Related Experiment Video
Updated: Jan 15, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
EfficientNet-B3-Based Automated Deep Learning Framework for Multiclass Endoscopic Bladder Tissue Classification
A A Abd El-Aziz1, Mahmood A Mahmood1, Sameh Abd El-Ghany1
1Department of Information Systems, College of Computer and Information Sciences, Jouf University, Sakaka 72388, Saudi Arabia.
Diagnostics (Basel, Switzerland)
|October 16, 2025
Summary
This study introduces an automated deep learning system for bladder cancer detection. The EfficientNet-B3 model accurately classifies smooth muscle images, aiding early diagnosis and reducing healthcare costs.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Bladder cancer (BLCA) diagnosis is challenging due to tumor heterogeneity and complex histopathology.
- Manual classification of tissues is time-consuming, prone to errors, and lacks standardization.
- There is a critical need for automated, reliable systems for efficient BLCA detection.
Purpose of the Study:
- To develop and validate a deep learning model for automated early detection of bladder cancer.
- To improve the accuracy and efficiency of BLCA diagnosis compared to traditional methods.
- To reduce diagnostic time and costs for patients undergoing bladder cancer screening.
Main Methods:
- A deep learning approach using the EfficientNet-B3 model was employed for multiclass classification.
- The Endoscopic Bladder Tissue Classification (EBTC) dataset was utilized, with preprocessing including resizing and normalization.
- Five-fold cross-validation and ablation studies were performed to optimize hyperparameters and validate performance against other leading DL models.
Main Results:
- The EfficientNet-B3 model achieved high performance metrics: 99.03% accuracy, 99.30% specificity, 97.95% precision, 96.85% recall, and 97.36% F1-score.
- The proposed model significantly outperformed five other leading deep learning models in classifying bladder tissue images.
- The system demonstrated efficient and accurate identification of bladder cancer tissues.
Conclusions:
- The EfficientNet-B3 model shows significant potential as a valuable tool for accurate and efficient bladder cancer diagnosis.
- Automated classification using this DL model can streamline the diagnostic process, leading to timely interventions.
- This technology offers a promising solution to reduce the morbidity and mortality associated with bladder cancer.
