Supervised deep learning embeddings for the prediction of cervical cancer diagnosis

Cervical cancer remains a significant cause of mortality all around the world, even if it can be prevented and cured by removing affected tissues in early stages. Providing universal and efficient access to cervical screening programs is a challenge that requires identifying vulnerable individuals in the population, among other steps. In this work, we present a computationally automated strategy for predicting the outcome of the patient biopsy, given risk patterns from individual medical records. We propose a machine learning technique that allows a joint and fully supervised optimization of dimensionality reduction and classification models. We also build a model able to highlight relevant properties in the low dimensional space, to ease the classification of patients. We instantiated the proposed approach with deep learning architectures, and achieved accurate prediction results (top area under the curve AUC = 0.6875) which outperform previously developed methods, such as denoising autoencoders. Additionally, we explored some clinical findings from the embedding spaces, and we validated them through the medical literature, making them reliable for physicians and biomedical researchers.

Download Full-text

P63 Improving cervical smear uptake through an Automated Voice Messages Reminder

BJS Open ◽

10.1093/bjsopen/zrab032.062 ◽

2021 ◽

Vol 5 (Supplement_1) ◽

Author(s):

Francesca Bladt ◽

Felyx Wong ◽

Francesca Bladt

Keyword(s):

Cervical Cancer ◽

Low Cost ◽

Cervical Screening ◽

Telephone Triage ◽

Cervical Smear ◽

Screening Programs ◽

Smear Test ◽

Voice Message ◽

The Uk ◽

Busy Schedule

Abstract National cervical screening programs have played a pivotal role in the prevention of cervical cancer. However, practices across the UK have reached an all-time low in cervical screening uptake. This study aimed to assess the efficacy of implementing an automated voice message reminder within the local general practice (GP) telephone triage system and explore the reasons which deter eligible patients away from cervical screening. A 20-second voice-message reminder in the telephone queue was played, addressing key risk factors along with a message from a child who lost his mother to cervical cancer. From the anonymised GP database, weekly new smear test bookings were monitored from 4 weeks prior until 2 weeks after the intervention was implemented. To qualitatively assess factors which deter patients away from screening, female patients were randomly sampled to fill in an anonymous questionnaire. The use of a low-cost 20 second voice message in the telephone queue across UK GP practices could be an effective method to increase cervical smear test coverage towards the national target of 80%. 35 questionnaire responses were received, main themes reported for not attending screening include embarrassment(37%), busy schedule(32%) and cultural differences(24%). In the week following the intervention, cervical smear tests increased more than 2-fold, from an average of 12 to 26 smears per week. This could be partly due to the convenient timing of voice recording, reminding them to book both appointments simultaneously and the child’s emotive message.

Download Full-text

Role of Protein Biomarkers in the Detection of High-Grade Disease in Cervical Cancer Screening Programs

Journal of Oncology ◽

10.1155/2012/289315 ◽

2012 ◽

Vol 2012 ◽

pp. 1-11 ◽

Cited By ~ 12

Author(s):

Charlotte A. Brown ◽

Johnannes Bogers ◽

Shaira Sahebali ◽

Christophe E. Depuydt ◽

Frans De Prins ◽

...

Keyword(s):

Cervical Cancer ◽

Cancer Screening ◽

Cervical Cancer Screening ◽

Cervical Screening ◽

False Negative ◽

Pap Test ◽

Developed Countries ◽

Protein Biomarkers ◽

Ki 67 ◽

Screening Programs

Since the Pap test was introduced in the 1940s, there has been an approximately 70% reduction in the incidence of squamous cell cervical cancers in many developed countries by the application of organized and opportunistic screening programs. The efficacy of the Pap test, however, is hampered by high interobserver variability and high false-negative and false-positive rates. The use of biomarkers has demonstrated the ability to overcome these issues, leading to improved positive predictive value of cervical screening results. In addition, the introduction of HPV primary screening programs will necessitate the use of a follow-up test with high specificity to triage the high number of HPV-positive tests. This paper will focus on protein biomarkers currently available for use in cervical cancer screening, which appear to improve the detection of women at greatest risk for developing cervical cancer, including Ki-67,p16INK4a, BD ProEx C, and Cytoactiv HPV L1.

Download Full-text

Multi-Task Deep Learning Model for Classification of Dental Implant Brand and Treatment Stage Using Dental Panoramic Radiograph Images

Biomolecules ◽

10.3390/biom11060815 ◽

2021 ◽

Vol 11 (6) ◽

pp. 815

Author(s):

Shintaro Sukegawa ◽

Kazumasa Yoshii ◽

Takeshi Hara ◽

Tamamo Matsuyama ◽

Katsusuke Yamashita ◽

...

