Publication: Sperm morfolojisinin sınıflandırılmasında evrişimsel sinir ağ ve görü dönüştürücü modellerin performans analizi
Loading...
Files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Dünya genelinde kısırlık problemi gün geçtikçe daha büyük bir sorun haline gelmektedir ve problemin görülme sıklığının yaklaşık yarısında erkek faktörü etkilidir. Son zamanlarda, sperm kalitesini ölçmek için geliştirilen bilgisayar destekli sperm analiz sistemleri ile birlikte yapay zeka uygulamaları da kullanılmaya başlanmıştır. Özellikle evrişimsel sinir ağları (ESA), sperm hücrelerinin sınıflandırılmasında kullanılan en yaygın yöntemlerden biridir. Son yıllarda popüler hale gelen ve görüntü sınıflandırma problemlerinde kullanılmaya başlanan bir diğer yöntem ise görü dönüştürücü(GD) (vision transformer (ViT)) modelleridir. Önerilen çalışma kapsamında erişime açık olan insan sperm hücre görüntü veri setlerinden HuSHeM, SMIDS ve SCIAN veri setleri üzerinde ESA ve GD mimarilerinin karşılaştırmalı olarak performans analizleri yapılmıştır. Geleneksel 5 farklı ESA mimarisi ve 3 farklı varyantta GD modeli olmak üzere toplam 8 adet model belirtilen veri setlerinin sınıflandırma performansları üzerinden kıyaslanmıştır. 5 kat çapraz doğrulama ve veri artırımı yapılarak modeller eğitilmiş ve elde edilen sonuçlar sunulmuştur. Sonuçlar t-test yöntemi ile kıyaslanarak performans analizi yapılmıştır. Her bir model 3 farklı veri seti üzerinde 7 farklı model ile kıyaslanarak toplam 21 karşılaştırma işlemi yapılmıştır. Yapılan karşılaştırmalar sonucunda ViT-L16 modeli ile 12 Kazanma, 9 Beraberlik ve 0 yenilgi alınmıştır. Kazanma oranları karşılaştırıldığında en yakın modelden yaklaşık %38 daha fazla kazanma oranına sahiptir.
The problem of infertility around the world is becoming more severe day by day. About half of the problem incidence is male-related. Computer Aided Sperm Analysis systems, accompanied by the usage of artificial intelligence, have been recently developed to measure the sperm quality. In particular, convolutional neural networks (CNNs) are one of the most widely used method for classifying sperm cells. Another method that has become popular in recent years and has started to be used in image classification problems is the vision transformer (ViT) models. In the proposed study, a comparative performance analysis of CNN and ViT architectures on open source HuSHeM, SMIDS and SCIAN human sperm cell image datasets was performed. The results obtained from 8 models, including 5 different traditional CNN architectures and 3 different variants of ViT models, were compared with each other. After 5-fold cross validation and data augmentation, the models were trained and the results were obtained. The results obtained were validated with t-test and the performance analysis was performed. Each model was compared with 7 different models on 3 different datasets for a total of 21 comparisons. As a result of the comparisons, 12 wins, 9 draws and 0 defeats were obtained with the ViT-L16 model. When comparing win rates, it has about %38 more win rates than its closest model.
The problem of infertility around the world is becoming more severe day by day. About half of the problem incidence is male-related. Computer Aided Sperm Analysis systems, accompanied by the usage of artificial intelligence, have been recently developed to measure the sperm quality. In particular, convolutional neural networks (CNNs) are one of the most widely used method for classifying sperm cells. Another method that has become popular in recent years and has started to be used in image classification problems is the vision transformer (ViT) models. In the proposed study, a comparative performance analysis of CNN and ViT architectures on open source HuSHeM, SMIDS and SCIAN human sperm cell image datasets was performed. The results obtained from 8 models, including 5 different traditional CNN architectures and 3 different variants of ViT models, were compared with each other. After 5-fold cross validation and data augmentation, the models were trained and the results were obtained. The results obtained were validated with t-test and the performance analysis was performed. Each model was compared with 7 different models on 3 different datasets for a total of 21 comparisons. As a result of the comparisons, 12 wins, 9 draws and 0 defeats were obtained with the ViT-L16 model. When comparing win rates, it has about %38 more win rates than its closest model.
Description
Keywords
Bilgisayar Bilimleri, Yapay Zeka, Bilgisayarda Öğrenme ve Örüntü Tanıma, Biyomedikal Mühendisliği, Biyomedikal Görüntü İşleme, Mühendislik ve Teknoloji, Computer Sciences, Artificial Intelligence, Computer Learning and Pattern Recognition, Biomedical Engineering, Biomedical Image Processing, Engineering and Technology, Mühendislik, Bilişim ve Teknoloji (ENG), Bilgisayar Bilimi, Mühendislik, BİLGİSAYAR BİLİMİ, YAPAY ZEKA, MÜHENDİSLİK, BİYOMEDİKAL, Engineering, Computing & Technology (ENG), COMPUTER SCIENCE, ENGINEERING, COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE, ENGINEERING, BIOMEDICAL, Biyomedikal mühendisliği, Bilgisayarla Görme ve Örüntü Tanıma, Bilgisayar Bilimi Uygulamaları, Yapay Zeka, Bilgisayar Bilimi (çeşitli), Genel Bilgisayar Bilimi, Biyomühendislik, Fizik Bilimleri, Computer Vision and Pattern Recognition, Computer Science Applications, Artificial Intelligence, Computer Science (miscellaneous), General Computer Science, Bioengineering, Physical Sciences
Citation
Aktas A., Serbes G., Ilhan H. O., \"Sperm Morfolojisinin Sınıflandırılmasında Evrişimsel Sinir Ağ ve Görü Dönüştürücü Modellerin Performans Analizi\", 2023 8th International Conference on Computer Science and Engineering (UBMK), Burdur, Türkiye, 13 - 15 Eylül 2023, ss.330-335
