Achieving reliable radiological image classification requires not only high accuracy, but also well-calibrated uncertainty estimates, robustness in the face of domain shifts, and generalizability across various imaging modalities. Current deep learning methodologies for medical image classification frequently display overconfident predictions and experience significant degradation under realistic clinical conditions, such as label noise, limited training data, and distribution shifts between institutions. Hybrid quantum–classical machine learning has emerged as a promising paradigm; however, existing studies in medical imaging predominantly focus on single-modality and accuracy-centric approaches. In this study, we introduced Q-CaMIR (Cross-Modal, Uncertainty-Aware Hybrid Quantum–Classical Learning), a novel framework that integrates chest X-ray and brain MRI classification within a unified hybrid quantum–classical architecture, incorporating explicit uncertainty quantification. Q-CaMIR utilizes a pretrained classical vision backbone to extract modality-agnostic latent representations, aligns cross-modal features through a contrastive representation module, and channels the resulting compact embeddings through a variational quantum circuit (VQC) decision head, enhanced with evidential deep learning for uncertainty-aware prediction. We evaluated Q-CaMIR on three public benchmarks: CheXpert and MIMIC-CXR (external validation) for chest radiograph classification, and BraTS 2021 for brain tumor classification, employing a comprehensive suite of metrics including accuracy, macro-F1, AUROC, AUPRC, expected calibration error (ECE), and Brier score. Extensive experiments under low-data regimes, synthetic label noise, intensity corruption, and cross-institutional domain shifts demonstrate that Q-CaMIR achieves competitive classification performance compared to robust classical baselines while significantly improving calibration (up to 47% ECE reduction) and robustness. Ablation studies confirm the complementary contributions of the quantum decision head and uncertainty-aware training objective. Our findings suggest that hybrid quantum–classical learning provides a meaningful inductive bias for uncertainty-sensitive radiological classification, advancing towards clinically deployable AI systems for radiation-related diagnostic workflows.
Achieving reliable radiological image classification requires not only high accuracy, but also well-calibrated uncertainty estimates, robustness in the face of domain shifts, and generalizability across various imaging modalities. Current deep learning methodologies for medical image classification frequently display overconfident predictions and experience significant degradation under realistic clinical conditions, such as label noise, limited training data, and distribution shifts between institutions. Hybrid quantum–classical machine learning has emerged as a promising paradigm; however, existing studies in medical imaging predominantly focus on single-modality and accuracy-centric approaches. In this study, we introduced Q-CaMIR (Cross-Modal, Uncertainty-Aware Hybrid Quantum–Classical Learning), a novel framework that integrates chest X-ray and brain MRI classification within a unified hybrid quantum–classical architecture, incorporating explicit uncertainty quantification. Q-CaMIR utilizes a pretrained classical vision backbone to extract modality-agnostic latent representations, aligns cross-modal features through a contrastive representation module, and channels the resulting compact embeddings through a variational quantum circuit (VQC) decision head, enhanced with evidential deep learning for uncertainty-aware prediction. We evaluated Q-CaMIR on three public benchmarks: CheXpert and MIMIC-CXR (external validation) for chest radiograph classification, and BraTS 2021 for brain tumor classification, employing a comprehensive suite of metrics including accuracy, macro-F1, AUROC, AUPRC, expected calibration error (ECE), and Brier score. Extensive experiments under low-data regimes, synthetic label noise, intensity corruption, and cross-institutional domain shifts demonstrate that Q-CaMIR achieves competitive classification performance compared to robust classical baselines while significantly improving calibration (up to 47% ECE reduction) and robustness. Ablation studies confirm the complementary contributions of the quantum decision head and uncertainty-aware training objective. Our findings suggest that hybrid quantum–classical learning provides a meaningful inductive bias for uncertainty-sensitive radiological classification, advancing towards clinically deployable AI systems for radiation-related diagnostic workflows.
