Classification of White Blood Cells using the Squeeze-Excitation Residual Network

Hüseyin Fırat

Makale detayları

Classification of White Blood Cells using the Squeeze-Excitation Residual Network

Yazarlar:

Hüseyin Fırat

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Bilişim Teknolojileri Dergisi Vol.16 Issue 3, pp.189-205

Belge Tipi:

Makale

Yayın Tarihi:

7/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001296199

pISSN:

1307-9697

Anahtar kelimeler:

Özet:

 White blood cells WBCs are an important component of the immune system that plays an active role in protecting the body against microorganisms such as parasites, bacteria and viruses. The structural features of WBCs, the shapes and numbers of their subtypes can provide important information about human health. Accurate WBC detection is clinically very important in the diagnosis of the disease. Accordingly, an accurate WBC classification method is of critical importance. In this study, a CNN-based method for WBC classification is proposed. The proposed method is a hybrid method consisting of a combination of squeeze-excitation SE network and residual network ResNet architecture. The problems that may occur as the number of layers increase in deep network architectures can be solved with ResNet. The use of the SE block with ResNet increases the classification accuracy while minimally increasing the total number of parameters. At the same time, combining the SE block with the ResNet improves the performance of traditional ResNets. The BCCD dataset from the Kaggle database was used to test the performance of the proposed method. As a result of the applications, an average of 99.92 accuracy, 99.85 precision, recall and F1-score were obtained. These results were compared with the results obtained by the CNN methods in recent studies using the BCCD dataset from the literature, and it was seen that the proposed method gave better results with less trainable parameters.

 White blood cells WBCs are an important component of the immune system that plays an active role in protecting the body against microorganisms such as parasites, bacteria and viruses. The structural features of WBCs, the shapes and numbers of their subtypes can provide important information about human health. Accurate WBC detection is clinically very important in the diagnosis of the disease. Accordingly, an accurate WBC classification method is of critical importance. In this study, a CNN-based method for WBC classification is proposed. The proposed method is a hybrid method consisting of a combination of squeeze-excitation SE network and residual network ResNet architecture. The problems that may occur as the number of layers increase in deep network architectures can be solved with ResNet. The use of the SE block with ResNet increases the classification accuracy while minimally increasing the total number of parameters. At the same time, combining the SE block with the ResNet improves the performance of traditional ResNets. The BCCD dataset from the Kaggle database was used to test the performance of the proposed method. As a result of the applications, an average of 99.92 accuracy, 99.85 precision, recall and F1-score were obtained. These results were compared with the results obtained by the CNN methods in recent studies using the BCCD dataset from the literature, and it was seen that the proposed method gave better results with less trainable parameters.

 White blood cells WBCs are an important component of the immune system that plays an active role in protecting the body against microorganisms such as parasites, bacteria and viruses. The structural features of WBCs, the shapes and numbers of their subtypes can provide important information about human health. Accurate WBC detection is clinically very important in the diagnosis of the disease. Accordingly, an accurate WBC classification method is of critical importance. In this study, a CNN-based method for WBC classification is proposed. The proposed method is a hybrid method consisting of a combination of squeeze-excitation SE network and residual network ResNet architecture. The problems that may occur as the number of layers increase in deep network architectures can be solved with ResNet. The use of the SE block with ResNet increases the classification accuracy while minimally increasing the total number of parameters. At the same time, combining the SE block with the ResNet improves the performance of traditional ResNets. The BCCD dataset from the Kaggle database was used to test the performance of the proposed method. As a result of the applications, an average of 99.92 accuracy, 99.85 precision, recall and F1-score were obtained. These results were compared with the results obtained by the CNN methods in recent studies using the BCCD dataset from the literature, and it was seen that the proposed method gave better results with less trainable parameters.

 Beyaz kan hücreleri, vücudun parazitler, bakteriler, virüsler gibi mikroorganizmalara karşı korunmasında etkin rol oynayan bağışıklık sisteminin önemli bir bileşenidir. Beyaz kan hücrelerinin yapısal özellikleri, alt türlerinin şekilleri ve sayıları insan sağlığı hakkında önemli bilgiler verebilmektedir. Hastalık teşhisinde doğru beyaz kan hücre tespiti klinik olarak oldukça önemlidir. Bu yüzden, doğru beyaz kan hücre sınıflandırma yöntemi kritik öneme sahiptir. Bu çalışmada, beyaz kan hücre sınıflandırması için Evrişimsel sinir ağı ESA tabanlı bir yöntem önerilmiştir. Önerilen yöntem sıkma-uyarma ağı ile artık ağ mimarisinin birleşiminden oluşan hibrit bir yöntemdir. Derin ağ mimarilerinde katman sayısı arttıkça oluşabilecek problemler artık ağ ile çözülebilmektedir. Sıkma-uyarma SU bloğunun artık ağ ile birlikte kullanımı, toplam parametre sayısını minimum düzeyde arttırırken sınıflandırma doğruluğunu arttırmakatdır. Aynı zamanda, SU bloğunun artık ağ ile birleştirilmesi geleneksel artık ağların performansını da arttırmaktadır. Önerilen yöntemin performansını test etmek için Kaggle veritabanından alınan BCCD veriseti kullanılmıştır. Uygulamalar sonucunda ortalama 99,92 doğruluk, 99,85 kesinlik, duyarlılık ve F1-skoru elde edilmiştir. Bu sonuçlar, literatürden BCCD verisetini kullanan son yıllardaki çalışmalarda yer alan ESA yöntemlerinin elde ettiği sonuçlarla karşılaştırıldı ve önerilen yöntemin daha az eğitilebilir parametre ile daha iyi sonuçlar verdiği görülmüştür.

Makale detayları

Classification of White Blood Cells using the Squeeze-Excitation Residual Network

Yazarlar:

Hüseyin Fırat

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Bilişim Teknolojileri Dergisi Vol.16 Issue 3, pp.189-205

Belge Tipi:

Makale

Yayın Tarihi:

7/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001296199

pISSN:

1307-9697

White blood cell

classification

squeeze-excitation residual network

convolutional neural networkBeyaz kan hücresi

sınıflandırma

sıkma-uyarma artık ağı

evrişimsel sinir ağı

Sayı

Makale detayları

Yazarlar: Hüseyin Fırat

Yayıncı: Gazi University

Ana Başlıklar: Bilgisayar Yazılımı

Dergi: Bilişim Teknolojileri Dergisi Vol.16 Issue 3, pp.189-205

Belge Tipi: Makale

Yayın Tarihi: 7/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001296199

pISSN: 1307-9697

Sayı

Yazarlar:

Hüseyin Fırat

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgisayar Yazılımı

Dergi:

Bilişim Teknolojileri Dergisi Vol.16 Issue 3, pp.189-205

Belge Tipi:

Makale

Yayın Tarihi:

7/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001296199

pISSN:

1307-9697