ABSTRACT
Visual assessment is currently used for primary screening or triage of screen-positive individuals in cervical cancer screening programs. Most guidelines recommend screening and triage up to at least age 65 years old. We examined cervical images from participants in three National Cancer Institute funded cervical cancer screening studies: ALTS (2864 participants recruited between 1996 to 1998) in the United States (US), NHS (7548 in 1993) in Costa Rica, and the Biopsy study (684 between 2009 to 2012) in the US. Specifically, we assessed the visibility of the squamocolumnar junction (SCJ), which is the susceptible zone for precancer/cancer by age, as reported by colposcopist reviewers either at examination or review of cervical images. The visibility of the SCJ declined substantially with age: by the late 40s the majority of people screened had at most partially visible SCJ. On longitudinal analysis, the change in SCJ visibility from visible to not visible was largest for participants from ages 40-44 in ALTS and 50-54 in NHS. Of note, in the Biopsy study, the live colposcopic exam resulted in significantly higher SCJ visibility as compared to review of static images (Weighted kappa 0.27 (95% Confidence Interval: 0.21, 0.33), Asymmetry chi-square P-value<0.001). Lack of SCJ visibility leads to increased difficulty in diagnosis and management of cervical precancers. Therefore, cervical cancer screening programs reliant on visual assessment might consider lowering the upper age limit for screening if there are not adequately trained personnel and equipment to evaluate and manage participants with inadequately visible SCJ.
Subject(s)
Uterine Cervical Dysplasia , Uterine Cervical Neoplasms , Female , Humans , Aged , Uterine Cervical Neoplasms/diagnosis , Uterine Cervical Neoplasms/prevention & control , Uterine Cervical Neoplasms/pathology , Early Detection of Cancer/methods , Uterine Cervical Dysplasia/pathology , BiopsyABSTRACT
Cervical cancer screening and treatment of screen positives is an important and effective strategy to reduce cervical cancer morbidity and mortality. In order to have an accurate cervical cancer screening and evaluation of positives, the entire Squamocolumnar Junction (SCJ) must be visible. Throughout the life course, the position of the SCJ changes and affects its visibility. SCJ visibility was analyzed among participants screened at the League Against Cancer Clinic in Lima, Peru. Of the 4247 participants screened, the SCJ was fully visible in 49.7% of participants, partially visible in 23.1%, and not visible in 27.2%. Visibility decreased with age, and by age 45 years old, the SCJ was not fully visible in over 50% of participants. Our results show that a high percentage of participants at ages still recommended for screening do not have totally visible SCJ, and we may need to reconsider the upper age limit for screening and find new strategies for evaluation of those with a positive screening test and non-visible SCJ.
Subject(s)
Uterine Cervical Neoplasms , Female , Humans , Middle Aged , Uterine Cervical Neoplasms/diagnosis , Early Detection of Cancer , Peru , Mass ScreeningABSTRACT
Accelerated cervical cancer control will require widespread human papillomavirus (HPV) vaccination and screening. For screening, sensitive HPV testing with an option of self-collection is increasingly desirable. HPV typing predicts risk of precancer/cancer, which could be useful in management, but most current typing assays are expensive and/or complicated. An existing 15-type isothermal amplification assay (AmpFire, Atila Biosystems, USA) was redesigned as a 13-type assay (ScreenFire) for public health use. The redesigned assay groups HPV types into four channels with differential cervical cancer risk: (a) HPV16, (b) HPV18/45, (c) HPV31/33/35/52/58 and (d) HPV39/51/56/59/68. Since the assay will be most useful in resource-limited settings, we chose a stratified random sample of 453 provider-collected samples from a population-based screening study in rural Nigeria that had been initially tested with MY09-MY11-based PCR with oligonucleotide hybridization genotyping. Frozen residual specimens were masked and retested at Atila Biosystems. Agreement on positivity between ScreenFire and prior PCR testing was very high for each of the channels. When we simulated intended use, that is, a hierarchical result in order of clinical importance of the type groups (HPV16 > 18/45 > 31/33/35/52/58 > 39/51/56/59/68), the weighted kappa for ScreenFire vs PCR was 0.90 (95% CI: 0.86-0.93). The ScreenFire assay is mobile, relatively simple, rapid (results within 20-60 minutes) and agrees well with reference testing particularly for the HPV types of greatest carcinogenic risk. If confirmed, ScreenFire or similar isothermal amplification assays could be useful as part of risk-based screening and management.
