CLASSIFICATION OF DERMATOLOGIC MANIFESTATIONS OF CARDIOVASCULAR DISEASE USING EFFICIENTNETV2 CNN MODEL | ||
| International Journal of Intelligent Computing and Information Sciences | ||
| Volume 23, Issue 1, March 2023, Pages 115-127 PDF (718.71 K) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/ijicis.2023.184311.1242 | ||
| Authors | ||
| Abraham Eseoghene Evwiekpaefe* 1; Oghenegueke Fortune Amrevuawho2 | ||
| 1Department of Computer Science, Faculty of Military Science and Interdisciplinary Studies, Nigerian Defence Academy, Kaduna, Nigeria | ||
| 2Department of Computer Science, Faculty of Military Science and Interdisciplinary Studies, Nigerian Defence Academy, Kaduna, Nigeria. | ||
| Abstract | ||
| The skin is one of the organs of the human body where various internal health problems including cardiovascular diseases tend to show some notable signs and symptoms. The dermatologist may be one of the first clinician to recognize that someone does have cardiovascular disease because warning signs can develop on the skin. The aim of this research is to use the efficientNetV2 model for the classification of dermatologic manifestations of cardiovascular disease based on transfer learning. The EfficientNetV2 model was modified and trained as a classifier for the selected images of dermatologic manifestations of cardiovascular disease. A total of 2665 images consisting of 430 for Cyanosis, 480 for Liverdo reticularis, 780 for Xanthoma, 430 for Stasis dermatitis, 540 for fingernails clubbing, and other 1100 images of both normal skin and general objects were used in the training of the model. Data augmentation was also used to increase the amount of training images and finetuning was employed on the model. Google Collaboratory was used as the platform to train the model. The trained model with fine-tuning was able to obtain a considerable accuracy of 96.04%. The EfficientNetV2 convolutional neural network (CNN) model performed exceptionally well in the image classification. | ||
| Keywords | ||
| deep neural network; images; model; skin disease; transfer learning | ||
|
Statistics Article View: 250 PDF Download: 237 |
||
