Enhancing advanced cervical cell categorization with cluster-based intelligent systems by a novel integrated CNN approach with skip mechanisms and GAN-based augmentation DOI Creative Commons

Gunjan Shandilya,

Sheifali Gupta, Ahmad Almogren

и другие.

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Ноя. 23, 2024

Cervical cancer is one of the biggest challenges in global health, thus it forms a critical need for early detection technologies that could improve patient prognosis and inform treatment decisions. This development form an mechanism increases chances successful survival, as diagnosis promptly offers interventions can dramatically reduce rate deaths attributed to this disease. Here, customized Convolutional Neural Network (CNN) model proposed cervical cancerous cell detection. It includes three convolutional layers with increasing filter sizes max-pooling layers, followed by dropout dense improved feature extraction robust learning. By using ResNet models inspiration, further innovates incorporating skip connections into CNN design. enabling direct transmission from earlier later links enhance gradient flow help preserve important spatial information. boosting propagation, integration model's ability recognize minute patterns images, hence classification accuracy. In our methodology, SIPaKMeD dataset has been employed which contains 4049 images are arranged five different categories. To address class imbalance, Generative Adversarial Networks (GANs) have applied data augmentation; is, synthetic created, diversity robustness same. The present astonishingly accurate classifying types: koilocytes, superficial-intermediate, parabasal, dyskeratotic, metaplastic, significantly enhancing cancer. gives excellent performance because validation accuracy 99.11% training 99.82%. reliable cells ensures advancement computer-assisted system.

Язык: Английский

A low-cost platform for automated cervical cytology: addressing health and socioeconomic challenges in low-resource settings DOI Creative Commons

José Ocampo-López-Escalera,

Héctor Ochoa‐Díaz‐López, Xariss M. Sánchez‐Chino

и другие.

Frontiers in Medical Technology, Год журнала: 2025, Номер 7

Опубликована: Март 31, 2025

Cervical cancer remains a significant health challenge around the globe, with particularly high prevalence in low- and middle-income countries. This disease is preventable curable if detected early stages, making regular screening critically important. cytology, most widely used method, has proven highly effective reducing cervical incidence mortality income However, its effectiveness low-resource settings been limited, among other factors, by insufficient diagnostic infrastructure shortage of trained healthcare personnel. paper introduces development low-cost microscopy platform designed to address these limitations enabling automatic reading cytology slides. The system features robotized microscope capable slide scanning, autofocus, digital image capture, while supporting integration artificial intelligence (AI) algorithms. All at production cost below 500 USD. A dataset nearly 2,000 images, captured custom-built covering seven distinct cellular types relevant cytologic analysis, was created. then fine-tune test several pre-trained models for classifying between images containing normal abnormal cell subtypes. Most tested showed good performance properly cells, sensitivities above 90%. Among models, MobileNet demonstrated highest accuracy detecting types, achieving 98.26% 97.95%, specificities 88.91% 88.72%, F-scores 96.42% 96.23% on validation sets, respectively. results indicate that might be suitable model real-world deployment platform, offering precision efficiency images. presents first step towards promising solution improving settings.

Язык: Английский

Процитировано

0
Enhancing advanced cervical cell categorization with cluster-based intelligent systems by a novel integrated CNN approach with skip mechanisms and GAN-based augmentation DOI Creative Commons

Gunjan Shandilya,

Sheifali Gupta, Ahmad Almogren

и другие.

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Ноя. 23, 2024

Cervical cancer is one of the biggest challenges in global health, thus it forms a critical need for early detection technologies that could improve patient prognosis and inform treatment decisions. This development form an mechanism increases chances successful survival, as diagnosis promptly offers interventions can dramatically reduce rate deaths attributed to this disease. Here, customized Convolutional Neural Network (CNN) model proposed cervical cancerous cell detection. It includes three convolutional layers with increasing filter sizes max-pooling layers, followed by dropout dense improved feature extraction robust learning. By using ResNet models inspiration, further innovates incorporating skip connections into CNN design. enabling direct transmission from earlier later links enhance gradient flow help preserve important spatial information. boosting propagation, integration model's ability recognize minute patterns images, hence classification accuracy. In our methodology, SIPaKMeD dataset has been employed which contains 4049 images are arranged five different categories. To address class imbalance, Generative Adversarial Networks (GANs) have applied data augmentation; is, synthetic created, diversity robustness same. The present astonishingly accurate classifying types: koilocytes, superficial-intermediate, parabasal, dyskeratotic, metaplastic, significantly enhancing cancer. gives excellent performance because validation accuracy 99.11% training 99.82%. reliable cells ensures advancement computer-assisted system.

Язык: Английский

Процитировано

1