Subject(s)
Papillomavirus Infections , Uterine Cervical Neoplasms , Cervix Uteri , DNA, Viral/analysis , DNA, Viral/genetics , Early Detection of Cancer/methods , Female , Genotype , Human papillomavirus 16/genetics , Humans , Papillomaviridae/geneticsABSTRACT
There is limited access to effective cervical cancer screening programs in many resource-limited settings, resulting in continued high cervical cancer burden. Human papillomavirus (HPV) testing is increasingly recognized to be the preferable primary screening approach if affordable due to superior long-term reassurance when negative and adaptability to self-sampling. Visual inspection with acetic acid (VIA) is an inexpensive but subjective and inaccurate method widely used in resource-limited settings, either for primary screening or for triage of HPV-positive individuals. A deep learning (DL)-based automated visual evaluation (AVE) of cervical images has been developed to help improve the accuracy and reproducibility of VIA as assistive technology. However, like any new clinical technology, rigorous evaluation and proof of clinical effectiveness are required before AVE is implemented widely. In the current article, we outline essential clinical and technical considerations involved in building a validated DL-based AVE tool for broad use as a clinical test.
Subject(s)
Deep Learning , Early Detection of Cancer/methods , Uterine Cervical Neoplasms/diagnosis , Algorithms , Female , Humans , Papillomaviridae/isolation & purification , Reproducibility of Results , Uterine Cervical Neoplasms/virologyABSTRACT
OBJECTIVE: The US screening and management guidelines for cervical cancer are based on the absolute risk of precancer estimated from large clinical cohorts and trials. Given the widespread transition toward screening with human papillomavirus (HPV) testing, it is important to assess which additional factors to include in clinical risk assessment to optimize management of HPV-infected women. MATERIALS AND METHODS: We analyzed data from HPV-infected women, ages 30-65 years, in the National Cancer Institute-Kaiser Permanente Northern California Persistence and Progression study. We estimated the influence of HPV risk group, cytology result, and selected cofactors on immediate risk of cervical intraepithelial neoplasia grade 3 or higher (CIN 3+) among 16,094 HPV-positive women. Cofactors considered included, age, race/ethnicity, income, smoking, and hormonal contraceptive use. RESULTS: Human papillomavirus risk group and cytology test result were strongly correlated with CIN 3+ risk. After considering cytology and HPV risk group, other cofactors (age, race/ethnicity, income, smoking, and hormonal contraceptive use) had minimal impact on CIN 3+ risk and did not change recommended management based on accepted risk thresholds. We had insufficient data to assess the impact of long-duration heavy smoking, parity, history of sexually transmitted infection, or immunosuppression. CONCLUSIONS: In our study at the Kaiser Permanente Northern California, the risk of CIN 3+ was determined mainly by HPV risk group and cytology results, with other cofactors having limited impact in adjusted analyses. This supports the use of HPV and cytology results in risk-based management guidelines.
