Koronavirüs (COVID-19), solunum yolu enfeksiyonuna neden olan ve insandan insana geçebilen bulaşıcı bir virüstür. Bu virüs dünyada kısa sürede etkili olmuş ve bir salgına dönüşmüştür. Bu tür bulaşıcı hastalıkların erken teşhisi ve gerekli tedavinin erken süreçte başlatılması gerekmektedir. COVID-19 hastalığı tespiti için akciğer görüntülerinden ve ağız yoluyla alınan tükrük ile tespit edilmektedir. COVID-19 hastasını RT-PCR (Reverse Transcription- Polymerase Chain Reaction) ile tespit etmek için yaklaşık 4-6 saat sürmektedir. Pandeminin büyüklüğüne bakıldığında çok ta hızlı sayılmamaktadır. Aynı zamanda test kitinin de bir maliyeti bulunmaktadır. Ekonomik olarak güçlü olmayan ülkeler RT-PCR kitlerine erişmekte sorun yaşamaktadır. Pandemi döneminde zorlu süreçlerden bir tanesi her raporu manuel olarak incelemek için, birden fazla radyoloji uzmanı gerekmektedir. Bu çalışmada makine öğrenmesi algoritmaları ile farklı kategorilerdeki akciğer tomografisi görüntülerinden COVID-19 olan görüntü tespit edilmiştir. Orange Data Mining Veri analizi programında makine öğrenmesi algoritması olan K-En Yakın Komşuluk, Yapay Sinir Ağları, Rastgele Orman ve Destek Vektör algoritmaları ile Akciğer veri setinden COVİD-19 hastalığına ait görüntüler sınıflandırılmış, en iyi sonucu Destek Vektör Algoritması ile elde edilmiştir.
Coronavirus (COVID-19) is a infectious virus that causes respiratory infection and can be transmitted from man to man. This virus has become effective in the world shortly and has become an epidemic. Early diagnosis of these infectious diseases and the necessary treatment must be initiated in the early process. COVID-19 is detected by lung images and sputum taken through the mouth to detect the disease. COVID-19 takes about 4-6 hours to detect the patient with RT-PCR (Reverse Transcription-Polymerase Chain Reaction). When it comes to the size of the pandemic, it is not so fast. The test kit also has a cost. Economically non-powerful countries have trouble accessing RT-PCR kit. One of the challenging processes during the pandemic period requires more than one radiology specialist to manually examine each report. In this study, the machine learning algorithms and the image COVID-19 was detected from lung tomography images in different categories. In the Orange Data Mining Data Analysis program, the machine learning algorithm is classified with the K-Last Neighborhood, Artificial Neural Networks, Rastgage Forest and Support Vector algorithms and the images of the lung data set of COVID-19 disease, the best result is obtained with the Support Vector algorithm.
Coronavirus (COVID-19) is a contagious virus that causes respiratory tract infection and can be passed from person to person. This virus became effective in the world in a short time and turned into an epidemic. Early diagnosis of such infectious diseases and the necessary treatment should be initiated in the early period. For the detection of COVID-19 disease, it is detected by lung images and oral saliva. It takes approximately 4-6 hours to detect a COVID-19 patient by RT-PCR (Reverse Transcription- Polymerase Chain Reaction). Considering the size of the pandemic, it is not considered very fast. At the same time, the test kit has a cost. Countries that are not economically strong have problems accessing RT-PCR kits. One of the challenging processes during the pandemic period is to manually review each report, requiring multiple radiologists. In this study, the images with COVID-19 were detected from different categories of lung tomography images with machine learning algorithms. In the Orange Data Mining data analysis program, the images of the COVID-19 disease from the Lung data set were classified with the machine learning algorithm K-Nearest Neighborhood, Artificial Neural Networks, Random Forest and Support Vector algorithms, and the best result was obtained with the Support Vector Algorithm.
Alan : Fen Bilimleri ve Matematik; Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|