Search

Many women around the world die due to breast cancer. If breast cancer is treated in the early phase, mortality rates may significantly be reduced. Quite a number of approaches have been proposed to help in the early detection of breast cancer. A novel hybrid feature selection model is suggested in this study. This novel hybrid model aims to build an efficient feature selection method and successfully classify breast lesions. A combination of relief and binary Harris hawk optimization (BHHO) hybrid model is used for feature selection. Then, k-nearest neighbor (k-NN), support vector machine (SVM), logistic regression (LR) and naive Bayes (NB) methods are preferred for the classification task. The suggested hybrid model is tested by three different breast cancer datasets which are Wisconsin diagnostic breast cancer dataset (WDBC), Wisconsin breast cancer dataset (WBCD) and mammographic breast cancer dataset (MBCD). According to the experimental results, the relief and BHHO hybrid model improves the performance of all classification algorithms in all three datasets. For WDBC, relief-BHO-SVM model shows the highest classification rates with an of accuracy of 98.77%, precision of 97.17%, recall of 99.52%, F1-score of 98.33%, specificity of 99.72% and balanced accuracy of 99.62%. For WBCD, relief-BHO-SVM model achieves of accuracy of 99.28%, precision of 98.76%, recall of 99.17%, F1-score of 98.96%, specificity of 99.56% and balanced accuracy of 99.36%. Relief-BHO-SVM model performs the best with an accuracy of 97.44%, precision of 97.41%, recall of 98.26%, F1-score of 97.84%, specificity of 97.47% and balanced accuracy of 97.86% for MBCD. Furthermore, the relief-BHO-SVM model has achieved better results than other known approaches. Compared with recent studies on breast cancer classification, the suggested hybrid method has achieved quite good results.

COVID-19 is a deadly virus that first appeared in late 2019 and spread rapidly around the world. Understanding and classifying computed tomography images (CT) is extremely important for the diagnosis of COVID-19. Many case classification studies face many problems, especially unbalanced and insufficient data. For this reason, deep learning methods have a great importance for the diagnosis of COVID-19. Therefore, we had the opportunity to study the architectures of NasNet-Mobile, DenseNet and Nasnet-Mobile+DenseNet with the dataset we have merged. The dataset we have merged for COVID-19 is divided into 3 separate classes: Normal, COVID-19, and Pneumonia. We obtained the accuracy 87.16%, 93.38% and 93.72% for the NasNet-Mobile, DenseNet and NasNet-Mobile+DenseNet architectures for the classification, respectively. The results once again demonstrate the importance of Deep Learning methods for the diagnosis of COVID-19.

Kidney cancer is difficult to diagnose and it can be quite complicated for physicians to diagnose. In this study, while providing information about multiple sources to help people who are dealing with the challenges of the diagnosis of kidney cancer, in order to serve as a guide the principles of kidney cancer are tried to be explained. In recent years, many new methods of treatment have been developed for kidney cancer, and some are under development by scientists. These studies provide treatment information that offers new hope to the lives of kidney cancer patients. In this study, it is aimed to get acquainted with kidney cancer cells by using machine learning, and deep learning algorithms. In this way, an application can be developed to guide patients and physicians through early diagnosis and classification.

