Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Investigations on Color Normalization Technique Using CycleGAN Based Machine Learning Algorithms for Breast Cancer Detection-Data Deployment
https://orcid.org/http://orcid.org/0000-0002-7692-810X
Breast cancer is a widespread malignancy affecting women globally, and early-stage diagnosis offers a cure reduces the death rate. Accurate discern between benign and malignant cells poses significant challenges. Early-stage breast cancer detection offers high cure rates, highlighting the importance of accurate diagnosis. However, distinguishing malignant from benign cells in histopathological images remains challenging due to staining inconsistencies and the expertise required. Improving the accuracy of breast cancer diagnosis is essential for effective treatment and reducing mortality rates through timely therapeutic interventions. To address the challenges of uneven staining in pathological images and improve accuracy in diagnosis, the proposed solution involves color normalization using a CycleGAN (Generative adversarial network) and (Convolution neural network) classification algorithm framework. The suggested methodology has the potential to help pathologists diagnose patients more accurately by combining images taken at various magnifications during the clinical stage. To evaluate the proposed solution’s performance, different architectures such as Resnet150, AlexNet, VGG19, Decision tree and (support vector machine) algorithms have been compared. The comparison is done based accuracy, recall, precision and F1score as performance metrics. The model Resnet150 model has outperformed with an accuracy of 96.91%. A Front-end user-friendly webpage has been developed using Streamlit to facilitate the input of patient data and generate accurate initial prediction results, potentially leading to improved treatment outcomes and survival rates for patients.
Investigations on Color Normalization Technique Using CycleGAN Based Machine Learning Algorithms for Breast Cancer Detection-Data Deployment
https://orcid.org/http://orcid.org/0000-0002-7692-810X
Breast cancer is a widespread malignancy affecting women globally, and early-stage diagnosis offers a cure reduces the death rate. Accurate discern between benign and malignant cells poses significant challenges. Early-stage breast cancer detection offers high cure rates, highlighting the importance of accurate diagnosis. However, distinguishing malignant from benign cells in histopathological images remains challenging due to staining inconsistencies and the expertise required. Improving the accuracy of breast cancer diagnosis is essential for effective treatment and reducing mortality rates through timely therapeutic interventions. To address the challenges of uneven staining in pathological images and improve accuracy in diagnosis, the proposed solution involves color normalization using a CycleGAN (Generative adversarial network) and (Convolution neural network) classification algorithm framework. The suggested methodology has the potential to help pathologists diagnose patients more accurately by combining images taken at various magnifications during the clinical stage. To evaluate the proposed solution’s performance, different architectures such as Resnet150, AlexNet, VGG19, Decision tree and (support vector machine) algorithms have been compared. The comparison is done based accuracy, recall, precision and F1score as performance metrics. The model Resnet150 model has outperformed with an accuracy of 96.91%. A Front-end user-friendly webpage has been developed using Streamlit to facilitate the input of patient data and generate accurate initial prediction results, potentially leading to improved treatment outcomes and survival rates for patients.
Investigations on Color Normalization Technique Using CycleGAN Based Machine Learning Algorithms for Breast Cancer Detection-Data Deployment
https://orcid.org/http://orcid.org/0000-0002-7692-810X
Breast cancer is a widespread malignancy affecting women globally, and early-stage diagnosis offers a cure reduces the death rate. Accurate discern between benign and malignant cells poses significant challenges. Early-stage breast cancer detection offers high cure rates, highlighting the importance of accurate diagnosis. However, distinguishing malignant from benign cells in histopathological images remains challenging due to staining inconsistencies and the expertise required. Improving the accuracy of breast cancer diagnosis is essential for effective treatment and reducing mortality rates through timely therapeutic interventions. To address the challenges of uneven staining in pathological images and improve accuracy in diagnosis, the proposed solution involves color normalization using a CycleGAN (Generative adversarial network) and (Convolution neural network) classification algorithm framework. The suggested methodology has the potential to help pathologists diagnose patients more accurately by combining images taken at various magnifications during the clinical stage. To evaluate the proposed solution’s performance, different architectures such as Resnet150, AlexNet, VGG19, Decision tree and (support vector machine) algorithms have been compared. The comparison is done based accuracy, recall, precision and F1score as performance metrics. The model Resnet150 model has outperformed with an accuracy of 96.91%. A Front-end user-friendly webpage has been developed using Streamlit to facilitate the input of patient data and generate accurate initial prediction results, potentially leading to improved treatment outcomes and survival rates for patients.
Investigations on Color Normalization Technique Using CycleGAN Based Machine Learning Algorithms for Breast Cancer Detection-Data Deployment
J. Inst. Eng. India Ser. B
Kakarla, Deepti (Autor:in) / Sahaja, P. (Autor:in) / Vaishnvai, K. (Autor:in) / Srileka, V. (Autor:in) / Anusha, B. (Autor:in)
Journal of The Institution of Engineers (India): Series B ; 106 ; 315-325
01.02.2025
11 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Data Augmentation by an Additional Self-Supervised CycleGAN-Based for Shadowed Pavement Detection
| DOAJ | 2022
| British Library Online Contents | 2012