Segmentation of Histopathological Images with LinkNet Model Supported by Vgg16 Backbone

Furkan Atlan

Makale detayları

Segmentation of Histopathological Images with LinkNet Model Supported by Vgg16 Backbone

Yazarlar:

Furkan Atlan

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Artificial Intelligence (Other)

Dergi:

Scientific Journal of Mehmet Akif Ersoy University Vol.8 Issue 1, pp.35-46

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001338994

pISSN:

2651-3722

Anahtar kelimeler:

Özet:

 Nuclei segmentation in histopathological images is crucial for the processing and analysis of medical images. Manual segmentation of nuclei images is challenging due to subjective errors by experts and image noise. Before the use of artificial intelligence in medical image analysis, segmentation tasks were performed with common classical methods such as thresholding and watershed. The development of deep learning has led to the emergence of models specifically designed for segmentation tasks. In this study, LinkNet model supported with Vgg16 backbone is proposed for segmenting histopathological images in CryoNuSeg dataset created for nucleus segmentation. After a small number of images are multiplied with data augmentation, feature maps are generated using the Vgg16 model integrated into the encoder of the LinkNet architecture. The results obtained in this study, with F1 Score, Intersection over Union IoU, and Aggregated Jaccard Index AJI values of 0.8447, 0.7312, and 0.7312 respectively, demonstrate superior performance compared to recent studies utilizing the same dataset.

 Nuclei segmentation in histopathological images is crucial for the processing and analysis of medical images. Manual segmentation of nuclei images is challenging due to subjective errors by experts and image noise. Before the use of artificial intelligence in medical image analysis, segmentation tasks were performed with common classical methods such as thresholding and watershed. The development of deep learning has led to the emergence of models specifically designed for segmentation tasks. In this study, LinkNet model supported with Vgg16 backbone is proposed for segmenting histopathological images in CryoNuSeg dataset created for nucleus segmentation. After a small number of images are multiplied with data augmentation, feature maps are generated using the Vgg16 model integrated into the encoder of the LinkNet architecture. The results obtained in this study, with F1 Score, Intersection over Union IoU, and Aggregated Jaccard Index AJI values of 0.8447, 0.7312, and 0.7312 respectively, demonstrate superior performance compared to recent studies utilizing the same dataset.

 Nuclei segmentation in histopathological images is crucial for the processing and analysis of medical images. Manual segmentation of nuclei images is challenging due to subjective errors by experts and image noise. Before the use of artificial intelligence in medical image analysis, segmentation tasks were performed with common classical methods such as thresholding and watershed. The development of deep learning has led to the emergence of models specifically designed for segmentation tasks. In this study, LinkNet model supported with Vgg16 backbone is proposed for segmenting histopathological images in CryoNuSeg dataset created for nucleus segmentation. After a small number of images are multiplied with data augmentation, feature maps are generated using the Vgg16 model integrated into the encoder of the LinkNet architecture. The results obtained in this study, with F1 Score, Intersection over Union IoU, and Aggregated Jaccard Index AJI values of 0.8447, 0.7312, and 0.7312 respectively, demonstrate superior performance compared to recent studies utilizing the same dataset.

 Nuclei segmentation in histopathological images is crucial for the processing and analysis of medical images. Manual segmentation of nuclei images is challenging due to subjective errors by experts and image noise. Before the use of artificial intelligence in medical image analysis, segmentation tasks were performed with common classical methods such as thresholding and watershed. The development of deep learning has led to the emergence of models specifically designed for segmentation tasks. In this study, LinkNet model supported with Vgg16 backbone is proposed for segmenting histopathological images in CryoNuSeg dataset created for nucleus segmentation. After a small number of images are multiplied with data augmentation, feature maps are generated using the Vgg16 model integrated into the encoder of the LinkNet architecture. The results obtained in this study, with F1 Score, Intersection over Union IoU, and Aggregated Jaccard Index AJI values of 0.8447, 0.7312, and 0.7312 respectively, demonstrate superior performance compared to recent studies utilizing the same dataset.

Makale detayları

Segmentation of Histopathological Images with LinkNet Model Supported by Vgg16 Backbone

Yazarlar:

Furkan Atlan

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Artificial Intelligence (Other)

Dergi:

Scientific Journal of Mehmet Akif Ersoy University Vol.8 Issue 1, pp.35-46

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001338994

pISSN:

2651-3722

Medical image segmentation

CryoNuSeg

backbone

Vgg16

LinkNet

Sayı

Makale detayları

Segmentation of Histopathological Images with LinkNet Model Supported by Vgg16 Backbone

Yazarlar: Furkan Atlan

Yayıncı: Burdur Mehmet Akif Ersoy University

Ana Başlıklar: Artificial Intelligence (Other)

Dergi: Scientific Journal of Mehmet Akif Ersoy University Vol.8 Issue 1, pp.35-46

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: İngilizce

Document ID (Org No.): 2001338994

pISSN: 2651-3722

Medical image segmentation

CryoNuSeg

backbone

Vgg16

LinkNet

Sayı

Yazarlar:

Furkan Atlan

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Artificial Intelligence (Other)

Dergi:

Scientific Journal of Mehmet Akif Ersoy University Vol.8 Issue 1, pp.35-46

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001338994

pISSN:

2651-3722