Leveraging Quantum Computing to Revolutionize Deep Learning: A Focus on Hybrid Algorithms for Medical Image Classification

Abstract

This thesis investigates the amalgamation of quantum computing and deep learning to mitigate computational constraints in image classification tasks. As deep learning models get more complicated, classical computing methods have a lot of problems with resources, energy use, and scalability. Quantum computing, with its built-in parallelism and ability to solve problems in more than one dimension, could help get around these problems. This study concentrates on the advancement of hybrid quantum-classical algorithms capable of functioning efficiently within the limitations of contemporary Noisy Intermediate-Scale Quantum (NISQ) devices while improving the precision and efficacy of image classification tasks. The goal of the work is to close the gap between the theoretical benefits of quantum computing and the real-world problems that come up when using it in machine learning. Quantum Learning (QL) has developed as a promising method for classifying medical images by using quantum mechanics to make machine learning algorithms work better and faster. This systematic review offers an extensive critical evaluation of the present state of QL techniques formulated for medical image classification, emphasising trends, methodologies, and prospective developments in this swiftly advancing domain. A comprehensive literature search was performed across five principal databases, yielding a total of 28 pertinent studies published between 2018 and 2024. The studies were examined and categorised according to the type of quantum algorithm, the medical imaging modality, and the performance metrics employed. The analysis identified quantum learning (QL) techniques, such as Quantum Support Vector Machines (QSVM), Quantum Convolutional Neural Networks (QCNN), and several hybrid quantum-classical methodologies. These methods have been utilised for various medical image classification tasks, including brain tumour classification, skin lesion classification, and COVID-19 detection, yielding encouraging outcomes regarding accuracy, sensitivity, and specificity. Nonetheless, various challenges were recognised, such as the preprocessing and encoding of medical images for quantum processing, the restricted scalability of existing quantum hardware, and the necessity for interpretable and explicable quantum learning models. This review highlights the significant potential of QL to transform medical image classification while also stressing the importance of interdisciplinary collaborations and additional research to address current challenges and promote the incorporation of QL techniques into clinical practice.