Subject(s)
Alphapapillomavirus , Papillomavirus Infections , Uterine Cervical Dysplasia , Uterine Cervical Neoplasms , Adult , Aged , Female , Humans , Mass Screening/methods , Middle Aged , Papillomaviridae , Papillomavirus Infections/diagnosis , Uterine Cervical Neoplasms/diagnosis , Uterine Cervical Neoplasms/epidemiology , Uterine Cervical Neoplasms/prevention & control , Vaginal Smears , Uterine Cervical Dysplasia/diagnosis , Uterine Cervical Dysplasia/epidemiologyABSTRACT
Health decision models are the only available tools designed to consider the lifetime natural history of human papillomavirus (HPV) infection and pathogenesis of cervical cancer, and the estimated long-term impact of preventive interventions. Yet health decision modeling results are often considered a lesser form of scientific evidence due to the inherent needs to rely on imperfect data and make numerous assumptions and extrapolations regarding complex processes. We propose a new health decision modeling framework that de-emphasizes cytologic-colposcopic-histologic diagnoses due to their subjectivity and lack of reproducibility, relying instead on HPV type and duration of infection as the major determinants of subsequent transition probabilities. We posit that the new model health states (normal, carcinogenic HPV infection, precancer, cancer) and corollary transitions are universal, but that the probabilities of transitioning between states may vary by population. Evidence for this variability in host response to HPV infections can be inferred from HPV prevalence patterns in different regions across the lifespan, and might be linked to different average population levels of immunologic control of HPV infections. By prioritizing direct estimation of model transition probabilities from longitudinal data (and limiting reliance on model-fitting techniques that may propagate error when applied to multiple transitions), we aim to reduce the number of assumptions for greater transparency and reliability. We propose this new microsimulation model for critique and discussion, hoping to contribute to models that maximally inform efficient strategies towards global cervical cancer elimination.
Subject(s)
Papillomavirus Infections , Papillomavirus Vaccines , Uterine Cervical Dysplasia , Uterine Cervical Neoplasms , Female , Humans , Papillomaviridae , Papillomavirus Infections/epidemiology , Papillomavirus Infections/prevention & control , Prevalence , Reproducibility of Results , Uterine Cervical Neoplasms/epidemiology , Uterine Cervical Neoplasms/prevention & controlABSTRACT
Novel screening and diagnostic tests based on artificial intelligence (AI) image recognition algorithms are proliferating. Some initial reports claim outstanding accuracy followed by disappointing lack of confirmation, including our own early work on cervical screening. This is a presentation of lessons learned, organized as a conceptual step-by-step approach to bridge the gap between the creation of an AI algorithm and clinical efficacy. The first fundamental principle is specifying rigorously what the algorithm is designed to identify and what the test is intended to measure (eg, screening, diagnostic, or prognostic). Second, designing the AI algorithm to minimize the most clinically important errors. For example, many equivocal cervical images cannot yet be labeled because the borderline between cases and controls is blurred. To avoid a misclassified case-control dichotomy, we have isolated the equivocal cases and formally included an intermediate, indeterminate class (severity order of classes: case>indeterminate>control). The third principle is evaluating AI algorithms like any other test, using clinical epidemiologic criteria. Repeatability of the algorithm at the borderline, for indeterminate images, has proven extremely informative. Distinguishing between internal and external validation is also essential. Linking the AI algorithm results to clinical risk estimation is the fourth principle. Absolute risk (not relative) is the critical metric for translating a test result into clinical use. Finally, generating risk-based guidelines for clinical use that match local resources and priorities is the last principle in our approach. We are particularly interested in applications to lower-resource settings to address health disparities. We note that similar principles apply to other domains of AI-based image analysis for medical diagnostic testing.
