Parameterized Quantum Circuits (PQCs) in Medical Analysis PDF 5/2/23.
Parameterized Quantum Circuits (PQCs) run on quantum simulators are being utilized in literature to improve medical image dataset analysis. PQCs support larger classical convolutional neural networks, with the hybrid model referred to as a quantum convolutional neural network, or quanvolutional neural network (QNN). Image data is typically trained classically, and then passes through the quantum circuit for quantum benefits.
Authors have reported up to a 3-5% boost in image classification testing accuracies. Both the PQCs running on cpu or gpu based hardware, and methods are expected to improve. Small increases in number of qubits are expected, due to exponential classical RAM requirements. The quantum simulator with PQC and classical network method will likely be used for some time regarding patient related data, as the approach is noise free. Real quantum computers do not yet have significantly more qubits but are faster than simulators, both are considerations for further implementation.
The Mathur, N., et al. paper Fig. 16 describes the logarithmic time advantage expected as quantum computers emerge. Ion-trap hardware may be the first candidate for models featuring a greater dependency on quantum mechanics. The medical field will likely weigh in on which applications will be appropriate for patient data, given real quantum computer error.
Paper links are accessible through the downloadable PDF. Several of the works featured in the attachment, as well as additional quantum advantages for image analysis are available through "Neuroradiology Discussions" on the company website. Efforts are being made to construct a beneficial classification model that features a PQC and Neuroimage dataset. PQCs run on classical hardware are also gaining considerable attention for quantum cybersecurity defense.