Development of algorithms for automated detection of cervical pre-cancers with a low-cost, point-of-care, Pocket colposcope

Asiedu, Mercy Nyamewaa; Simhal, Anish K.; Chaudhary, Usamah; Mueller, Jenna L.; Lam, Christopher T.; Schmitt, John W.; Venegas, Gino; Sapiro, Guillermo; Ramanujam, Nimmi

doi:10.1101/324541

Cited by 12 publications

(14 citation statements)

References 43 publications

(46 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, the number of patients and images included in this study was the largest from all colposcopic machine learning models created till date. Previous studies have only included 51–158 subjects with 170–485 colposcopic images 22 , 23 , 35 , 36 , whereas our study included 791 subjects with 791 colposcopic images. Second, the colposcopic images were obtained from three medical centers (the Kangnam Sacred Heart Hospital, Dongtan Sacred Heart Hospital, and Seoul St. Mary’s Hospital).…”

Section: Discussionmentioning

confidence: 99%

Classification of cervical neoplasms on colposcopic photography using deep learning

Cho

Choi

Lee

et al. 2020

Sci Rep

View full text Add to dashboard Cite

Colposcopy is widely used to detect cervical cancers, but experienced physicians who are needed for an accurate diagnosis are lacking in developing countries. Artificial intelligence (AI) has been recently used in computer-aided diagnosis showing remarkable promise. In this study, we developed and validated deep learning models to automatically classify cervical neoplasms on colposcopic photographs. Pre-trained convolutional neural networks were fine-tuned for two grading systems: the cervical intraepithelial neoplasia (CIN) system and the lower anogenital squamous terminology (LAST) system. The multi-class classification accuracies of the networks for the CIN system in the test dataset were 48.6 ± 1.3% by Inception-Resnet-v2 and 51.7 ± 5.2% by Resnet-152. The accuracies for the LAST system were 71.8 ± 1.8% and 74.7 ± 1.8%, respectively. The area under the curve (AUC) for discriminating high-risk lesions from low-risk lesions by Resnet-152 was 0.781 ± 0.020 for the CIN system and 0.708 ± 0.024 for the LAST system. The lesions requiring biopsy were also detected efficiently (AUC, 0.947 ± 0.030 by Resnet-152), and presented meaningfully on attention maps. These results may indicate the potential of the application of AI for automated reading of colposcopic photographs. Cervical cancer is the fourth most common cancer in women worldwide, and the second most common cancer among females in developing countries 1. Screening is the principal prevention method aimed at reducing mortality rates. Screening includes certain steps, including population-based Papanicolaou (Pap) testing, colposcopydirected biopsy of suspicious lesions, and the treatment of confirmed pre-cancer lesions 2,3. In women with low-grade intraepithelial lesions (LSIL) or high-grade intraepithelial lesions (HSIL), the risk of pre-cancer is medium to high, and immediate referral for colposcopy is necessary. However, referring all women with atypical squamous cells of undetermined significance (ASC-US) is considered inefficient, as the risk of such cases being pre-cancerous is lower 4. Screening programs have been successful in the developed countries, leading to an approximately 80% decrease in the cervical cancer incidence over the past 4 decades. In contrast, the increase in cervical cancer incidence reported in developing countries 5 has been attributed to the unsuccessful implementation of screening programs. This, has been attributed to logistics in health systems, infrastructural inadequacies, and the lack of expert physicians capable of introducing screening programs and follow-up 6 .

show abstract

Section: Discussionmentioning

confidence: 99%

Classification of cervical neoplasms on colposcopic photography using deep learning

Cho

Choi

Lee

et al. 2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Early in 2009, Acosta et al 21 used K-NN algorithm to automatically distinguish normal and abnormal cervical tissue in aceto-white pattern, and gained a sensitivity of 71% and the specificity of 59%. Years later, Asiedu et al 22 achieved the sensitivity, specificity, and accuracy of 81.3%, 78.6%, and 80.0% to distinguish CIN+ and benign tissues apart. Liming Hu et al 23 established a cohort and followed up for 7 years, using images shot by cervicography, to train and validate deep learning algorithm and gained higher accuracy compared to pap smear.…”

mentioning

confidence: 99%

The application of deep learning based diagnostic system to cervical squamous intraepithelial lesions recognition in colposcopy images