Subject(s)
Artificial Intelligence , Uterine Cervical Neoplasms , Female , Humans , Early Detection of Cancer , Uterine Cervical Neoplasms/diagnosis , Algorithms , Image Processing, Computer-AssistedABSTRACT
Background: The HPV-automated visual evaluation (PAVE) Study is an extensive, multinational initiative designed to advance cervical cancer prevention in resource-constrained regions. Cervical cancer disproportionally affects regions with limited access to preventive measures. PAVE aims to assess a novel screening-triage-treatment strategy integrating self-sampled HPV testing, deep-learning-based automated visual evaluation (AVE), and targeted therapies. Methods: Phase 1 efficacy involves screening up to 100,000 women aged 25-49 across nine countries, using self-collected vaginal samples for hierarchical HPV evaluation: HPV16, else HPV18/45, else HPV31/33/35/52/58, else HPV39/51/56/59/68 else negative. HPV-positive individuals undergo further evaluation, including pelvic exams, cervical imaging, and biopsies. AVE algorithms analyze images, assigning risk scores for precancer, validated against histologic high-grade precancer. Phase 1, however, does not integrate AVE results into patient management, contrasting them with local standard care.Phase 2 effectiveness focuses on deploying AVE software and HPV genotype data in real-time clinical decision-making, evaluating feasibility, acceptability, cost-effectiveness, and health communication of the PAVE strategy in practice. Results: Currently, sites have commenced fieldwork, and conclusive results are pending. Conclusions: The study aspires to validate a screen-triage-treat protocol utilizing innovative biomarkers to deliver an accurate, feasible, and cost-effective strategy for cervical cancer prevention in resource-limited areas. Should the study validate PAVE, its broader implementation could be recommended, potentially expanding cervical cancer prevention worldwide. Funding: The consortial sites are responsible for their own study costs. Research equipment and supplies, and the NCI-affiliated staff are funded by the National Cancer Institute Intramural Research Program including supplemental funding from the Cancer Cures Moonshot Initiative. No commercial support was obtained. Brian Befano was supported by NCI/ NIH under Grant T32CA09168.
Subject(s)
Papillomavirus Infections , Uterine Cervical Neoplasms , Humans , Female , Uterine Cervical Neoplasms/diagnosis , Uterine Cervical Neoplasms/prevention & control , Early Detection of Cancer , Papillomavirus Infections/diagnosis , Vagina , AlgorithmsABSTRACT
Deepening understanding of cervical cancer pathogenesis has yielded one-dose prophylactic human papillomavirus (HPV) vaccines and accurate HPV-based cervical screening tests. Knowing the heterogeneous carcinogenic potential of the individual high-risk HPV types permits prioritization of vaccination and screening strategies. However, "correct" (i.e., safe and effective) treatment of women found to have precancer is still undefined, forcing reliance on one or more rounds of untargeted destructive/excisional treatment. Both over-treatment and under-treatment are common results. Until safe and effective anti-HPV therapies are invented, defining optimal destructive/excisional treatment of precancer remains a fundamental and under-researched challenge, especially in resource-constrained settings. See related article by King et al., p. 681.
Subject(s)
Papillomavirus Infections , Papillomavirus Vaccines , Uterine Cervical Dysplasia , Uterine Cervical Neoplasms , Female , Humans , Uterine Cervical Neoplasms/diagnosis , Uterine Cervical Neoplasms/prevention & control , Uterine Cervical Neoplasms/pathology , Early Detection of Cancer/methods , Uterine Cervical Dysplasia/diagnosis , Cervix Uteri/surgery , Cervix Uteri/pathology , Papillomavirus Vaccines/therapeutic use , Mass Screening , PapillomaviridaeABSTRACT
Cervical cancer is a leading cause of cancer mortality, with approximately 90% of the 250,000 deaths per year occurring in low- and middle-income countries (LMIC). Secondary prevention with cervical screening involves detecting and treating precursor lesions; however, scaling screening efforts in LMIC has been hampered by infrastructure and cost constraints. Recent work has supported the development of an artificial intelligence (AI) pipeline on digital images of the cervix to achieve an accurate and reliable diagnosis of treatable precancerous lesions. In particular, WHO guidelines emphasize visual triage of women testing positive for human papillomavirus (HPV) as the primary screen, and AI could assist in this triage task. Published AI reports have exhibited overfitting, lack of portability, and unrealistic, near-perfect performance estimates. To surmount recognized issues, we implemented a comprehensive deep-learning model selection and optimization study on a large, collated, multi-institutional dataset of 9,462 women (17,013 images). We evaluated relative portability, repeatability, and classification performance. The top performing model, when combined with HPV type, achieved an area under the Receiver Operating Characteristics (ROC) curve (AUC) of 0.89 within our study population of interest, and a limited total extreme misclassification rate of 3.4%, on held-aside test sets. Our work is among the first efforts at designing a robust, repeatable, accurate and clinically translatable deep-learning model for cervical screening.