Keyword(s):

Deep Learning ◽

Dental Implant ◽

Area Under The Curve ◽

Treatment Phase ◽

Panoramic Radiograph ◽

Radiographic Images ◽

Treatment Stage ◽

Prefectural Central Hospital ◽

Dental Panoramic Radiograph

It is necessary to accurately identify dental implant brands and the stage of treatment to ensure efficient care. Thus, the purpose of this study was to use multi-task deep learning to investigate a classifier that categorizes implant brands and treatment stages from dental panoramic radiographic images. For objective labeling, 9767 dental implant images of 12 implant brands and treatment stages were obtained from the digital panoramic radiographs of patients who underwent procedures at Kagawa Prefectural Central Hospital, Japan, between 2005 and 2020. Five deep convolutional neural network (CNN) models (ResNet18, 34, 50, 101 and 152) were evaluated. The accuracy, precision, recall, specificity, F1 score, and area under the curve score were calculated for each CNN. We also compared the multi-task and single-task accuracies of brand classification and implant treatment stage classification. Our analysis revealed that the larger the number of parameters and the deeper the network, the better the performance for both classifications. Multi-tasking significantly improved brand classification on all performance indicators, except recall, and significantly improved all metrics in treatment phase classification. Using CNNs conferred high validity in the classification of dental implant brands and treatment stages. Furthermore, multi-task learning facilitated analysis accuracy.

Download Full-text

The local Universe in the era of large surveys – I. Spectral classification of S0 galaxies

Monthly Notices of the Royal Astronomical Society ◽

10.1093/mnras/staa1408 ◽

2020 ◽

Vol 495 (4) ◽

pp. 4135-4157 ◽

Cited By ~ 1

Author(s):

J L Tous ◽

J M Solanes ◽

J D Perea

Keyword(s):

Dimensional Space ◽

Spectral Characteristics ◽

Principal Component ◽

Morphological Type ◽

Physical Parameters ◽

Star Forming ◽

Interesting Possibility ◽

Low Dimensional ◽

Free Machine

ABSTRACT This is the first paper in a series devoted to review the main properties of galaxies designated S0 in the Hubble classification system. Our aim is to gather abundant and, above all, robust information on the most relevant physical parameters of this poorly understood morphological type and their possible dependence on the environment, which could later be used to assess their possible formation channel(s). The adopted approach combines the characterization of the fundamental features of the optical spectra of $68\, 043$ S0 with heliocentric z ≲ 0.1 with the exploration of a comprehensive set of their global attributes. A principal component analysis is used to reduce the huge number of dimensions of the spectral data to a low-dimensional space facilitating a bias-free machine-learning-based classification of the galaxies. This procedure has revealed that objects bearing the S0 designation consist, despite their similar morphology, of two separate subpopulations with statistically inconsistent physical properties. Compared to the absorption-dominated S0, those with significant nebular emission are, on average, somewhat less massive, more luminous with less concentrated light profiles, have a younger, bluer, and metal-poorer stellar component, and avoid high-galaxy-density regions. Noteworthy is the fact that the majority of members of this latter class, which accounts for at least a quarter of the local S0 population, show star formation rates and spectral characteristics entirely similar to those seen in late spirals. Our findings suggest that star-forming S0 might be less rare than hitherto believed and raise the interesting possibility of identifying them with plausible progenitors of their quiescent counterparts.

Download Full-text

Self-Collection for Cervical Screening Programs: From Research to Reality

Cancers ◽

10.3390/cancers12041053 ◽

2020 ◽

Vol 12 (4) ◽

pp. 1053 ◽

Cited By ~ 2

Author(s):

David Hawkes ◽

Marco H. T. Keung ◽

Yanping Huang ◽

Tracey L. McDermott ◽

Joanne Romano ◽

...

Keyword(s):

Cervical Cancer ◽

Cervical Screening ◽

Scale Up ◽

Public Health Problem ◽

Routine Laboratory ◽

Major Barrier ◽

Screening Programs ◽

Speculum Examination ◽

Primary Screening ◽

High Negative Predictive Value

In 2018, there were an estimated 570,000 new cases of cervical cancer globally, with most of them occurring in women who either had no access to cervical screening, or had not participated in screening in regions where programs are available. Where programs are in place, a major barrier for women across many cultures has been the requirement to undergo a speculum examination. With the emergence of HPV-based primary screening, the option of self-collection (where the woman takes the sample from the vagina herself) may overcome this barrier, given that such samples when tested using a PCR-based HPV assay have similar sensitivity for the detection of cervical pre-cancers as practitioner-collected cervical specimens. Other advantages of HPV-based screening using self-collection, beyond the increase in acceptability to women, include scalability, efficiency, and high negative predictive value, allowing for long intervals between negative tests. Self-collection will be a key strategy for the successful scale up of cervical screening programs globally in response to the WHO call for all countries to work towards the elimination of cervical cancer as a public health problem. This review will examine self-collection for HPV-based cervical screening including the collection devices, assays and possible routine laboratory processes considering how they can be utilized in cervical screening programs.