Yuan

Yao

Cheng

et al. 2020

Sci Rep

View full text Add to dashboard Cite

Background Deep learning has presented considerable potential and is gaining more importance in computer assisted diagnosis. As the gold standard for pathologically diagnosing cervical intraepithelial lesions and invasive cervical cancer, colposcopy-guided biopsy faces challenges in improving accuracy and efficiency worldwide, especially in developing countries. To ease the heavy burden of cervical cancer screening, it is urgent to establish a scientific, accurate and efficient method for assisting diagnosis and biopsy. Methods The data were collected to establish three deeplearning-based models. For every case, one saline image, one acetic image, one iodine image and the corresponding clinical information, including age, the results of human papillomavirus testing and cytology, type of transformation zone, and pathologic diagnosis, were collected. The dataset was proportionally divided into three subsets including the training set, the test set and the validation set, at a ratio of 8:1:1. The validation set was used to evaluate model performance. After model establishment, an independent dataset of high-definition images was collected to further evaluate the model performance. In addition, the comparison of diagnostic accuracy between colposcopists and models weas performed. Results The sensitivity, specificity and accuracy of the classification model to differentiate negative cases from positive cases were 85.38%, 82.62% and 84.10% respectively, with an AUC of 0.93. The recall and DICE of the segmentation model to segment suspicious lesions in acetic images were 84.73% and 61.64%, with an average accuracy of 95.59%. Furthermore, 84.67% of highgrade lesions were detected by the acetic detection model. Compared to colposcopists, the diagnostic system performed better in ordinary colposcopy images but slightly unsatisfactory in high-definition images. Implications The deep learning-based diagnostic system could help assist colposcopy diagnosis and biopsy for HSILs.

show abstract

“…Kim et al trained a support vector machine (SVM) with 2000 cervical images and achieved the sensitivity and specificity of 75% for CIN2+ and normal/CIN1 classification [19]. Asiedu et al trained and validated a SVM model with 134 patients and achieved 80% accuracy, 81.3% sensitivity, and 78.6% specificity for detecting CIN1+ against normal/benign tissues [20]. Miyagi et al trained and validated AI classifier by using 310 images, which showed a high accuracy of 82.3%, sensitivity of 80%, and specificity of 88.2% for the classification of CIN1 and CIN2+ [21].…”

Section: The Advancements In Computer Algorithms Applying To Cervicalmentioning

confidence: 99%

The challenges of colposcopy for cervical cancer screening in LMICs and solutions by artificial intelligence

Xue

Qiao

2020

BMC Med

110

View full text Add to dashboard Cite

Background: The World Health Organization (WHO) called for global action towards the elimination of cervical cancer. One of the main strategies is to screen 70% of women at the age between 35 and 45 years and 90% of women managed appropriately by 2030. So far, approximately 85% of cervical cancers occur in low-and middleincome countries (LMICs). The colposcopy-guided biopsy is crucial for detecting cervical intraepithelial neoplasia (CIN) and becomes the main bottleneck limiting screening performance. Unprecedented advances in artificial intelligence (AI) enable the synergy of deep learning and digital colposcopy, which offers opportunities for automatic image-based diagnosis. To this end, we discuss the main challenges of traditional colposcopy and the solutions applying AI-guided digital colposcopy as an auxiliary diagnostic tool in low-and middle-income countries (LMICs). Main body: Existing challenges for the application of colposcopy in LMICs include strong dependence on the subjective experience of operators, substantial inter-and intra-operator variabilities, shortage of experienced colposcopists, consummate colposcopy training courses, and uniform diagnostic standard and strict quality control that are hard to be followed by colposcopists with limited diagnostic ability, resulting in discrepant reporting and documentation of colposcopy impressions. Organized colposcopy training courses should be viewed as an effective way to enhance the diagnostic ability of colposcopists, but implementing these courses in practice may not always be feasible to improve the overall diagnostic performance in a short period of time. Fortunately, AI has the potential to address colposcopic bottleneck, which could assist colposcopists in colposcopy imaging judgment, detection of underlying CINs, and guidance of biopsy sites. The automated workflow of colposcopy examination could create a novel cervical cancer screening model, reduce potentially false negatives and false positives, and improve the accuracy of colposcopy diagnosis and cervical biopsy. Conclusion: We believe that a practical and accurate AI-guided digital colposcopy has the potential to strengthen the diagnostic ability in guiding cervical biopsy, thereby improves cervical cancer screening performance in LMICs and accelerates the process of global cervical cancer elimination eventually.

show abstract

Development of algorithms for automated detection of cervical pre-cancers with a low-cost, point-of-care, Pocket colposcope

Cited by 12 publications

References 43 publications

Classification of cervical neoplasms on colposcopic photography using deep learning

Classification of cervical neoplasms on colposcopic photography using deep learning

The application of deep learning based diagnostic system to cervical squamous intraepithelial lesions recognition in colposcopy images

The challenges of colposcopy for cervical cancer screening in LMICs and solutions by artificial intelligence

Contact Info

Product

Resources

About