Transfer learning based CNN models for classification of abnormalities in kidney CT images

Kidney diseases can severely impact the health of a person. The early diagnosis and medication play an import role in the control of disease and to avoid any undue effects. Convolution neural networks have been proven to be effective in the identification of diseases from medical images. In the pres...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Malhotra, Priyanka, Kaur, Swapandeep, Singh, Rajvir
Format: Tagungsbericht
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Kidney diseases can severely impact the health of a person. The early diagnosis and medication play an import role in the control of disease and to avoid any undue effects. Convolution neural networks have been proven to be effective in the identification of diseases from medical images. In the present study, transfer learning techniques are used to fine tune four different pre-trained CNN models: VGG16, InceptionV3, ResNet50 and ResNet101 to classify the kidney CT-scan images into different categories. The pre-trained models with transfer learning shows effectiveness in classifying CT images into the category of: Normal, Cyst, Stone and Tumor images. The results demonstrate that the proficiency of ResNet101 model for classification of CT images with overall model accuracy of 0.99.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0228316