An Effective Bladder Cancer Diagnosis Strategy (BCDS) based on Deep learning and Artificial intelligence techniques | ||||
| Nile Journal of Communication and Computer Science | ||||
| Volume 9, Issue 1, June 2025 PDF (2.19 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/njccs.2025.374720.1048 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
Rana A El-atier  1; Ahmed Saleh2; asmaa hamdy2; M. Sabry Saraya3
| ||||
| 11 Computers and control Dept. Faculty of engineering Department, Faculty of engineering, Mansoura University, Mansoura 35516, Egypt | ||||
| 2Computers and control Dept. Faculty of engineering Department, Faculty of engineering, Mansoura University, Mansoura 35516, Egypt | ||||
| 3Computers and Control systems Dept. Faculty of Engineering, Mansoura University, Mansoura, Egypt | ||||
| Abstract | ||||
| Bladder cancer, one of the most common tumors of the urinary system, requires an early and correct diagnosis to optimize survival and treatment results. This research introduces the Bladder Cancer Diagnosis Strategy (BCDS), a two-phase methodology that uses AI-powered image processing to improve bladder cancer detection. In the Pre-Processing Phase (PP), picture data is balanced and enhanced to guarantee variety, followed by feature extraction using pre-trained deep learning models like GoogleNet, DenseNet, and AlexNet. These models are chosen based on their complementing architectural qualities, which ensure a balance of performance and computational economy. The retrieved features are refined using Leopard Seal Optimization (LSO) for feature selection and the Interquartile Range (IQR) approach for outlier removal. During the Bladder Cancer Diagnosis Phase (BDP), the K-Nearest Neighbors (KNN) classifier is optimized using Grid Search cross-validation, which investigates alternative combinations of hyper parameters (such as the number of neighbors and distance measure) to maximize classification accuracy. A comparison study using ResNet-50 was also done, which revealed that the suggested technique performed similarly or slightly better at a lower computational cost. The experimental findings show that the suggested BCDS has an accuracy of 97%, precision of 94%, recall of 95%, and an F1-score of 96%, indicating that it is effective for clinical use. Future research will focus on increasing dataset variety, implementing real-time diagnostic capabilities, and investigating advanced AI models to improve bladder cancer diagnosis. | ||||
| Keywords | ||||
| Artificial intelligence; Deep Learning; Bladder Cancer; Classification; Diagnosis | ||||
|
Statistics Article View: 94 PDF Download: 80 |
||||
