EUVIMEDEUVIMED entdecken
Crossref

A Lightweight Dual–Path Convolutional Neural Network for Accurate Classification of Kidney Diseases in CT Scans

Sekenov Zhalgas, Zubaer Ibna Mannan, Nur Alam MD

BioMed and BioSci Advances · 2025

DOI: 10.69626/bba.2025.0236 ↗

Kurz gefasst

Fachartikel in der Zeitschrift BioMed and BioSci Advances, 2025. Verfasst von Sekenov Zhalgas und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Kidney diseases, including cysts, stones, tumors, and healthy tissue, present considerable diagnostic difficulties due to inherent class imbalances, dataset constraints, and the intricate anatomical variability observed in renal imaging. This study introduces a practical application of the Crossbar-Net, a computationally efficient dual-path convolutional neural network tailored for the classification of kidney conditions using CT images. To address dataset imbalance, a curated subset was created from an original collection of 12,500 images by sampling 1,377 examples per diagnostic category, yielding a balanced dataset of 5,508 images distributed across four classes: cyst, normal, stone, and tumor. The experimental pipeline incorporated systematic preprocessing and augmentation techniques such as resizing, normalization, random rotation, and horizontal flipping to enhance model robustness. Training was performed over 10 epochs using an NVIDIA A100 GPU within the Google Colab environment. The proposed model attained a final test accuracy of 99.47%, substantially surpassing established architectures such as ResNet, VGG, U-Net, and Inception in both predictive performance and operational efficiency. Detailed descriptions of the data preparation, model configuration, training protocols, and evaluation metrics are provided to ensure reproducibility and to underscore the framework’s potential for real-world clinical implementation.

Zitieren

APA

Zhalgas, S., Mannan, Z. I., & MD, N. A. (2025). A Lightweight Dual–Path Convolutional Neural Network for Accurate Classification of Kidney Diseases in CT Scans. BioMed and BioSci Advances. https://doi.org/10.69626/bba.2025.0236

Vancouver

Zhalgas S, Mannan ZI, MD NA. A Lightweight Dual–Path Convolutional Neural Network for Accurate Classification of Kidney Diseases in CT Scans. BioMed and BioSci Advances. 2025. doi:10.69626/bba.2025.0236

BibTeX
@article{Zhalgas2025,
  author = {Zhalgas, Sekenov and Mannan, Zubaer Ibna and MD, Nur Alam},
  title = {A Lightweight Dual–Path Convolutional Neural Network for Accurate Classification of Kidney Diseases in CT Scans},
  journal = {BioMed and BioSci Advances},
  year = {2025},
  doi = {10.69626/bba.2025.0236}
}
Kostenlos starten →