A Quantum CNN Approach for MRI Based Multi-Class Brain Tumor Classification
Rajyalakshmi Pulicherla, Gireesh Namineni
The evolution of quantum computing has paved way for several evolutions, protruding into medical diagnosis too. This paper focuses on Quantum Convolutional Neural Network is compared with existing methods like Support Vector Machine, Convolutional Neural Network and ResNet18 used for brain tumor classification. A detailed study of various types of brain tumor is provided along with possible diagnosis and treatment. The frameworks corresponding to SVM, CNN, ResNet18 and QCNN are provided. The metrics used to evaluate the methods are provided like accuracy, precision, recall, F1 score, confusion matrix and quantum related metrics like Quantum Fidelity, Quantum Cross Entropy Loss, etc. The QCNN is observed to improved accuracy by 10.12%, precision by 10.96%, recall by 3.8%, specificity by 5.4%, F1 score by 7.1%, MSE by 65.5% good results in all parameters varying from 0.1% to 10.96% when compared with SVM, CNN and ResNet18, for multiclass brain tumor classification of Glioma, Meningioma, Pituitary and No Tumor. The quantum fidelity is about 1.0 and the quantum cross entropy is about 0.054. Hence QCNN proves to be a better choice for brain tumor classification.
IEEE Madhya Pradesh Section Conference (MPCON)•Mar 14, 2026•0 citations