Fuzzy Enhanced Kidney Tumor Detection using Transferable Networks and Ensemble Learning

Dr.S.Suryakumari 1,G.Gnana Prakash 2,Kalimuthu Murthy 3,M.Mythri 4,B.Dilli kumar 5
1Assistant Professor, Dept of Information Technology, SV College of Engineering, Tirupati, India.
2B. Tech, Dept of Information Technology, SV college of Engineering, Tirupati, India.
3B. Tech, Dept of Information Technology, SV college of Engineering, Tirupati, India.
4B. Tech, Dept of Information Technology, SV college of Engineering, Tirupati, India.
5B. Tech, Dept of Information Technology, SV college of Engineering, Tirupati, India.
Email: 1suryakumari.s@svce.edu.in, 2gnanaprakashgudimetla@gmail.com,
3kalimuthumurthy19@gmail.com,
4mythrireddy345@gmail.com, 5dilliburra399@gmail.com

Abstract-Kidney tumors are among several factors that can affect patient survival rates if detected early; however, current automated approaches for detecting renal masses on computed tomography (CT) suffer from non-generalizability across datasets, potential unethicaluse and adverse clinical effects, lack of spatial localization, no interpretability by clinicians, and limited broader medical applicability. The existing STREAMLINERS system integrates fuzzy logic-based image enhancement, twin transferable deep neural networks (DenseNet121 and ResNet101), and aweighted ensemble machine learning classifier (SVM and Random Forest) for kidney tumor detection in CT scans, achieving high accuracy and robustness withMLOps for scalability. However, it faces limitations  such as limited dataset size andgeneralizability, lack of spatial tumor localization, sensitivity to data imbalance and noise, and limited explainability. The proposedsystem addresses these by incorporating advanced image segmentation (e.g., U-Net or Mask R-CNN) for precise localization, training on a larger, diverse dataset,integrating explainable AI (e.g., Grad-CAM or SHAP)  for transparent predictions, employing advanced dataaugmentation and class-balancing techniques for robustness, and exploring multi-modal fusion (e.g., CT with MRI or ultrasound) for comprehensive assessment.These enhancements improve localization,generalizability, explainability, handling of imbalanced data, and diagnostic accuracy, ultimately supportingbetter clinical decision-making and patient outcomes Keywords: computed tomography, twin transferabledeep neural networks, multi-modal fusion, ensemblemachine learning, Kidney tumors