Download Full-text

Learning representations of Web entities for entity resolution

International Journal of Web Information Systems ◽

10.1108/ijwis-07-2018-0059 ◽

2019 ◽

Vol 15 (3) ◽

pp. 346-358

Author(s):

Luciano Barbosa

Keyword(s):

Deep Learning ◽

Dimensional Space ◽

Entity Resolution ◽

Data Sets ◽

Content Type ◽

Learning Framework ◽

Document Frequency ◽

Deep Learning Network ◽

Low Dimensional ◽

Vector Representations

Purpose Matching instances of the same entity, a task known as entity resolution, is a key step in the process of data integration. This paper aims to propose a deep learning network that learns different representations of Web entities for entity resolution. Design/methodology/approach To match Web entities, the proposed network learns the following representations of entities: embeddings, which are vector representations of the words in the entities in a low-dimensional space; convolutional vectors from a convolutional layer, which capture short-distance patterns in word sequences in the entities; and bag-of-word vectors, created by a bow layer that learns weights for words in the vocabulary based on the task at hand. Given a pair of entities, the similarity between their learned representations is used as a feature to a binary classifier that identifies a possible match. In addition to those features, the classifier also uses a modification of inverse document frequency for pairs, which identifies discriminative words in pairs of entities. Findings The proposed approach was evaluated in two commercial and two academic entity resolution benchmarking data sets. The results have shown that the proposed strategy outperforms previous approaches in the commercial data sets, which are more challenging, and have similar results to its competitors in the academic data sets. Originality/value No previous work has used a single deep learning framework to learn different representations of Web entities for entity resolution.

Download Full-text

Identifying Disease Related Genes by Network Representation and Convolutional Neural Network

Frontiers in Cell and Developmental Biology ◽

10.3389/fcell.2021.629876 ◽

2021 ◽

Vol 9 ◽

Author(s):

Bolin Chen ◽

Yourui Han ◽

Xuequn Shang ◽

Shenggui Zhang

Keyword(s):

Neural Network ◽

Deep Learning ◽

Convolutional Neural Network ◽

Biological Networks ◽

Dimensional Space ◽

Scale Free ◽

Network Representation ◽

Disease Related Genes ◽

Representation Method ◽

Low Dimensional

The identification of disease related genes plays essential roles in bioinformatics. To achieve this, many powerful machine learning methods have been proposed from various computational aspects, such as biological network analysis, classification, regression, deep learning, etc. Among them, deep learning based methods have gained big success in identifying disease related genes in terms of higher accuracy and efficiency. However, these methods rarely handle the following two issues very well, which are (1) the multifunctions of many genes; and (2) the scale-free property of biological networks. To overcome these, we propose a novel network representation method to transfer individual vertices together with their surrounding topological structures into image-like datasets. It takes each node-induced sub-network as a represented candidate, and adds its environmental characteristics to generate a low-dimensional space as its representation. This image-like datasets can be applied directly in a Convolutional Neural Network-based method for identifying cancer-related genes. The numerical experiments show that the proposed method can achieve the AUC value at 0.9256 in a single network and at 0.9452 in multiple networks, which outperforms many existing methods.

Download Full-text

A Frequency-Based Approach for the Detection and Classification of Structural Changes Using t-SNE †

Sensors ◽

10.3390/s19235097 ◽

2019 ◽

Vol 19 (23) ◽

pp. 5097 ◽

Cited By ~ 7

Author(s):

David Agis ◽

Francesc Pozo

Keyword(s):

Structural Changes ◽

Dimensional Space ◽

Principal Component ◽

Majority Voting ◽

Data Set ◽

Centroid Distance ◽

Low Dimensional ◽

Reduced Data

This work presents a structural health monitoring (SHM) approach for the detection and classification of structural changes. The proposed strategy is based on t-distributed stochastic neighbor embedding (t-SNE), a nonlinear procedure that is able to represent the local structure of high-dimensional data in a low-dimensional space. The steps of the detection and classification procedure are: (i) the data collected are scaled using mean-centered group scaling (MCGS); (ii) then principal component analysis (PCA) is applied to reduce the dimensionality of the data set; (iii) t-SNE is applied to represent the scaled and reduced data as points in a plane defining as many clusters as different structural states; and (iv) the current structure to be diagnosed will be associated with a cluster or structural state based on three strategies: (a) the smallest point-centroid distance; (b) majority voting; and (c) the sum of the inverse distances. The combination of PCA and t-SNE improves the quality of the clusters related to the structural states. The method is evaluated using experimental data from an aluminum plate with four piezoelectric transducers (PZTs). Results are illustrated in frequency domain, and they manifest the high classification accuracy and the strong performance of this method.

