Early Detection and Diagnosis of Breast Cancer Based on Machine Learning Supported by Feature Selection

Cihan Akyel; Hüseyin Polat; Bünyamin Ciylan

Makale detayları

Early Detection and Diagnosis of Breast Cancer Based on Machine Learning Supported by Feature Selection

Yazarlar:

Cihan Akyel | Hüseyin Polat | Bünyamin Ciylan

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Sistemleri Geliştirme Metodolojileri ve Uygulamaları | Karar Desteği ve Grup Destek Sistemleri

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 2, pp.675-690

Belge Tipi:

Makale

Yayın Tarihi:

6/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001321572

pISSN:

2147-9526

Anahtar kelimeler:

Özet:

 Although the exact cause of cancer is not known, it is known that many factors such as lifestyle, environmental factors, nutrition and genetics may contribute to the development of cancer. Among the cancer types, especially breast cancer is a disease with a high incidence among women worldwide. Methods such as physical examination and examination of mammography images are used in the diagnosis of breast cancer. With the developing technology, the use of machine learning applications in the field of medicine is increasing. In this way, there is an increasing number of promising studies that can help doctors diagnose breast cancer at an earlier and faster stage. In this study, the performances of 4 different feature selection and 6 different machine learning methods to be used in the early detection of breast cancer were compared. In the first phase of the study, the number of features in the dataset was reduced by selecting the features that have the most impact on the target feature in the dataset with Principal Component Analysis PCA, Recursive Feature Elimination, Variance inflation factors VIF and Univariate feature selection methods. In the second stage, K Nearest Neighbors KNN, Naive Bayes, Decision Tree, SVM and Random Forest machine learning algorithms were trained and tested based on original and feature-selected datasets. The highest model test performance was obtained by using Variance inflation factors VIF and Random Forest algorithm. According to the test results, 98.83 accuracy, 99 precision and 99 recall values were obtained. In addition, the machine learning model that was trained and tested in the study was turned into an application with a web interface using flask framework.

 Although the exact cause of cancer is not known, it is known that many factors such as lifestyle, environmental factors, nutrition and genetics may contribute to the development of cancer. Among the cancer types, especially breast cancer is a disease with a high incidence among women worldwide. Methods such as physical examination and examination of mammography images are used in the diagnosis of breast cancer. With the developing technology, the use of machine learning applications in the field of medicine is increasing. In this way, there is an increasing number of promising studies that can help doctors diagnose breast cancer at an earlier and faster stage. In this study, the performances of 4 different feature selection and 6 different machine learning methods to be used in the early detection of breast cancer were compared. In the first phase of the study, the number of features in the dataset was reduced by selecting the features that have the most impact on the target feature in the dataset with Principal Component Analysis PCA, Recursive Feature Elimination, Variance inflation factors VIF and Univariate feature selection methods. In the second stage, K Nearest Neighbors KNN, Naive Bayes, Decision Tree, SVM and Random Forest machine learning algorithms were trained and tested based on original and feature-selected datasets. The highest model test performance was obtained by using Variance inflation factors VIF and Random Forest algorithm. According to the test results, 98.83 accuracy, 99 precision and 99 recall values were obtained. In addition, the machine learning model that was trained and tested in the study was turned into an application with a web interface using flask framework.

 Although the exact cause of cancer is not known, it is known that many factors such as lifestyle, environmental factors, nutrition and genetics may contribute to the development of cancer. Among the cancer types, especially breast cancer is a disease with a high incidence among women worldwide. Methods such as physical examination and examination of mammography images are used in the diagnosis of breast cancer. With the developing technology, the use of machine learning applications in the field of medicine is increasing. In this way, there is an increasing number of promising studies that can help doctors diagnose breast cancer at an earlier and faster stage. In this study, the performances of 4 different feature selection and 6 different machine learning methods to be used in the early detection of breast cancer were compared. In the first phase of the study, the number of features in the dataset was reduced by selecting the features that have the most impact on the target feature in the dataset with Principal Component Analysis PCA, Recursive Feature Elimination, Variance inflation factors VIF and Univariate feature selection methods. In the second stage, K Nearest Neighbors KNN, Naive Bayes, Decision Tree, SVM and Random Forest machine learning algorithms were trained and tested based on original and feature-selected datasets. The highest model test performance was obtained by using Variance inflation factors VIF and Random Forest algorithm. According to the test results, 98.83 accuracy, 99 precision and 99 recall values were obtained. In addition, the machine learning model that was trained and tested in the study was turned into an application with a web interface using flask framework.