Dünya çapında her yıl yüz binlerce insana böbrek kanseri teşhisi konulmaktadır. Böbrek kanseri özellikle gelişmiş ve sanayileşmiş ülkelerde daha yaygın olarak görülmektedir. Günümüzde böbrek kanseri tanısı alan hastaların yaklaşık %30'u maalesef metastatik hastalık aşamasında (uzak organlara yayılmış) iken fark edilmektedir. Böbrek tümörleri, ileri yaştaki insanlar da görülme olasılığı daha yüksek olan bir kanser türüdür. Bu nedenle yaşamın ilerleyen dönemlerinde dikkatli olmak ve tanısal testler yaptırmak son derece önemlidir. Tıbbi görüntüleme ve derin öğrenme yöntemleri bu anlamda giderek daha önemli hale gelmektedir. Bu sebeple doktorların tümörleri başarılı bir şekilde fark edebilmesine yardımcı olacak derin öğrenme modellerinin geliştirilmesi büyük önem taşımaktadır. Bununla birlikte, böbrek ve prostat gibi, kanser tespiti nispeten zor olan yumuşak doku organları için başarılı sistemler oldukça azdır. Böbrek hastalıklarının bilgisayar destekli tedavi planlamasında segmentasyon önemli bir süreçtir. Doğru segmentasyon elde etmek için genellikle derin öğrenme tabanlı çalışmaların yapılması gerekir. Yapılan tez çalışmasında, böbrek tümörleri ile mücadele eden uzman kişilere yardımcı olunabilmesi için mevcut U-Net ve V-Net modelleri geliştirmiş ve yeni modeller önerilmiştir. Bilindiği üzere segmentasyonun zor olduğu pek çok durumda çoğunlukla U-Net ve V-Net tabanlı modeller kullanılmaktadır. Tez çalışmasında ise U-Net ve V-Net modellerinin üstün özelliklerini kullanarak daha iyi sonuçlar üretebilen yeni modeller önerilmiştir. Önerilen modeller, daha önce uygulanmamış kodlayıcı ve kod çözücü aşamalarında iyileştirmelerle daha başarılı bir sistemi temsil etmektedir. Yeni modellerin, özellikle böbrek ve böbrek tümörü segmentasyonuna odaklanan hekimlere yardımcı olabileceği düşünülmektedir. Önerilen her iki model de segmentasyonda mevcut görüntüleme modellerine göre daha iyi performans göstermiş, esnek yapısı ve uygulanabilirliği sayesinde tüm sistemlere kolayca entegre edilebilir şekilde hazırlanmıştır. Bu amaçla geliştirilmiş U-Net modeli böbrek ve tümörleri için sırasıyla %97,8 ve %85,6, Hibrit V-Net modeli ise sırasıyla %97,7 ve %86,5'lik ortalama zar katsayısına ulaşmıştır. Her iki modelde mevcut sonuçlar doğrultusunda yumuşak doku organ segmentasyonu için güvenilir bir yöntem olarak rahatlıkla kullanılabilir. Kidney cancer is a common disease, especially in developed and industrialized countries. Since renal tumors, noticed in the metastatic stage (spread to distant organs) in nearly 30% of patients with kidney cancer, are more likely to develop in the elderly, it is necessary to be cautious and to provide diagnostic tests for people of advanced age Medical imaging and deep learning methods are increasingly becoming more important in this sense. However, not many successful systems exist for soft-tissue organs such as the kidney and prostate, where tumors are relatively difficult to detect. Segmentation is an important process in the computer-aided treatment planning of kidney diseases. For this reason, deep learning-based studies are often required to achieve correct segmentation. As U-Net /V-Net- based models are mostly used for many difficult segmentation processes, this study focused on new models designed with the superior features of the existing U-Net and V-Net models. The proposed hybrid models represent more successful systems with improvements to the encoder and decoder stages that were not previously implemented. We believe that physicians who focus specifically on kidney and kidney tumor segmentation will find these new models helpful. The new models performed better than existing imaging models in segmentation and were prepared to be easily integrated into all systems thanks to their flexible structure and applicability. The hybrid U-Net model developed in this study reached average Dice coefficients of 97.8% and 85.6% for kidney and tumor segmentation, respectively. The Hybrid V-Net model designed here showed average Dice coefficients of 97.7% and 86.5% for kidney and tumor segmentation, respectively. Both architectures proved to be reliable methods for soft tissue organ segmentation.