Download Full-text

Opportunities and challenges in cervical cancer screening program in Kazakhstan: Results after imPACT review.

Journal of Clinical Oncology ◽

10.1200/jco.2020.38.15_suppl.e18020 ◽

2020 ◽

Vol 38 (15_suppl) ◽

pp. e18020-e18020

Author(s):

Dilyara Kaidarova ◽

Raikhan Bolatbekova ◽

Alma Zhylkaidarova ◽

Tolkyn Sadykova ◽

Yerlan Kukubassov ◽

...

Keyword(s):

Cervical Cancer ◽

Mortality Rate ◽

Screening Program ◽

Early Stage ◽

Gynecological Cancer ◽

Cervical Screening ◽

Pap Test ◽

Target Population ◽

Developed Countries ◽

Screening Programs

e18020 Background: Cervical cancer is the second most common cancer in women worldwide, where the majority of registered patients are in developing countries. Screening programs in developed countries have reduced morbidity and mortality from cervical cancer by more than 2 times. Cervical cancer (CC) is the most common gynecological cancer in Kazakhstan (KZ). Standardized incidence rate of CC was 18.2 per 100,000, while the mortality rate was 6.2 per 100,000, in 2019. The National Cervical Screening program in KZ uses cytology (Pap test) from 2008. Screening program funded by the State budget. In 2016, Experts of imPACT Mission analyzed the CC screening and made recommendations for improvement. Since 2018 target age of CC screening expanded to 30-70 years and shortened the interval to 4 years, strengthened the control of patients with pre-cancerous pathology. Until 2018, people came to a fixed age; today we start CC screening within the target age at any age at the time of the first visit. The purpose of this study is to analyze cytological screening results in KZ after imPACT recommendations. Methods: Coverage, the number of screened women, the level of pre-cancer detection and cervical cancer during screening have been obtained from specific reports (form № 025, № 08) for 2008-2019. Results: The total number of screened women was in 6.775.975. There is a decrease in the number of screened women by 32% from 2008 to 2017. Since improvement of CC screening we increased coverage from 49.9% in 2017 (abs. number 409.124) to 89% in 2019 (abs. number 954.322). According to the results of screening, 2603 cases of CC were registered in 12 years. Analysis of screening results showed a marked increase in the detection of CC with an increasing by 67%. The persentage of registered cases of ASH+HSIL increased from 0.136% to 0.673%. Conclusions: there has been an increase in the coverage by screening of the target population since the screening update. During the study period, there has been an improvement in the detection of precancerous pathology and cancer in the early stage. Despite the positive results of screening, sufficient coverage by screening, certain successes in detecting the initial stage of CC, mortality rate from CC remain high, which makes it necessary to improve the screening of CC in KZ through the introduction of HPV-screening.

Download Full-text

Promising strategies for cervical cancer screening in the post-human papillomavirus vaccination era

Sexual Health ◽

10.1071/sh10022 ◽

2010 ◽

Vol 7 (3) ◽

pp. 376 ◽

Cited By ~ 20

Author(s):

Joseph Tota ◽

Salaheddin M. Mahmud ◽

Alex Ferenczy ◽

François Coutlée ◽

Eduardo L. Franco

Keyword(s):

Cervical Cancer ◽

Human Papillomavirus ◽

Human Error ◽

Pap Smear ◽

Cervical Screening ◽

Screening Tests ◽

Hpv Testing ◽

Human Papillomavirus Vaccination ◽

Screening Programs

Human papillomavirus (HPV) vaccination is expected to reduce the burden of cervical cancer in most settings; however, it is also expected to interfere with the effectiveness of screening. In the future, maintaining Pap cytology as the primary cervical screening test may become too costly. As the prevalence of cervical dysplasias decreases, the positive predictive value of the Pap test will also decrease, and, as a result, more women will be referred for unnecessary diagnostic procedures and follow-up. HPV DNA testing has recently emerged as the most likely candidate to replace cytology for primary screening. It is less prone to human error and much more sensitive than the Pap smear in detecting high-grade cervical lesions. Incorporating this test would improve the overall quality of screening programs and allow spacing out screening tests, while maintaining safety and lowering costs. Although HPV testing is less specific than Pap cytology, this issue could be resolved by reserving the latter for the more labour-efficient task of triaging HPV-positive cases. Because most HPV-positive smears would contain relevant abnormalities, Pap cytology would be expected to perform with sufficient accuracy under these circumstances. HPV Pap triage would also provide a low-cost strategy to monitor long-term vaccine efficacy. Although demonstration projects could start implementing HPV testing as a population screening tool, more research is needed to determine the optimal age to initiate screening, the role of HPV typing and other markers of disease progression, and appropriate follow-up algorithms for HPV-positive and Pap-negative women.

Download Full-text