 Kanserin tam nedeni bilinmemekle birlikte, yaşam tarzı, çevresel faktörler, beslenme ve genetik gibi birçok faktörün kanser gelişimine katkıda bulunabileceği bilinmektedir. Kanser türleri arasında özellikle göğüs kanseri, dünya genelinde kadınlar arasında görülme sıklığı yüksek olan bir hastalıktır. Göğüs kanserinin teşhisinde fiziksel muayene ve mamografi görüntülerinin incelenmesi gibi yöntemler kullanılmaktadır. Gelişen teknolojiyle birlikte makine öğrenmesi uygulamalarının tıp alanında kullanımı giderek artmaktadır. Bu sayede göğüs kanserinin daha erken aşamada ve hızlı şekilde teşhisi konusunda doktorlara yardımcı olabilecek umut verici çalışmalar giderek artmaktadır. Bu çalışmada, göğüs kanserinin erken teşhisinde kullanmak için 4 farklı öznitelik seçimi ve 5 farklı makine öğrenme yönteminin performansları karşılaştırılmıştır. Çalışmanın ilk aşamasında, Principal Component Analysis PCA, Recursive feature elimination, Variance inflation factors VIF ve Univariate feature selection yöntemleri ile veri kümesinde hedef özniteliğe en çok etki eden öznitelikler seçilerek veri kümesindeki öznitelik sayısı azaltılmıştır. İkinci aşamada, K Nearest Neighbors KNN, Naive Bayes, Decision Tree, Support Vector Machine SVM ve Random Forest makine öğrenme algoritmaları orijinal ve öznitelik seçimi yapılmış veri kümelerine dayalı olarak eğitilmiş ve test edilmiştir. Test sonuçlarına göre 98,83 doğruluk, 99 kesinlik ve 99 duyarlılık değerleri ile Variance inflation factors VIF öznitelik seçimi ve Random Forest algoritması kullanılarak elde edilmiştir. Daha az öznitelik kullanımı sayesinde eğitim ve test aşamalarında benzer başarı değerleri, kaynak kullanımı ile sağlanmıştır. Çalışmada eğitilip test edilen makine öğrenme modeli Flask framework kullanılarak bir web ara yüzüne sahip uygulama haline getirilmiştir.

Makale detayları

Early Detection and Diagnosis of Breast Cancer Based on Machine Learning Supported by Feature Selection

Yazarlar:

Cihan Akyel | Hüseyin Polat | Bünyamin Ciylan

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Sistemleri Geliştirme Metodolojileri ve Uygulamaları | Karar Desteği ve Grup Destek Sistemleri

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 2, pp.675-690

Belge Tipi:

Makale

Yayın Tarihi:

6/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001321572

pISSN:

2147-9526

Breast Cancer

Feature Selection

Machine Learning

Decision Support Systems

Information Systems

Data ManagementGöğüs Kanseri

Öznitelik Seçimi

Makine Öğrenmesi

Karar Destek Sistemleri

Bilişim Sistemleri Veri Yönetimi

Sayı

Makale detayları

Yazarlar: Cihan Akyel | Hüseyin Polat | Bünyamin Ciylan

Yayıncı: Gazi University

Ana Başlıklar: Bilgi Sistemleri Geliştirme Metodolojileri ve Uygulamaları | Karar Desteği ve Grup Destek Sistemleri

Dergi: Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 2, pp.675-690

Belge Tipi: Makale

Yayın Tarihi: 6/2024

Belge Dili: Türkçe

Document ID (Org No.): 2001321572

pISSN: 2147-9526

Sayı

Yazarlar:

Cihan Akyel | Hüseyin Polat | Bünyamin Ciylan

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Sistemleri Geliştirme Metodolojileri ve Uygulamaları | Karar Desteği ve Grup Destek Sistemleri

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 2, pp.675-690

Belge Tipi:

Makale

Yayın Tarihi:

6/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001321572

pISSN:

2147-9526