Browsing by Author "Dogan, Refika Sultan"

Now showing 1 - 4 of 4

Comparison of deep learning and conventional machine learning methods for classification of colon polyp types
(SCIENDOBOGUMILA ZUGA 32A, WARSAW, MAZOVIA, POLAND, 2021) Dogan, Refika Sultan; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Dogan, Refika Sultan; Yilmaz, Bulent
Determination of polyp types requires tissue biopsy during colonoscopy and then histopathological examination of the microscopic images which tremendously time-consuming and costly. The first aim of this study was to design a computer-aided diagnosis system to classify polyp types using colonoscopy images (optical biopsy) without the need for tissue biopsy. For this purpose, two different approaches were designed based on conventional machine learning (ML) and deep learning. Firstly, classification was performed using random forest approach by means of the features obtained from the histogram of gradients descriptor. Secondly, simple convolutional neural networks (CNN) based architecture was built to train with the colonoscopy images containing colon polyps. The performances of these approaches on two (adenoma & serrated vs. hyperplastic) or three (adenoma vs. hyperplastic vs. serrated) category classifications were investigated. Furthermore, the effect of imaging modality on the classification was also examined using white-light and narrow band imaging systems. The performance of these approaches was compared with the results obtained by 3 novice and 4 expert doctors. Two-category classification results showed that conventional ML approach achieved significantly better than the simple CNN based approach did in both narrow band and white-light imaging modalities. The accuracy reached almost 95% for white-light imaging. This performance surpassed the correct classification rate of all 7 doctors. Additionally, the second task (three-category) results indicated that the simple CNN architecture outperformed both conventional ML based approaches and the doctors. This study shows the feasibility of using conventional machine learning or deep learning based approaches in automatic classification of colon types on colonoscopy images.
Investigation of Hepatocellular Carcinoma Molecular Mechanisms via in Silico Analyses
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2020) Dogan, Refika Sultan; Saka, Samed; Gungor, Burcu Bakir; AGÜ, Mühendislik Fakültesi, Bilgisayar Mühendisliği Bölümü
Hepatocellular carcinoma (HCC) is the most common cause of cancer-related death in the world. The molecular changes in the organism during the development of HCC are not fully understood. The aim of the present study is to contribute to the identification of critical genes and pathways associated with HCC via integrating various bioinformatics methods. In this study, experiments were conducted on gene expression data of 14 HCC tissues and non-cancerous control tissues. A total of 1229 genes, which show a statistically significant change between the groups, were identified. Among these, 681 genes were upregulated and 548 genes were downregulated. Via mapping the detected genes into protein protein interaction networks, active subnetwork search, subnetwork topological analysis and functional enrichment analyses were carried out. The interactions between the molecular pathways affected by HCC were also presented.
Kolon Polipleri için Kolonoskopi ve Histopatoloji Görüntülerinden Yapay Zekâ Destekli Prognostik Belirteç Tespiti
(ELEKTRİK, ELEKTRONİK VE ENFORMATİK ARAŞTIRMA DESTEK GRUBU GRUBU: EEEAG, 2023) Yılmaz, Bülent; Akay, Ebru; Doğan, Serkan; Aydın, Zafer; Dogan, Refika Sultan; Yengec-Tasdemir, Sena Busra; Güzel, Ömer Faruk; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yılmaz, Bülent; Aydın, Zafer; Dogan, Refika Sultan
Kolon kanseri vakalarının çoğu kolon mukozasında anormal hücre çoğalmasından kaynaklanan poliplerle başlar. Bu projede Kayseri Şehir Hastanesi gastroenteroloji kliniğine gelen 201 hastada tespit edilen poliplere dair kolonoskopi video ve görüntülerinden ve biyopsi örneklerinden elde edilen patoloji raporu ve immunohistokimyasal (İHK) gen ve protein analizi sonuçlarını içeren kapsamlı bir veri seti oluşturulmuştur. Bu projede, elde ettiğimiz veri setinde yer alan görüntülerden kolon poliplerinin evresini/patolojisini tahmin etmek için yenilikçi derin öğrenme ve makine öğrenmesi yöntemlerini temel alan çevrim içi veya dışı kullanılabilen kapsamlı bir yapay zekâ destekli bilgisayarlı görü sistemi geliştirilmiştir. Bu proje kapsamında; kolonoskopi videolarından gerçek zamanlı polip lokalizasyonu, videolardan görüntülerin elde edilmesi, polip görüntülerinden hiperplastik ve tübüler polip ayrımının otomatik yapılması ve hekim performansıyla karşılaştırılması, bu görüntüler üzerinde ayırt edici özniteliklerin tespit edilmesi, farklı büyütmelerde alınan histopatoloji görüntülerinden adenomatöz olan ve olmayan poliplerin ve poliplerin alt tiplerinin yenilikçi derin öğrenme yöntemleriyle tespiti, Ki-67, p53, VEGF, PDL-lenfosit ve PDL-epitel, BRAF ve cd34 isimli gen ve proteinlerin İHK analizlerinin sonuçlarının polip tipleri ve alt tipleri için yorumlanması ve poliplerin bu bilgilere göre etiketlenmesi gerçekleştirilmiştir.
Polyp Localization in Colonoscopy Images Using Vessel Density
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Dogan, Refika Sultan; Yilmaz, Bulent; 0000-0001-8416-1765; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
In this paper, we present a new approach for polyp localization in colonoscopy images. This approach is based on the determination of the polyp location using the vessel density in colon images. Primarily, we used pre-processing procedures on the colon images, and then blood vessel extraction techniques were employed. Later, segmentation of the vessel boundaries was performed. With the help of vessel boundaries we calculated the vessel density, and used this for the localization of the polyps. We tested the success of this approach using a publicly available image set (CVC-ClinicDB database). This database consisted of 612 images from 29 different polyps. This approach succeeds in correct detection of 24 out of 29 different polyps.