KLASIFIKASI PENYAKIT GINJAL MENGGUNAKAN ALGORITMA HIBRIDA CNN-ELM

Bik, Ahmad Hasby (2024) KLASIFIKASI PENYAKIT GINJAL MENGGUNAKAN ALGORITMA HIBRIDA CNN-ELM. Undergraduate thesis, UPN Veteran Jatim.

	Text (cover) 20081010032_COVER.pdf Download (987kB)
	Text (bab 1) 20081010032-BAB1.pdf Download (110kB)
	Text (bab 2) 20081010032-BAB2.pdf Restricted to Repository staff only until 30 May 2028. Download (1MB)
	Text (bab 3) 20081010032-BAB3.pdf Restricted to Repository staff only until 30 May 2028. Download (1MB)
	Text (bab 4) 20081010032-BAB4.pdf Restricted to Repository staff only until 30 May 2028. Download (2MB)
	Text (bab 5) 20081010032-BAB5.pdf Download (101kB)
	Text (daftar pustaka) 20081010032-DAFTAR PUSTAKA.pdf Download (125kB)

Abstract

Kidney disease is a serious health issue requiring early detection. This study explores a hybrid model combining Convolutional Neural Network (CNN) and Extreme Learning Machine (ELM) to classify CT images of kidney disease, highlighting the importance of activation function selection. This approach promises accurate diagnosis with high accuracy, supporting everyday clinical practice. By leveraging medical imaging technology through computerized tomography (CT), this research aims to improve the efficiency of kidney disease diagnosis through automatic image analysis using artificial intelligence. The goal of this research is to enhance the accuracy of kidney disease CT image classification by combining the strengths of CNN in feature extraction and ELM in fast classification. To address this issue, a series of experiments were conducted with variations in the number of filters in CNN and hidden neurons in ELM. The hybrid CNN-ELM model was tested using ReLU and Tanh activation functions to determine the optimal configuration that yields the highest accuracy. This hybrid approach is expected to overcome the limitations of CNNs, such as the need for large datasets and the risk of overfitting. Test results indicate that the hybrid CNN-ELM model with the ReLU activation function achieved the highest accuracy of 0.9963, while Tanh reached 0.8419. Model accuracy ranged from a low of 0.8419 with the Tanh activation function to a high of 0.9963 with ReLU. The performance evaluation of the highest model using the ROC curve also showed a perfect AUC value (1.00) for all classes (cyst, normal, stone, tumor), indicating high sensitivity and specificity. The conclusion of this study is that the hybrid CNN-ELM approach is effective in improving the accuracy and efficiency of kidney disease CT image classification, making it a valuable tool in medical practice to support better clinical decision-making.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Anggraeny, Fetty Tri	NIDN19820211 2021212 005	fettyanggraeny.if@upnjatim.ac.id
Thesis advisor	Puspaningrum, Eva Yulia	NIDN19890705 2021212 002	evapuspaningrum.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Depositing User:

Ahmad Hasby Bik Bik

Date Deposited:

03 Jun 2024 04:28

Last Modified:

03 Jun 2024 04:28

URI:

https://repository.upnjatim.ac.id/id/eprint/23582