International Research Journal of Education and Technology

Adrenal tumor detection is critical for early diagnosis and effective treatment, but distinguishing between benign and malignant tumors in medical imaging is challenging. This project proposes a deep learning-based approach integrating CNNs, transfer learning, and ensemble learning to improve detection accuracy. The process involves preprocessing medical images (resizing, normalization, and contrast enhancement), segmentation using a U-Net model, and feature extraction with pre-trained models like VGG16, VGG19, ResNet-50, AlexNet, and Inception V3. Extracted features are classified using machine learning algorithms, including SVM, KNN, XGBoost, Decision Tree, Random Forest, AdaBoost, and Logistic Regression. Performance is evaluated using accuracy, error rate, F1 score, Jaccard index, and G-Mean. Results show that combining deep learning and machine learning enhances tumor detection. This approach can assist clinicians in making timely and precise diagnoses, improving patient outcomes. Future research may explore hybrid models and real-time clinical deployment. 

Key Words: Segmentation, transfer learning, feature extraction , deep learning, machine  learning, adrenal tumor, clinical deployment,  Logistic Regression.