Optimised hybrid deep learning classification model for kidney stone diagnosis
The kidney plays a vital role in maintaining homeostasis within the human body. In recent years, the prevalence of nephrolithiasis (kidney stone formation) characterized by the accumulation of crystalline solids within the renal system has emerged as a significant health concern. Early detection is...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-06-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025012939 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The kidney plays a vital role in maintaining homeostasis within the human body. In recent years, the prevalence of nephrolithiasis (kidney stone formation) characterized by the accumulation of crystalline solids within the renal system has emerged as a significant health concern. Early detection is critical for effective treatment and prevention of complications. Diagnostic imaging techniques such as computed tomography (CT), ultrasonography, and Doppler imaging are routinely employed for this purpose. To enhance the precision and reliability of early diagnosis, Deep Learning (DL) models are increasingly being integrated into the diagnostic workflow, offering superior accuracy through advanced image analysis and pattern recognition capabilities. The proposed work combines two deep learning models, AlexNet and Gated Recurrent Unit (GRU) for feature extraction and classification. These models are integrated to deliver optimal training parameter performance. An optimized AlexNet-GRU model is introduced in this work for detection of kidney stone, feature extraction, and classification. The Elephant Herding Optimizer (EHO) is utilized to fine-tune the hyperparameters of the AlexNet-GRU model. by performing this EHO fine tuning, the performance metrics of the proposed work have provided a high optimal result. Finally, the proposed evaluation metrics like precision, recall, accuracy, and F1 score are evaluated and compared with the traditional models to prove their efficient performances. The proposed model achieved a precision of 98.67 %, a recall of 97.68 %, an accuracy of 98.82 %, and an F1 score of 97.54 %. |
|---|---|
| ISSN: | 2590-1230 |