[2605.23324] Enhancing Blood Cells Classification utilizing Hybrid Quantum Neural Networks

View a PDF of the paper titled Enhancing Blood Cells Classification utilizing Hybrid Quantum Neural Networks, by Guilherme Cruz and 4 different authors

View PDF

Summary:Correct classification of microscopic blood cells remains to be a crucial job in medical picture evaluation, the place refined variations and restricted information can problem typical deep studying fashions. As such, we examine on this work the potential of Hybrid Quantum-Classical Neural Networks (HQNNs) to reinforce function illustration and enhance classification efficiency on this area. We suggest a modular structure combining a pre-trained ResNet-50 spine with a low-dimensional latent bottleneck and a variational quantum circuit, enabling a direct comparability between quantum-enhanced and purely classical transformation mechanisms. To isolate the contribution of the quantum part, we consider three architectures: a HQNN mannequin, a Classical Matched Mannequin with a further nonlinear transformation layer of comparable capability, and a baseline mannequin with out an intermediate transformation stage. Experiments carried out on two publicly out there blood cell datasets, specifically the Blood Cell Photos dataset and the PBC dataset, exhibit that HQNNs persistently obtain superior or extra balanced efficiency throughout analysis metrics. Within the Blood Cell Photos Dataset, the proposed strategy improves macro F1-score by as much as 3.7% in comparison with classical baselines, whereas enhancing the F1-score from 98.54% to 98.69% within the tougher 8-class situation with near-saturated efficiency. Extra analysis on IBM quantum {hardware} reveals that the mannequin stays strong below noise, with solely a modest efficiency degradation relative to simulated outcomes. These outcomes point out that quantum function transformations can improve discriminative representations, significantly in difficult classification eventualities, and spotlight the sensible potential of HQNN fashions for medical imaging duties.