ABSTRACT
Objective: To describe the HPV-Automated Visual Evaluation (PAVE) Study, an international, multi-centric study designed to evaluate a novel cervical screen-triage-treat strategy for resource-limited settings as part of a global strategy to reduce cervical cancer burden. The PAVE strategy involves: 1) screening with self-sampled HPV testing; 2) triage of HPV-positive participants with a combination of extended genotyping and visual evaluation of the cervix assisted by deep-learning-based automated visual evaluation (AVE); and 3) treatment with thermal ablation or excision (Large Loop Excision of the Transformation Zone). The PAVE study has two phases: efficacy (2023-2024) and effectiveness (planned to begin in 2024-2025). The efficacy phase aims to refine and validate the screen-triage portion of the protocol. The effectiveness phase will examine acceptability and feasibility of the PAVE strategy into clinical practice, cost-effectiveness, and health communication within the PAVE sites. Study design: Phase 1 Efficacy: Around 100,000 nonpregnant women, aged 25-49 years, without prior hysterectomy, and irrespective of HIV status, are being screened at nine study sites in resource-limited settings. Eligible and consenting participants perform self-collection of vaginal specimens for HPV testing using a FLOQSwab (Copan). Swabs are transported dry and undergo testing for HPV using a newly-redesigned isothermal DNA amplification HPV test (ScreenFire HPV RS), which has been designed to provide HPV genotyping by hierarchical risk groups: HPV16, else HPV18/45, else HPV31/33/35/52/58, else HPV39/51/56/59/68. HPV-negative individuals are considered negative for precancer/cancer and do not undergo further testing. HPV-positive individuals undergo pelvic examination with collection of cervical images and targeted biopsies of all acetowhite areas or endocervical sampling in the absence of visible lesions. Accuracy of histology diagnosis is evaluated across all sites. Cervical images are used to refine a deep learning AVE algorithm that classifies images as normal, indeterminate, or precancer+. AVE classifications are validated against the histologic endpoint of high-grade precancer determined by biopsy. The combination of HPV genotype and AVE classification is used to generate a risk score that corresponds to the risk of precancer (lower, medium, high, highest). During the efficacy phase, clinicians and patients within the PAVE sites will receive HPV testing results but not AVE results or risk scores. Treatment during the efficacy phase will be performed per local standard of care: positive Visual Inspection with Acetic Acid impression, high-grade colposcopic impression or CIN2+ on colposcopic biopsy, HPV positivity, or HPV 16,18/45 positivity. Follow up of triage negative patients and post treatment will follow standard of care protocols. The sensitivity of the PAVE strategy for detection of precancer will be compared to current SOC at a given level of specificity.Phase 2 Effectiveness: The AVE software will be downloaded to the new dedicated image analysis and thermal ablation devices (Liger Iris) into which the HPV genotype information can be entered to provide risk HPV-AVE risk scores for precancer to clinicians in real time. The effectiveness phase will examine clinician use of the PAVE strategy in practice, including feasibility and acceptability for clinicians and patients, cost-effectiveness, and health communication within the PAVE sites. Conclusion: The goal of the PAVE study is to validate a screen-triage-treat protocol using novel biomarkers to provide an accurate, feasible, cost-effective strategy for cervical cancer prevention in resource-limited settings. If validated, implementation of PAVE at larger scale can be encouraged. Funding: The consortial sites are responsible for their own study costs. Research equipment and supplies, and the NCI-affiliated staff are funded by the National Cancer Institute Intramural Research Program including supplemental funding from the Cancer Cures Moonshot Initiative. No commercial support was obtained. Brian Befano was supported by NCI/NIH under Grant T32CA09168. Date of protocol latest review: September 24 th 2023.