Kidney cancer is an important type of cancer that spreads rapidly today. Although many treatment methods for kidney cancer have been developed in recent years, current studies are still ongoing. These studies enable treatment information that offers new hope to the lives of kidney cancer patients. When the studies are examined, it seems to be an important alternative in medical segmentation. Although the disease can progress insidiously, sometimes patients may not even have a serious complaint until the last stage. Therefore, segmentation is important for early diagnosis and diagnosis. In this study, it has been prepared in mind in order to help physicians. Here, successful results were obtained by making improvements on the Multi Depth V-Net model. The membrane coefficient of 0.949 and 0.944 for Multi Depth V-Net model and V-Net model kidney segmentation, and 0.841 and 0.830 for tumor segmentation, respectively. In line with the data obtained, we can say that V-Net models for kidney and tumor segmentation can be applied and give accurate results., Böbrek kanseri günümüzde hızla yayılan önemli bir kanser türüdür. Son yıllarda, böbrek kanseri için birçok tedavi yöntemi geliştirilmekle birlikte mevcut çalışmalar halen devam etmektedir. Bu çalışmalar, böbrek kanseri hastalarının hayatlarına yeni bir umut sunan tedavi bilgilerini mümkün kılmaktadır. Çalışmalar incelendiğinde tıbbi segmentasyonda önemli bir alternatif gibi gözükmektedir. Hastalık sinsi ilerleyebilmekle beraber bazen son evreye kadar hastalarda ciddi bir şikâyet bile olmayabilir. Bu yüzden segmentasyon erken tanı ve teşhis için önem arz etmektedir. Bu çalışmada da hekimlere yardımcı olabilmek amacıyla düşünülerek hazırlanmıştır. Burada Multi Depth V-Net modeli üzerinde iyileştirmeler yapılarak başarılı sonuçlar elde edilmiştir. Multi Depth V-Net model ve V-Net model böbrek segmentasyonu için sırasıyla 0,949 ve 0,944 zar katsayısı, tümör segmentasyonu için de 0,841 ve 0,830 zar katsayısına ulaşmıştır. Elde edilen veriler doğrultusunda böbrek ve tümör segmentasyonu için V-Net modellerin uygulanabilir ve doğru sonuçlar verebildiğini söyleyebiliriz.

Worldwide, hundreds of thousands of people are diagnosed with kidney cancer and this disease is more common in developed and industrialized countries. Previously, kidney cancer was known as an elderly disease and was seen in people over a certain age; nowadays it is also seen in younger individuals and it is easier to diagnose thanks to new radiological diagnostic methods. A kidney tumor is a type of cancer that is extremely aggressive and needs surgical treatment rapidly. Today, approximately 30% of patients diagnosed with kidney cancer are unfortunately noticed at the stage of metastatic disease (spread to distant organs). The biggest factor that pushes us to this study is that kidney tumors progress unlike other cancer types with little or no symptoms. Therefore, conducting such studies is extremely important for early diagnosis. In this study, we compare the Unet3D models in order to help people who are dealing with difficulties in the diagnosis of kidney cancer. Unet, Unet+ResNet and Unet++ models were compared for image segmentation.

3rd International Symposium on Multidisciplinary Studies and Innovative Technologies, ISMSIT 2019 -- 11 October 2019 through 13 October 2019 -- -- 156063 Kidney cancer is one of the types of cancer that can be difficult to diagnose and can be very complicated for physicians to diagnose. Especially in recent years, many new treatment methods for kidney cancer have been developed and some of them are still under development by scientists. These studies enable new treatment modalities for kidney cancer patients. In addition, renal tumors are one of the most insidious progressive tumor types. Many times it can be mistaken for other diseases. Especially until the last stage, patients may not even have a serious complaint. Therefore, conducting such studies is very important for early diagnosis. In this study, it is tried to segmentation with deep learning methods in order to help people who are dealing with difficulties of kidney cancer diagnosis. For this reason, Unet and Unet-ResNet models were compared. The Unet-ResNet model achieved 90.2% success for renal tumor segmentation, while the Unet model achieved 44.3% success for renal tumor segmentation. These results shed light on how successful and necessary the Unet-ResNet model can be in particular in studies on image segmentation. © 2019 IEEE.