International Conference on Methods and Models in Automation and Robotics
Quantum-Enhanced Fine-Grained Hematologic Cell Classification: A Hybrid ResNet18 Architecture with Adjoint-Differentiable Variational Quantum Heads
Muktadir Hasan Rakebb, Raian Bin Dhali Mujahid, Mohammad Atiqul Islam, Souvik Pramanik, Ahmed Faizul Haque Dhrubo, Mohammad Abdul Qayum, Mohsin Sajjad, Mohammad Ashrafuzzaman Khan
Automated identification of blood cancer cells via morphological properties is extremely important for diagnosis; however, it poses significant challenges owing to subtle morphological variations, variations in staining techniques, and unbalanced class distributions. In this work, we propose an innovative pipeline combining classical deep learning architectures with quantum models to perform 16-way cell classification using angle embedding and entangled quantum variational layers trained with PennyLane. In addition to training various classical models for benchmarking purposes, ResNet18 was chosen as our backbone. The proposed hybrid approach demonstrates a performance improvement compared to the classical counterpart (ResNet18), achieving 85% test accuracy and a macro-F1 score of 0.83 compared to 79% accuracy and a macro-F1 of 0.77, respectively. The contribution of quantum variational circuits consists in better classification of precursors (MYB, LYA, ERY).
Automated identification of blood cancer cells via morphological properties is extremely important for diagnosis; however, it poses significant challenges owing to subtle morphological variations, variations in staining techniques, and unbalanced class distributions. In this work, we propose an innovative pipeline combining classical deep learning architectures with quantum models to perform 16-way cell classification using angle embedding and entangled quantum variational layers trained with PennyLane. In addition to training various classical models for benchmarking purposes, ResNet18 was chosen as our backbone. The proposed hybrid approach demonstrates a performance improvement compared to the classical counterpart (ResNet18), achieving 85% test accuracy and a macro-F1 score of 0.83 compared to 79% accuracy and a macro-F1 of 0.77, respectively. The contribution of quantum variational circuits consists in better classification of precursors (MYB, LYA, ERY).
Accurate classification of microscopic blood cells is still a critical task in medical image analysis, where subtle variations and limited data can challenge conventional deep learning models. As such, we investigate in this work the potential of Hybrid Quantum-Classical Neural Networks (HQNNs) to enhance feature representation and improve classification performance in this domain. We propose a modular architecture combining a pre-trained ResNet-50 backbone with a low-dimensional latent bottleneck and a variational quantum circuit, enabling a direct comparison between quantum-enhanced and purely classical transformation mechanisms. To isolate the contribution of the quantum component, we evaluate three architectures: a HQNN model, a Classical Matched Model with an additional nonlinear transformation layer of comparable capacity, and a baseline model without an intermediate transformation stage. Experiments conducted on two publicly available blood cell datasets, namely the Blood Cell Images dataset and the PBC dataset, demonstrate that HQNNs consistently achieve superior or more balanced performance across evaluation metrics. In the Blood Cell Images Dataset, the proposed approach improves macro F1-score by up to 3.7% compared to classical baselines, while improving the F1-score from 98.54% to 98.69% in the more challenging 8-class scenario with near-saturated performance. Additional evaluation on IBM quantum hardware shows that the model remains robust under noise, with only a modest performance degradation relative to simulated results. These results indicate that quantum feature transformations can enhance discriminative representations, particularly in challenging classification scenarios, and highlight the practical potential of HQNN models for medical imaging tasks.
Classifying malaria cell images relies on examining blood-smear microscopy slides to separate healthy red blood cells from those carrying the parasite to recognize the stage of infection. Manually examining the cells under a microscope is slow, as it depends heavily on expert skill, and can lead to inconsistent results, especially in resource-limited settings. Machine learning and deep learning help overcome these difficulties by automatically detecting subtle patterns that are hard for the human eye to distinguish, enabling faster, more accurate, and scalable diagnosis. This work compares the performance of classical deep learning models and hybrid quantum-classical systems for malaria cell classification problem, for a publicly available Kaggle dataset of 27,558 labelled images. The pipeline combines classical feature extraction from a subset of 5000 images with quantum feature mapping and variational quantum circuits (2, 3 and 4 qubits) through angle encoding and strongly entangling layers. All the models have been trained over 40 epochs with a 70-15-15 train-validation-test split, and accuracy, precision, recall, and F1-score have been evaluated. Classical CNN achieved 92.93% accuracy, while ResNet18 and MobileNetV2 achieved 87.87% and 60.67 % respectively.4-qubit Quantum CNN achieves the highest score of 93.47%, surpassing all the classical baselines. The hybrid MobileNetV2 achieved an accuracy of 87.07%, which is a significant improvement compared with its classical counterpart, indicating that quantum feature transformation can repurpose weak classical representations.
