linked biopsies, with three outcomes: Normal vs Atypical, Normal vs Low-Grade Atypical vs High-Grade Atypical, and Non-High-Grade Atypical vs High-Grade Atypical. A time-trend analysis was performed. The accuracy of identifiable colposcopists related to experience was analyzed.

RESULTS

In total, 82 289 colposcopic assessments with linked biopsies were included for analysis of the outcome Normal vs Atypical; average accuracy was 63%. Overrating colposcopic findings was four times more common than underrating. No time trend in accuracy was noted during the study period. Accuracy in distinguishing High-Grade from Non-High-Grade lesions was better: 76%. Among identifiable colposcopists, overall accuracy was 67%. Some had significantly better accuracy than others, but no correlation with experience was found.

CONCLUSIONS

Colposcopy, including in a referral setting, has low accuracy in distinguishing Normal from Atypical. Increased experience alone does not lead to improvement. This is supported by the substantial differences in performance between colposcopists.

Clinical performance of primary HPV screening cut‐off for colposcopy referrals in HPV vaccinated cohort: observational study

OBJECTIVE

To understand the effect of changing from cytology-based to primary HPV screening on the positive predictive value (PPV) of colposcopy referrals for cervical intraepithelial neoplasia (CIN) in a cohort offered HPV vaccination.

DESIGN

Retrospective pre/post observational cohort study.

SETTING

Scotland.

POPULATION OR SAMPLE

2193 women referred to colposcopy between September 2019 and February 2020 from cytology-based screening and between September 2020 and February 2021 from primary high-risk HPV (hrHPV) screening.

METHODS

Calculating positive predictive values (PPVs) for two cohorts of women; one having liquid-based cytology screening and the other, the subsequent hrHPV cervical screening as a pre/post observational study.

MAIN OUTCOME MEASURES

Positive predictive values of LBC and hrHPV cut-offs for colposcopy referral for CIN at colposcopy.

RESULTS

Three papers fitted our criteria; these reported results only for cytology-based screening. The PPV was lower for women in HPV-vaccinated cohorts indicating a lower prevalence of disease. Vaccination under the age of 17 had the lowest PPV reported. Scottish colposcopy data concerning hrHPV and cytology showed a non-significant difference between PPV (17.5%, 95% CI 14.3-20.7, and 20.6, 95% CI 16.7-24.5, respectively) for referrals with a cut-off of low grade dyskaryosis (LGD); both met the standard set of 8-25%. The hrHPV PPV (66.7, 95% CI 56.8-76.6) was comparable to cytology (64.1, 95% CI 55.8-72.4) for referrals with a cut-off of high grade dyskaryosis (HGD) but neither met the standard set of 77-92%.

CONCLUSIONS

Current literature only provides PPVs for LBC and, overall, the vaccinated cohort had lower PPVs. Only LG dyskaryosis met PHE criteria. The PPV for HPV-vaccinated women undergoing either LBC or HR-HPV screening were not statistically different. However, similar to papers in the current literature, HG dyskaryosis (HGD) PPVs of both techniques did not meet the PHE threshold of 76.6-91.6% outlined in the cervical standards data report.

Comparison of accuracy and reproducibility of colposcopic impression based on a single image versus a two-minute time series of colposcopic images

OBJECTIVE

Colposcopy is an important part of cervical screening/management programs. Colposcopic appearance is often classified, for teaching and telemedicine, based on static images that do not reveal the dynamics of acetowhitening. We compared the accuracy and reproducibility of colposcopic impression based on a single image at one minute after application of acetic acid versus a time-series of 17 sequential images over two minutes.

METHODS

Approximately 5000 colposcopic examinations conducted with the DYSIS colposcopic system were divided into 10 random sets, each assigned to a separate expert colposcopist. Colposcopists first classified single two-dimensional images at one minute and then a time-series of 17 sequential images as 'normal,' 'indeterminate,' 'high grade,' or 'cancer'. Ratings were compared to histologic diagnoses. Additionally, 5 colposcopists reviewed a subset of 200 single images and 200 time series to estimate intra- and inter-rater reliability.

RESULTS

Of 4640 patients with adequate images, only 24.4% were correctly categorized by single image visual assessment (11% of 64 cancers; 31% of 605 CIN3; 22.4% of 558 CIN2; 23.9% of 3412 < CIN2). Individual colposcopist accuracy was low; Youden indices (sensitivity plus specificity minus one) ranged from 0.07 to 0.24. Use of the time-series increased the proportion of images classified as normal, regardless of histology. Intra-rater reliability was substantial (weighted kappa = 0.64); inter-rater reliability was fair ( weighted kappa = 0.26).

