Quantum Fruit Fly algorithm and ResNet50-VGG16 for medical diagnosis

Medical data are present in large amount and this is difficult to process for the diagnosis and Healthcare organization requires effective technique to handle big data. Existing techniques in medical diagnosis have limitations of imbalance data and overfitting problem. This research applies Quantum Fruit Fly Algorithm (QFFA) technique for feature selection to improve the effectiveness of classification in medical diagnosis. The Min-Max Normalization technique is applied to normalize the images to reduce the difference in pixel values and enhance the images. The ResNet50 and VGG16 deep learning models were applied for the feature extraction. The QFFA technique applies Archimedes spiral to increases the exploitation of the model to select unique features for classification. The Archimedes spiral provides spiral search in the top solutions of the Fruit Fly algorithm that helps to overcome local optima trap and increases exploitation. The QFFA technique selected features were applied to SVM model for the effective classification of medical diseases. The QFFA unique feature selection helps to overcome imbalance and overfitting problem in classification. The QFFA model has achieved better results in terms of various performance metrics such as sensitivity (99.26 %), and accuracy (99.04%) than existing Deep 1D-CNN and GA-Decision tree models.(c) 2023 Elsevier B.V. All rights reserved.

Automated summary

Medical data are difficult to process due to their large amount, and effective techniques are needed for healthcare organizations to handle big data. Existing techniques in medical diagnosis have limitations such as imbalanced data and overfitting. This research uses the Quantum Fruit Fly Algorithm (QFFA) technique for feature selection to improve the effectiveness of classification in medical diagnosis. The QFFA model achieves better results in terms of sensitivity (99.26%) and accuracy (99.04%) compared to existing Deep 1D-CNN and GA-Decision Tree models.