ABSTRACT
Cervical cancer is a leading cause of cancer mortality, with approximately 90% of the 250,000 deaths per year occurring in low- and middle-income countries (LMIC). Secondary prevention with cervical screening involves detecting and treating precursor lesions; however, scaling screening efforts in LMIC has been hampered by infrastructure and cost constraints. Recent work has supported the development of an artificial intelligence (AI) pipeline on digital images of the cervix to achieve an accurate and reliable diagnosis of treatable precancerous lesions. In particular, WHO guidelines emphasize visual triage of women testing positive for human papillomavirus (HPV) as the primary screen, and AI could assist in this triage task. In this work, we implemented a comprehensive deep-learning model selection and optimization study on a large, collated, multi-geography, multi-institution, and multi-device dataset of 9462 women (17,013 images). We evaluated relative portability, repeatability, and classification performance. The top performing model, when combined with HPV type, achieved an area under the Receiver Operating Characteristics (ROC) curve (AUC) of 0.89 within our study population of interest, and a limited total extreme misclassification rate of 3.4%, on held-aside test sets. Our model also produced reliable and consistent predictions, achieving a strong quadratic weighted kappa (QWK) of 0.86 and a minimal %2-class disagreement (% 2-Cl. D.) of 0.69%, between image pairs across women. Our work is among the first efforts at designing a robust, repeatable, accurate and clinically translatable deep-learning model for cervical screening.
Subject(s)
Papillomavirus Infections , Uterine Cervical Neoplasms , Humans , Female , Cervix Uteri/pathology , Papillomavirus Infections/epidemiology , Artificial Intelligence , Early Detection of Cancer/methods , Mass Screening/methods , Neural Networks, ComputerABSTRACT
Background: High-risk human papillomavirus (hrHPV) may cause more than 99% of cervical cancers worldwide. Little is known about performance differences in tests for hrHPV. Objective: This study analysed agreement for detection of hrHPV between the established, clinically validated Xpert HPV assay and the novel isothermal amplification-based AmpFire HPV genotyping assay. Methods: This study was nested in a larger project on cervical cancer screening among approximately 5000 women living with HIV in Kigali, Rwanda. This sub-study included 298 participants who underwent initial screening for cervical cancer using the Xpert HPV assay and visual inspection with acetic acid in 2017 and tested positive by either or both. Participants were rescreened using colposcopy, and cervical samples were collected between June 2018 and June 2019. Samples were then tested for HPV using the Xpert HPV assay and AmpFire HPV genotyping assay. Agreement between results from both tests was analysed using an exact version of McNemar test and chi-square test. Results: Overall agreement and kappa value for detection of hrHPV by Xpert and AmpFire were 89% and 0.77 (95% confidence interval: 0.70-0.85). AmpFire was marginally more likely to diagnose hrHPV-positive than Xpert (p = 0.05), due primarily to the extra positivity for HPV16 (p < 0.001). Conclusion: Overall, there was good to excellent agreement between the Xpert and AmpFire when testing hrHPV types among women living with HIV. AmpFire was more likely to test extra cases of HPV16, the most carcinogenic HPV type, but the clinical meaning of detecting additional HPV16 infections remains unknown.
ABSTRACT
Image quality control is a critical element in the process of data collection and cleaning. Both manual and automated analyses alike are adversely impacted by bad quality data. There are several factors that can degrade image quality and, correspondingly, there are many approaches to mitigate their negative impact. In this paper, we address image quality control toward our goal of improving the performance of automated visual evaluation (AVE) for cervical precancer screening. Specifically, we report efforts made toward classifying images into four quality categories ("unusable", "unsatisfactory", "limited", and "evaluable") and improving the quality classification performance by automatically identifying mislabeled and overly ambiguous images. The proposed new deep learning ensemble framework is an integration of several networks that consists of three main components: cervix detection, mislabel identification, and quality classification. We evaluated our method using a large dataset that comprises 87,420 images obtained from 14,183 patients through several cervical cancer studies conducted by different providers using different imaging devices in different geographic regions worldwide. The proposed ensemble approach achieved higher performance than the baseline approaches.