CONCLUSION

Substantial variation exists in visual assessment of colposcopic images, even when a 17-image time series showing the two-minute process of acetowhitening is presented. We are currently evaluating whether deep-learning image evaluation can assist classification.

The Performance of Artificial Intelligence in Cervical Colposcopy: A Retrospective Data Analysis

Objective

We aimed to evaluate the performance of artificial intelligence (AI) system in detecting high-grade precancerous lesions.

Methods

A retrospective and diagnostic study was conducted in Chongqing Cancer Hospital. Anonymized medical records with cytology, HPV testing, colposcopy findings with images, and the histopathological results were selected. The sensitivity, specificity, and areas under the curve (AUC) in detecting CIN2+ and CIN3+ were evaluated for the AI system, the AI-assisted colposcopy, and the human colposcopists, respectively.

Results

Anonymized medical records from 346 women were obtained. The images captured under colposcopy of 194 women were found positive by the AI system; 245 women were found positive either by human colposcopists or the AI system. In detecting CIN2+, the AI-assisted colposcopy significantly increased the sensitivity (96.6% vs. 88.8%, p=0.016). The specificity was significantly lower for AI-assisted colposcopy (38.1%), compared with human colposcopists (59.5%, p < 0.001) or the AI system (57.6%, p < 0.001). The AUCs for the human colposcopists, AI system, and AI-assisted colposcopy were 0.741, 0.765, and 0.674, respectively. In detecting CIN3+, the sensitivities of the AI system and AI-assisted colposcopy were not significantly higher than human colposcopists (97.5% vs. 92.6%, p=0.13). The specificity was significantly lower for AI-assisted colposcopy (37.4%) compared with human colposcopists (59.2%, p < 0.001) or compared with the AI system (56.6%, p < 0.001). The AUCs for the human colposcopists, AI system, and AI-assisted colposcopy were 0.759, 0.674, and 0.771, respectively.

Conclusions

The AI system provided equally matched sensitivity to human colposcopists in detecting CIN2+ and CIN3+. The AI-assisted colposcopy significantly improved the sensitivity in detecting CIN2+.

Development of a prognostic prediction support system for cervical intraepithelial neoplasia using artificial intelligence-based diagnosis

Objective

Human papillomavirus subtypes are predictive indicators of cervical intraepithelial neoplasia (CIN) progression. While colposcopy is also an essential part of cervical cancer prevention, its accuracy and reproducibility are limited because of subjective evaluation. This study aimed to develop an artificial intelligence (AI) algorithm that can accurately detect the optimal lesion associated with prognosis using colposcopic images of CIN2 patients by utilizing objective AI diagnosis.

Methods

We identified colposcopic findings associated with the prognosis of patients with CIN2. We developed a convolutional neural network that can automatically detect the rate of high-grade lesions in the uterovaginal area in 12 segments. We finally evaluated the detection accuracy of our AI algorithm compared with the scores by multiple gynecologic oncologists.

Results

High-grade lesion occupancy in the uterovaginal area detected by senior colposcopists was significantly correlated with the prognosis of patients with CIN2. The detection rate for high-grade lesions in 12 segments of the uterovaginal area by the AI system was 62.1% for recall, and the overall correct response rate was 89.7%. Moreover, the percentage of high-grade lesions detected by the AI system was significantly correlated with the rate detected by multiple gynecologic senior oncologists (r=0.61).

Conclusion

Our novel AI algorithm can accurately determine high-grade lesions associated with prognosis on colposcopic images, and these results provide an insight into the additional utility of colposcopy for the management of patients with CIN2.

High-grade lesion occupancy in the uterovaginal area was significantly correlated with CIN2 patients’ prognosis. The number of high-grade lesions in 12 segments detected by an artificial intelligence (AI)-based system was comparable to that detected by senior colposcopists. The overall correct response rate of the AI algorithm for detecting high-grade lesions was 89.7%.

Value of multi-quadrants biopsy: Pooled analysis of 11 population-based cervical cancer screening studies

OBJECTIVE

The accuracy of colposcopy-guided biopsy is key to the success of colposcopic triage in cervical cancer screening programs. However, there is no widely adopted biopsy guideline up to date. Our study aimed to determine whether multi-quadrants biopsy improves the yield of cervical lesions.