Accurate classification of glioma grades from magnetic resonance imaging (MRI) is essential for clinical decision-making in neuro-oncology. Although deep learning performance has been impressive with classical models, they struggle with high-dimensional medical imaging data and generalise poorly beyond their training data, especially in time- and resource-constrained settings. In light of the aforementioned challenges, we propose QuantumMedDx, a hybrid quantum–classical learning framework for classifying gliomas using MRI. The framework combines quantum feature encoding and variational quantum circuits with classical neural inference to improve diagnostic performance. The base model, QImageNet, uses amplitude-based quantum encoding for writing, entanglement-enabled parameterised quantum circuits (EPQCs) as feature extractors, and classical dense layers for classifying HGG and LGG from multimodal MRI slices. We demonstrate the effectiveness of the proposed approach on the BraTS 2021 benchmark dataset using a patient-aware 5-fold cross-validation protocol. Experimental results show that QuantumMedDx achieves accuracies of 94.12%, 93.30%, and 96.42%; F1-scores of 93.30% and 96.42%; and AUCs of 96.42% and 96.42%, respectively, outperforming conventional CNN, DNN, and SVM baselines. Ablation studies provide additional evidence of the performance improvements enabled by quantum Fourier transform and entanglement layers. Such results suggest that quantum–classical learning can efficiently improve feature extraction and discrimination in medical imaging, thus providing a modular and scalable route towards quantum–inspired clinical decision–support systems of the future.
Although automated diagnosis in medical imaging has made tremendous breakthroughs with deep learning, current models also exhibit difficulties with representation stability, calibration, and robustness to domain heterogeneity. Traditional methods that use deeper CNN or transformer networks incur greater computational costs without fully addressing feature separability and uncertainty estimation. To address these shortcomings, the present study proposes a hybrid quantum-classical approach, VQMedNet, that integrates variational quantum circuits with classical feature extraction for supervised classification of MRI and CT images. The suggested QFeatEmbed-VQC module is an NISQ-conscious nonlinear quantum feature embedding that can be trained on tokenised and compressed image features, enabling effective learning with constrained quantum resources. A gradual training approach is followed to achieve consistent optimisation, and calibration-sensitive assessment is included to measure the predictability. The experiments with the brain tumour MRI (BraTS) and lung CT (LIDC-IDRI) datasets show a stable increase over the classical baseline, achieving 95.6% accuracy and 0.947 Macro-F1 on MRI classification, and 93.1% accuracy with improved AUC-PR on CT tasks. Additional analyses show enhanced embedding separability, robustness to domain perturbations, and reduced Expected Calibration Error. Moreover, the findings suggest that moderate-depth quantum circuits offer a good balance between performance and computational complexity. Although tested under simulation-based conditions, the proposed framework can indicate the potential of quantum-enhanced feature embedding for building stable and efficient medical AI systems. It may be seen as a promising approach to future NISQ-aware diagnostics. The implementation is publicly available at: https://github.com/gvvsrk/QuantumMed-FedDx . Clinical trial number: Not applicable.
Accurate and computationally efficient classification of brain tumors from MRI scans is critical for deployment in resource-constrained clinical settings, where conventional deep CNN architectures-despite achieving strong accuracy-often require large parameter counts that limit scalability. This work proposes a hybrid quantum-classical deep learning framework that combines a classical convolutional feature extractor with a variational quantum circuit (VQC) for multi-class tumor classification. Spatial features extracted by the CNN backbone will be dimensionality-reduced and encoded into quantum states using angle encoding; a parameterized quantum circuit will then process these features in a higher-dimensional Hilbert space, leveraging superposition and entanglement to capture non-linear feature correlations. Measurement outputs from the quantum layer will be passed to a classical fully connected layer for classification into glioma, meningioma, pituitary tumor, and no-tumor categories. The proposed framework will be evaluated on a publicly available benchmark MRI dataset and benchmarked against state-of-the-art classical CNN architectures in terms of accuracy, precision, recall, F1-score, and parameter efficiency. This study aims to demonstrate that quantum-enhanced layers can achieve competitive classification performance while substantially reducing trainable parameters, offering a practical direction for efficient, quantum-assisted medical image analysis.