ABSTRACT
Our retrospective cohort study assesses the survival probability and identifies the demographic and clinical predictors of mortality in HIV patients taking antiretroviral therapy using an antiretroviral therapy centre data in Western India. Secondary data on 7532 registered HIV-infected individuals between September 2006 and January 2013 were analysed. The probability of survival at 75 months was 84.9%. Significant indicators of poor chances of survival were greater age, lower occupation class, lower CD4 count, poor functional status; higher stage of disease, lower weight, the presence and type of opportunistic infections, co-trimoxazole therapy and poor adherence to antiretroviral therapy. We thus find that, in addition to pre-ART, antiretroviral therapy clinical status and treatment adherence, socioeconomic status plays an important influence on ultimate survival of HIV patients on antiretroviral therapy.
Subject(s)
Anti-HIV Agents/therapeutic use , Antiretroviral Therapy, Highly Active , HIV Infections/drug therapy , Adult , CD4 Lymphocyte Count , Demography , HIV Infections/mortality , Humans , Middle Aged , Retrospective Studies , Socioeconomic Factors , Survival Analysis , Survival Rate , Treatment OutcomeABSTRACT
Cervical cancer is a disease of significant concern affecting women's health worldwide. Early detection of and treatment at the precancerous stage can help reduce mortality. High-grade cervical abnormalities and precancer are confirmed using microscopic analysis of cervical histopathology. However, manual analysis of cervical biopsy slides is time-consuming, needs expert pathologists, and suffers from reader variability errors. Prior work in the literature has suggested using automated image analysis algorithms for analyzing cervical histopathology images captured with the whole slide digital scanners (e.g., Aperio, Hamamatsu, etc.). However, whole-slide digital tissue scanners with good optical magnification and acceptable imaging quality are cost-prohibitive and difficult to acquire in low and middle-resource regions. Hence, the development of low-cost imaging systems and automated image analysis algorithms are of critical importance. Motivated by this, we conduct an experimental study to assess the feasibility of developing a low-cost diagnostic system with the H&E stained cervical tissue image analysis algorithm. In our imaging system, the image acquisition is performed by a smartphone affixing it on the top of a commonly available light microscope which magnifies the cervical tissues. The images are not captured in a constant optical magnification, and, unlike whole-slide scanners, our imaging system is unable to record the magnification. The images are mega-pixel images and are labeled based on the presence of abnormal cells. In our dataset, there are total 1331 (train: 846, validation: 116 test: 369) images. We formulate the classification task as a deep multiple instance learning problem and quantitatively evaluate the classification performance of four different types of multiple instance learning algorithms trained with five different architectures designed with varying instance sizes. Finally, we designed a sparse attention-based multiple instance learning framework that can produce a maximum of 84.55% classification accuracy on the test set.
Subject(s)
Image Processing, Computer-Assisted , Uterine Cervical Neoplasms , Algorithms , Female , Humans , Microscopy , Uterine Cervical Neoplasms/diagnostic imagingABSTRACT
Visual inspection of the cervix with acetic acid (VIA), though error prone, has long been used for screening women and to guide management for cervical cancer. The automated visual evaluation (AVE) technique, in which deep learning is used to predict precancer based on a digital image of the acetowhitened cervix, has demonstrated its promise as a low-cost method to improve on human performance. However, there are several challenges in moving AVE beyond proof-of-concept and deploying it as a practical adjunct tool in visual screening. One of them is making AVE robust across images captured using different devices. We propose a new deep learning based clustering approach to investigate whether the images taken by three different devices (a common smartphone, a custom smartphone-based handheld device for cervical imaging, and a clinical colposcope equipped with SLR digital camera-based imaging capability) can be well distinguished from each other with respect to the visual appearance/content within their cervix regions. We argue that disparity in visual appearance of a cervix across devices could be a significant confounding factor in training and generalizing AVE performance. Our method consists of four components: cervix region detection, feature extraction, feature encoding, and clustering. Multiple experiments are conducted to demonstrate the effectiveness of each component and compare alternative methods in each component. Our proposed method achieves high clustering accuracy (97%) and significantly outperforms several representative deep clustering methods on our dataset. The high clustering performance indicates the images taken from these three devices are different with respect to visual appearance. Our results and analysis establish a need for developing a method that minimizes such variance among the images acquired from different devices. It also recognizes the need for large number of training images from different sources for robust device-independent AVE performance worldwide.
