In a paper published in Scientific Reports, researchers proposed a hybrid quantum-classical 3D convolutional neural network (CNN) framework to improve blood flow velocity prediction using laser speckle contrast imaging (LCSI).
Study: Quantum machine learning enhanced laser speckle analysis for precise speed prediction. Image Credit: Aleksandra Budnik/Shutterstock.com
By integrating variational quantum circuits in place of traditional pooling layers, the model preserved 3D spatiotemporal features, enhancing prediction accuracy. Cross-validation on experimental data demonstrated improvements in prediction accuracy and error metrics compared to classical models. The results showed the hybrid model's superior learning stability and generalization capabilities for low and high blood flow velocities.
Quantum-Enhanced Blood Flow Prediction
This study explored a new approach to blood flow prediction using quantum-enhanced methods. A tissue phantom and a full-field laser speckle contrast imaging (LSCI) system were used to simulate and measure blood flow. The phantom replicated the light scattering properties of human tissue, while a servo-controlled scattering plate mimicked red blood cell motion at varying speeds.
The LSCI system employed a 532 nm green laser, a CMOS sensor, and an FPGA platform to capture and process speckle patterns generated by the simulated blood flow. Data augmentation was performed on the captured frames, increasing the training data for analysis.
At the core of this research was a quantum-classical hybrid neural network framework designed to improve predictive performance. Variational quantum algorithms (VQAs) replaced global pooling layers in traditional 3D convolutional neural networks (CNNs), preserving critical spatial and temporal information. The quantum layer leveraged amplitude encoding to efficiently handle high-dimensional feature vectors, reducing data loss compared to classical methods.
This framework was applied to CNN-LSCI and ResNet-LSCI models, resulting in hybrid versions: QCNN-LSCI and QResNet-LSCI. These quantum-enhanced models addressed the limitations of traditional 3D CNNs by preserving detailed spatial and temporal features. The incorporation of VQAs also boosted predictive accuracy, particularly for tasks involving complex spatial-temporal patterns.
Furthermore, the flexibility of the variational quantum circuits enabled the models to adapt to a wide range of computational tasks. This adaptability, combined with superior predictive accuracy and data efficiency, allowed QCNN-LSCI and QResNet-LSCI to outperform their classical counterparts.
By integrating quantum computing techniques, this research marked a significant step toward leveraging quantum advantages for practical applications. The proposed frameworks provided a robust platform for advancing medical imaging, dynamic tissue analysis, and other fields reliant on precise, data-driven predictions.
Performance Comparison of Models
The experiments were conducted using classical and quantum models, specifically CNN-LSCI, ResNet-LSCI, QCNN-LSCI, and QResNet-LSCI, to evaluate their ability to predict blood flow velocity. Quantum layers were implemented with PennyLane, while classical layers were built using PyTorch. PennyLane's default qubit simulator served as the backend for executing quantum circuits, with both backpropagation and adjoint differentiation techniques employed to compute quantum gradients in the variational quantum circuits (VQC).
Model performance was assessed using k-fold cross-validation, with the dataset divided into training, validation, and test sets. The dataset consisted of 1440 samples, with 922 allocated for training, 230 for validation, and 288 for testing. Models were trained for 100 epochs with a batch size of 16, using the Adam optimizer and a learning rate of 0.001.
Mean squared error (MSE) and mean absolute percentage error (MAPE) were used as evaluation metrics. Results from the 5-fold cross-validation demonstrated that the quantum models consistently outperformed their classical counterparts in both training and validation phases.
The quantum models, QCNN-LSCI and QResNet-LSCI, demonstrated lower and more stable losses compared to their classical counterparts. QCNN-LSCI achieved a 14.8 % improvement in test loss and a 26.1 % improvement in MAPE over CNN-LSCI, while QResNet-LSCI showed improvements of 5.7 % in test loss and 18.5 % in MAPE over ResNet-LSCI.
Quantum models demonstrated exceptional performance under specific conditions. At high blood flow velocities, where classical models tended to underestimate speeds, QCNN-LSCI and QResNet-LSCI delivered significantly more accurate predictions. Even with limited data, the quantum models exhibited superior learning and generalization capabilities. Furthermore, they excelled at detecting subtle speckle patterns in low-speed blood flow, effectively overcoming noise and capturing fine details—areas where classical models struggled to achieve reliable predictions.
Conclusion
In summary, this study introduced a hybrid quantum-classical 3D CNN framework for predicting phantom velocity, effectively improving prediction accuracy by mitigating information loss from global pooling. The quantum models exhibited superior performance in terms of prediction accuracy, learning stability, and generalization, particularly under challenging conditions such as extreme blood flow velocities.
Despite computational constraints, the framework demonstrated significant potential and offers a foundation for extending its application to other LSCI tasks and beyond. Future research will focus on validating the framework with in vivo experiments, expanding the dataset, and exploring alternative quantum circuit architectures to further enhance performance.
Journal Reference
Chen, Y., et al. (2024). Quantum machine learning enhanced laser speckle analysis for precise speed prediction. Scientific Reports, 14:1, 1-17. DOI: 10.1038/s41598-024-78884-4, https://www.nature.com/articles/s41598-024-78884-4
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.