Abstract Classical deep learning models often struggle to fully capture complex visual patterns and to generalize effectively when trained on limited or imbalanced medical imaging data. To address this limitation, this paper presents a novel Hybrid Quantum Deep Learning (Hybrid QDL) framework that leverages the strengths of both deep neural networks and quantum circuits for advanced medical image interpretation. The proposed model integrates a pre-trained EfficientNet-B4 as the feature extractor and a variational quantum circuit as the classification head, enabling joint learning of hierarchical and quantum-enhanced representations. The Hybrid QDL model is evaluated on three clinically relevant datasets: COVIDx (for 3-class COVID-19 diagnosis using chest X-rays), PneumoniaMNIST (for pneumonia detection), and OrganAMNIST (for multi-organ abdominal classification). The model achieves state-of-the-art results, with 98.48% accuracy and 98.33% F1-score on COVIDx, outperforming both classical and prior quantum baselines. On PneumoniaMNIST and OrganAMNIST, it reaches 98.28% and 95.27% accuracy, respectively, along with high macro F1-scores. Ablation studies confirm that increasing quantum circuit depth and entanglement enhances discriminative power, particularly in complex multi-class scenarios. These results demonstrate that the Hybrid QDL framework is a robust, scalable solution for automated medical diagnosis and holds significant promise for real-world clinical decision support and COVID-19 detection.
Introduction: Medical image classification is a crucial task in cancer diagnosis, relying on the accurate analysis of high-dimensional imaging data. While Convolutional Neural Networks (CNNs) have shown great success in this domain, their performance is often limited by the shallow feature expressiveness and overfitting, particularly in small or heterogeneous datasets. Methods: Quantum machine learning offers new opportunities through high-dimensional representations and nonlinear transformations. In this work, we propose a Quantum-Enhanced Sandwich Convolutional Neural Network (QSCNN), a layered hybrid architecture that integrates quantum and classical modules. The model employs a quanvolutional layer for localized quantum feature extraction, followed by conventional convolution and pooling for hierarchical representation learning, and a variational quantum classifier for nonlinear decision-making. Results: QSCNN achieved higher accuracy and training stability than classical CNNs and QCCNN baselines across three medical imaging tasks.: brain tumor MRI, skin cancer dermoscopy, and lung cancer CT. Circuit depth analysis revealed a trade-off between expressiveness and robustness, and additional experiments with depolarizing noise confirmed the model’s resilience under realistic quantum error conditions. Discussion: This suggests that circuit design choices influence hybrid model behavior and generalization, supporting the feasibility of quantum-enhanced methods for small-sample medical imaging. However, the current evaluation is limited to relatively small benchmark datasets, and broader validation on large-scale data will be essential to confirm clinical applicability. Conclusion: In summary, QSCNN presents a feasible hybrid framework for enhancing medical image classification with quantum features. While preliminary, our results suggest potential advantages in accuracy and stability under NISQ conditions.
Furthermore, the physics and multimodality imaging characteristics of tumors, along with high variability in intensity, morphology and anatomical position of the tumor, make the diagnosis of tumors from magnetic resonance imaging (MRI) still quite difficult. In this study, we introduce a hybrid quantum-classical Quantum Convolutional Neural Network (QCNN) that applies Quantum processing in the localized stage of feature extraction rather than later in the classification stage when most current quantum neural networks (CNNs) are used, for automated tumor detection from multiparametric MRI data. The proposed framework makes use of the BraTS 2021 data set, comprising of T1-weighted, post-contrast T1-weighted (T1Gd), T2-weighted and T2-FLAIR images, along with patient-level partitioning, intensity normalization, tumor relevant slice selection, 4×4/8×8 patch generation, dimensionality reduction, quantum angle encoding, trainable variational quantum transformations and Pauli-Z expectation-value measurements. This produced quantum feature maps which were summed across classical layers, and compared to a standard CNN and transfer learning CNN and CNN with a quantum classifier. The proposed QCNN showed excellent performance with 96% accuracy, 93% sensitivity, 94% specificity, 94% precision, 94% F1-score, AUPRC of 0.94, AUROC of 0.96 and the MCC score is 0.82, which is the best among all the architectures considered. The use of architectural analysis revealed to be an option, since 4-6 qubits, 2-3 quantum layers, angle encoding and linear or ring entanglement prove to be adequate. The results of robustness analysis showed that the performance is degrades progressively as the level of quantum noise is increased but still reached a good AUROC (0.80) in the case of strong quantum noise, and was confirmed by an independent external test with a value of 0.85. The results here show trainable quantum convolution to be one potential approach to localized multimodal representation learning for MRI, and present an indication of the impact of quantum noise, finite shot execution, and dataset shift. The study provides a groundwork for the further expansion of research into the use of quantum-enhanced MRI for analysis, in the form of an experimentally controlled experiments design, in larger independent cohorts of patients and with real quantum infrastructure.