Characterization of IBM quantum computers for quantum machine learning applications on brain tumor datasets

Quantum computing is an emerging field with the potential to address challenges faced by classical computation, particularly in the management of large datasets. As the volume of data has increased, classical machine learning models, including large language models, face significant obstacles in training and performance. In contrast, quantum computers offer theoretical advantages in processing these large datasets comprised of metrics and vectors. However, to facilitate broader adoption, it is essential for quantum systems to demonstrate stability and reliability, akin to the consistency found in classical Turing machines. This paper evaluates quantum computers from the IBM Eagle family using a Quantum Support Vector Classifier (QSVC) model to classify brain tumors as malignant or benign. Using the Brats 2015 dataset available on Kaggle, the QSVC model achieves an impressive accuracy rate of 98.5% when analyzing two-dimensional data, a performance level consistently observed across the Eagle family of quantum computers. Moreover, our methodology involves creating sub-datasets of varying dimensions, contributing to the characterization of quantum hardware’s performance under increasing complexity. By focusing on the classification of brain tumors through Quantum Machine Learning (QML), this study aims to improve diagnostic accuracy and efficiency within the biomedical field, leveraging benefits over classical machine learning approaches. The findings confirm the hypothesis that the IBM Eagle backends are reliable and produce consistent results under different conditions. Additionally, the research demonstrates significant speedups in processing times for QML compared to classical methods, underscoring the potential of quantum computing in practical biomedical applications.