An Automatic Clinical System for Tumor Medical Image Classification Using Machine Learning Techniques

This study introduces an enhanced machine learning framework, referred to as the Robotic Clinical System for Brain MRI Image Classification (RCS-BMIC), and developed to enable automated detection of brain tumors from MRI scans. The architecture of the proposed system is organized into two fundamental stages: model training and performance evaluation. In the training stage, previously acquired brain MRI scans are compiled from clinical data sources to form a structured dataset. These images are then subjected to preprocessing procedures aimed at improving visual clarity and normalizing image dimensions through optimized computational techniques. Feature extraction is subsequently conducted using a Convolutional Neural Network (CNN), which learns discriminative representations such as structural boundaries and texture patterns within the brain images. The extracted features are further organized using the K-means clustering algorithm to group similar patterns into distinct clusters based on shared characteristics. During the evaluation stage, unseen MRI images are analyzed using the trained model. The classification process is performed through the K-Nearest Neighbor (K-NN) algorithm, which determines the presence or absence of a tumor based on feature similarity. The integration of deep feature learning, clustering, and instance-based classification provides a comprehensive and reliable framework for accurate brain tumor identification.