Chronic kidney disease (CKD) may be treated if diagnosed early, but recovery becomes difficult as the condition advances. Renal replacement treatment, such as transplantation or dialysis, will be required at some point. This paper proposes using VGG16 with principal component analysis (PCA) and only VGG16 for automatic kidney disease classification. A real dataset from Al-Diwaniyah General Teaching Hospital (Lithotripsy Center), Iraq is used in the presented work. The kidney classes are classified into a total of 1260 ultrasound images performed on normal, stone, hydronephrosis, and cyst images. Kidney abnormality detection involves three stages: feature extraction, feature selection, and then classification. The use of feature selection to decrease the retrieved features helps in quicker data training and machine learning. The accuracy, precision, recall, and F1 score of the suggested technique were all evaluated. The VGG16-PCA model has a classification accuracy of 95%.
CITATION STYLE
Alkordy, N. H., Aljobouri, H. K., & Wadi, Z. K. (2023). Feature Extraction and Selection of Kidney Ultrasound Images Using a Deep CNN and PCA. In Lecture Notes in Networks and Systems (Vol. 596 LNNS, pp. 104–114). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-21435-6_10
Mendeley helps you to discover research relevant for your work.