ABSTRACT
Uterine cervical cancer is a leading cause of women's mortality worldwide. Cervical tissue ablation is an effective surgical excision of high grade lesions that are determined to be precancerous. Our prior work on the Automated Visual Examination (AVE) method demonstrated a highly effective technique to analyze digital images of the cervix for identifying precancer. Next step would be to determine if she is treatable using ablation. However, not all women are eligible for the therapy due to cervical characteristics. We present a machine learning algorithm that uses a deep learning object detection architecture to determine if a cervix is eligible for ablative treatment based on visual characteristics presented in the image. The algorithm builds on the well-known RetinaNet architecture to derive a simpler and novel architecture in which the last convolutional layer is constructed by upsampling and concatenating specific RetinaNet pretrained layers, followed by an output module consisting of a Global Average Pooling (GAP) layer and a fully connected layer. To explain the recommendation of the deep learning algorithm and determine if it is consistent with lesion presentation on the cervical anatomy, we visualize classification results using two techniques: our (i) Class-selective Relevance Map (CRM), which has been reported earlier, and (ii) Class Activation Map (CAM). The class prediction heatmaps are evaluated by a gynecologic oncologist with more than 20 years of experience. Based on our observation and the expert's opinion, the customized architecture not only outperforms the baseline RetinaNet network in treatability classification, but also provides insights about the features and regions considered significant by the network toward explaining reasons for treatment recommendation. Furthermore, by investigating the heatmaps on Gaussian-blurred images that serve as surrogates for out-of-focus cervical pictures we demonstrate the effect of image quality degradation on cervical treatability classification and underscoring the need for using images with good visual quality.
ABSTRACT
BACKGROUND: Isothermal amplification-based tests were developed as rapid, low-cost, and simple alternatives to real-time reverse transcriptase-polymerase chain reaction (RT-PCR) tests for SARS-COV-2 detection. METHODS: Clinical performance of two isothermal amplification-based tests (Atila Biosystems iAMP COVID-19 detection test and OptiGene COVID-19 Direct Plus RT-LAMP test) was compared to clinical RT-PCR assays using different sampling strategies. A total of 1378 participants were tested across four study sites. RESULTS: Compared to standard of care RT-PCR testing, the overall sensitivity and specificity of the Atila iAMP test for detection of SARS-CoV-2 were 76.2% and 94.9%, respectively, and increased to 88.8% and 89.5%, respectively, after exclusion of an outlier study site. Sensitivity varied based on the anatomic collected site. Sensitivity for nasopharyngeal was 65.4% (range across study sites:52.8%-79.8%), mid-turbinate 88.2%, saliva 55.1% (range across study sites:42.9%-77.8%) and anterior nares 66.7% (range across study sites:63.6%-76.5%). The specificity for these anatomic collection sites ranged from 96.7% to 100%. Sensitivity improved in symptomatic patients (overall 82.7%) and those with a higher viral load (overall 92.4% for ct≤25). Sensitivity and specificity of the OptiGene Direct Plus RT-LAMP test, conducted at a single study-site, were 25.5% and 100%, respectively. CONCLUSIONS: The Atila iAMP COVID test with mid-turbinate sampling is a rapid, low-cost assay for detecting SARS-COV-2, especially in symptomatic patients and those with a high viral load, and could be used to reduce the risk of SARS-COV-2 transmission in clinical settings. Variation of performance between study sites highlights the need for site-specific clinical validation of these assays before clinical adoption.