Leveraging Orange for Efficient Implementation of VGG16/VGG19 for Neoplasticity Classification

This project investigates the classification of neoplasticity in brain images using VGG16 and VGG19 models in conjunction with Orange Data Mining. The dataset, sourced from the Whole Brain Atlas, includes PET images of gliomas, meningiomas, and metastatic tumors. Both models were evaluated with various classifiers, achieving high performance in distinguishing tumor types. VGG19 with the polynomial kernel attained the best AUC (0.986) and F1-score (0.895), while VGG16 with the polynomial kernel achieved the highest classification accuracy (0.913). The results highlight the potential of deep learning models in brain tumor classification and demonstrate the utility of Orange for model evaluation.