METHODS

Eleven population-based cervical cancer screening studies were conducted in China. Cytology, high-risk human papillomavirus (hrHPV) testing and visual inspection were performed for primary screening. Females positive on one or more tests were referred for colposcopy and biopsy. The proportion of detected cervical intraepithelial neoplasia (CIN)2+ and yields by quadrant lesion-targeted biopsy or 4-quadrant random biopsy were compared.

RESULTS

Among 4,923 females included, 1,606 had quadrant lesion-targeted biopsy, and 3,317 had 4-quadrant random biopsy. The cumulative CIN2+ yield increased from 0.10 for only one quadrant-targeted biopsy to 0.21, 0.34, and 0.58 for at most two, three and four quadrants targeted biopsies. Among hrHPV positive females with high-grade squamous intraepithelial lesion (HSIL)+ cytology, the cumulative CIN2+ yield of a second targeted biopsy in another quadrant was significantly increased (P<0.05). Among hrHPV-negative females, the yield of 4-quadrant random biopsies was 0.005, and the yield by lesion-targeted biopsies was 0.017. For hrHPV positive females who had 4-quadrant random biopsy, the additional CIN2+ yield for HSIL+, low-grade squamous intraepithelial lesion (LSIL) cytology, or abnormal visual inspection via acetic acid and Lugol's iodine (VIA/VILI) were 0.46, 0.11, 0.14.

CONCLUSIONS

A 4-quadrant random biopsy is recommended only for hrHPV positive females with HSIL cytology, and is acceptable if hrHPV positive with LSIL cytology or with abnormal VIA/VILI. Our findings add evidences for an objective and practical biopsy standard to guide colposcopy in cervical cancer screening programs in low- and middle-income countries.

MiRNA detection in cervical exfoliated cells for missed high-grade lesions in women with LSIL/CIN1 diagnosis after colposcopy-guided biopsy

Background

Low-grade squamous intraepithelial lesion/cervical intraepithelial neoplasia grade 1 (LSIL/CIN1) preceded by colposcopy guided biopsy is recommended conservative follow-up, although some of these lesions are actually high-grade lesions, which are missed on an initial colposcopy. Therefore, in this work, we evaluate the potential role of miRNA detection in cervical exfoliated cells in a clinic-based population for predicting missed high-grade lesions in women diagnosed with LSIL/CIN1 after colposcopy-guided biopsy.

Methods

A total number of 177 women with a diagnosis of LSIL/CIN1 obtained by colposcopy-guided biopsy were grouped into two categories according to the histology of the conization specimens: consistent LSIL/CIN1 group (surgical pathology consistent with colposcopic diagnosis) and missed high-grade lesion group (surgical pathology found high-grade lesion). The expression of eight miRNAs, such as miRNA195, miRNA424, miRNA375, miRNA218, miRNA34a, miRNA29a, miRNA16–2, and miRNA20a was detected by real time-quantitative polymerase chain reaction (RT-qPCR) in cervical exfoliated cells of the 177 patients. Pearson Chi-Square was used to compare the performance efficiency of patients’ characteristics. Nonparametric Man-Whitney U test was used to assess differences in miRNA expression. The receiver operating characteristic (ROC) curve was used to assess the performance of miRNA evaluation in detecting missed high-grade lesions.

Results

Among the 177 women with biopsy-confirmed CIN1, 15.3% (27/177) had CIN2+ in the conization specimen (missed high-grade lesion group) and 84.7% (150/177) had CIN1-(consistent LSIL/CIN1 group). The relative expression of miRNA-195 and miRNA-29a in the missed high-grade lesion group was significantly lower than that in the consistent LSIL/CIN1 group. The relative expression of miRNA16–2 and miRNA20a in the missed high-grade lesion group was significantly higher than that in the consistent LSIL/CIN1 group. No significant difference was observed between these two groups regarding the other four miRNAs. Of these significant miRNAs, miRNA29a detection achieved the highest Youden index (0.733), sensitivity (92.6%), positive predictive value (46.2%), negative predictive value (98.3%) and higher specificity (80.7%) when identifying missed high-grade lesions.

Conclusions

Detection of miRNA might provide a new triage for identifying a group at higher risk of missed high-grade lesions in women with colposcopy diagnosis of LSIL/CIN1.

Electronic supplementary material

The online version of this article (10.1186/s12885-019-5311-3) contains supplementary material, which is available to authorized users.