Accurate classification of blood cells is crucial for early diagnosis and monitoring of hematological disorders. However, manual microscopy is time-consuming, subjective, and prone to inter-observer variability. Deep learning has shown potential in automating this task, but its high computational demands limit its use in resource-constrained settings. This study aims to develop an efficient, hybrid quantum-classical model for blood cell classification using quantum simulation to reduce computational complexity without compromising accuracy. We propose a hybrid framework that integrates a 6-qubit variational quantum circuit (VQC) with a classical EfficientNet-B0 model. The model was trained and tested on the full BloodMNIST dataset, which includes 17,092 microscopic images across eight blood cell types. The proposed model achieved a test accuracy of 96.58% and a validation accuracy of 96.90%. F1-scores ranged from 91.97% to 99.84%, with notable results in neutrophils (98.80%) and lymphocytes (98.36%). These findings confirm the feasibility of combining quantum and classical approaches for medical image classification. The proposed hybrid model shows strong promise for practical use in diagnostic systems, especially in low-resource settings, and offers a pathway for future research in quantum-enhanced medical imaging.
With millions of medical data being produced daily, it is important to leverage current technology to improve the healthcare system. As traditional deep learning models are unable to effectively handle high-dimensional data, a hybrid classical-quantum approach is used to enhance traditional deep learning algorithms. The hybrid classical-quantum models are built upon the classical deep learning framework with quantum techniques. The classical layers are utilized for feature extraction while the quantum layer is used for computational tasks. This addresses computational limitations and data imbalance challenges in medical image analysis. Transfer learning models, such as ResNet18 and InceptionV3, are used in conjunction with a quantum variational circuit to perform the task. The model is trained on a chest X-ray dataset, which consists of two sets of data: binary and multiclass. The binary dataset consists of Normal and COVID-19 images, while the multiclass dataset has COVID-19, Normal, and Tuberculosis images for classification. PennyLane, as a simulator, has been used to perform quantum computing. The dataset has also been used to train the traditional transfer learning models, pre-trained ResNet18, and their results have been compared. Out of all the models tested, the hybrid classical-quantum model, namely Hybrid ResNet18, performed the best. It achieved an accuracy of 93.22% for the binary classification task and 93.91% for the multiclass classification. The results show that there is a statistically significant improvement in classification accuracy and F1-score of the hybrid approach over the classical transfer learning baselines under a simulation-based environment. This sets the stage for further research on real-world physical quantum hardware. Class imbalance and overlapping predictions have been addressed using methods such as data augmentation and confusion matrix analysis. Thus, quantum machine learning could be used as a vital tool to analyze medical data for better diagnostic results.
Classifying rice plant disease from leaf images presents a significant challenge due to the high visual similarity between different disease types, which can limit the effectiveness of purely classical deep learning models. This research addresses this problem by developing and evaluating a hybrid quantum neural network that integrates a pre-trained ResNet18 classical backbone with a quantum-enhanced classifier, implemented as a variational single-qubit circuit using the PennyLane framework. The model was trained on an augmented dataset and tested on original, non-augmented images to ensure a robust evaluation of its generalization capabilities. Experimental results demonstrate the superiority of the hybrid approach, which achieved an overall accuracy of 0.89, surpassing the 0.85 of the classical ResNet18 baseline, with notable performances on visually similar diseases, reaching an F1-score of 0.88 for rice stripes, 0.88 for rice blast, and 0.90 for rice tungro. These results lead to the conclusion that integrating even minimal quantum components can enhance classical deep learning models, enabling them to capture subtle feature representations. Future work will focus on exploring more complex quantum circuits with multi-qubit entanglement and deploying these models on Noisy Intermediate-Scale Quantum (NISQ) hardware to assess their real-world performance.
Background: Coronary angiography (CAG) is the cornerstone imaging modality for evaluating coronary artery stenosis and guiding interventional decision-making. However, interpretation based on single-frame angiographic images remains highly operator-dependent, and conventional deep learning models still face challenges in modeling complex vascular morphology and fine-grained texture patterns.Methods: We propose a Lightweight Quantum-Enhanced ResNet (LQER) for binary classification of coronary angiography images. A pretrained ResNet18 is employed as a classical feature extractor, while a parameterized quantum circuit (PQC) is introduced at the high-level semantic feature space for quantum feature enhancement. The quantum module utilizes data re-uploading and entanglement structures, followed by residual fusion with classical features, enabling end-to-end hybrid optimization with a strictly controlled number of qubits.Results: On an independent test set, the proposed LQER outperformed the classical ResNet18 baseline in accuracy, AUC, and F1-score, achieving a test accuracy exceeding 90%. The results demonstrate that lightweight quantum feature enhancement improves discrimination of positive lesions, particularly under class-imbalanced conditions.Conclusion: This study validates a practical hybrid quantum--classical learning paradigm for coronary angiography analysis, providing a feasible pathway for deploying quantum machine learning in medical imaging applications.