Toward Quantum-Enabled Medical Imaging: Optimized CNNs for Diagnosis

The recent accelerated development of quantum computing has offered new avenues of re-conceptualization of classical architecture-constrained computational models. Quantum Convolutional Neural Networks (QCNNs) can also be leveraged on the principles of quantum mechanics to provide a fundamentally improved framework on the analysis of medical images. In contrast to classical CNNs, QCNNs take advantage of quantum parallelism and entanglement to find discriminative features more effectively and allows analyzing high-dimensional image patterns simultaneously and does not highly rely on massive annotated datasets. This quantum benefit does not only improve the feature mapping quicker rate but also improves the classification performance in a particular situation of scarce or nonhomogeneous medical information. More so, QCNN models are also naturally resilient to noise, and can maintain a high diagnostic accuracy under the conditions of fluctuation in image quality, which is a critical attribute in clinical settings. Experimental assessment on dataset of chest X-rays with labels of samples of Types- Covid-19, Pneumonia and Normal results in significant gains in speed of inference, discrimination of features and generalization. The results reveal the potential of quantum-driven imaging designs in disrupting the diagnostic platform; which refined QCNN designs could lead to significant improvements in the diagnostic platforms to the extent of improving the reliability of the decision and clinical impact.