Evaluation of QCNN-LSTM for Disability Forecasting in Multiple Sclerosis Using Sequential Multisequence MRI

Abstract: Introduction Quantum Convolutional Neural Network (QCNN)-Long Short-Term Memory (LSTM) models were studied to provide sequential relationships for each timepoint in MRIs of patients with Multiple Sclerosis (MS). In this pilot study, we compared three QCNN-LSTM models for binary classification of MS disability benchmarked against classical neural network architectures. Our hypothesis is that quantum models will provide competitive performance. Methods Matrix Product State (MPS), reverse Multistate Entanglement Renormalization Ansatz (MERA), and Tree-Tensor Network (TTN) circuits were paired with LSTM layer to process near-annual MRI data of patients diagnosed with MS. These were benchmarked against a Visual Geometry Group (VGG)-LSTM and a Video Vision Transformer (ViViT). Predicted logits were measured against ground truth labels of each patient's Extended Disability Severity Score (EDSS) using binary cross-entropy loss. Training/validation/holdout testing was partitioned using 5-fold cross validation with a total split of 60:20:20. Levene's test of variance was used to measure statistical difference and Student's t-test for paired model differences in mean. Results The MPS-LSTM, reverse MERA-LSTM, and TTN-LSTM had holdout testing ROC-AUC of 0.70, 0.77, and 0.81, respectively (p-value 0.915). VGG16-LSTM and ViViT performed similarly with ROC-AUC of 0.73 and 0.77, respectively (p-value 0.631). Overall variance and mean were not statistically significant (p-value 0.713), however, time to train was significantly faster for the QCNN-LSTMs (39.4 sec per fold vs. 224 and 218, respectively, p-value <0.001). Conclusion QCNN-LSTM models perform competitively to their classical counterparts with greater efficiency in train time. Clinically, these can add value in terms of efficiency to time-dependent deep learning prediction of disease progression based upon medical imaging.

• The paper introduces a QCNN-LSTM approach that forecasts disability in MS using sequential MRI data.
• It integrates three quantum architectures (MPS, Reverse MERA, and TTN) with LSTM, achieving ROC-AUC values up to 0.81.
• It demonstrates significantly reduced training time compared to classical models, promising faster clinical decision-making.

Introduction

The interplay between quantum computing and machine learning is an area ripe for innovative breakthroughs, particularly in the medical imaging domain. A novel approach utilizing Quantum Convolutional Neural Networks (QCNNs) combined with Long Short-Term Memory (LSTM) networks has been introduced to aid in the disability forecasting for Multiple Sclerosis (MS) from sequential Multisequence MRI datasets. Leveraging quantum mechanics' unique computational properties, the QCNN-LSTM model is a fresh contender in the field of deep learning for medical prognosis, with the potential for competitive performance against existing classical neural network architectures.

Methodology

In the study, three QCNN architectures were employed: the Matrix Product State (MPS), Reverse Multistate Entanglement Renormalization Ansatz (MERA), and Tree-Tensor Network (TTN), each integrated with an LSTM layer. These quantum-inspired models were assessed against two robust classical neural networks — the Visual Geometry Group (VGG16)-LSTM and the Video Vision Transformer (ViViT). The models underwent a rigorous evaluation process, employing 5-fold cross-validation, and performance comparison using ROC-AUC metrics. An important outcome measured was training efficiency, a vital factor when comparing quantum and classical models.

Results and Analysis

In the quantum field, the MPS-LSTM, Reverse MERA-LSTM, and TTN-LSTM demonstrated promising ROC-AUC values of 0.70, 0.77, and 0.81, respectively. Interestingly, these models achieved a significant reduction in training time compared to their classical counterparts, showcasing the inherent speed advantage that quantum methods might possess. This advantage is critical as it suggests that QCNN-LSTM models can offer not just competitive accuracy but also greater efficiency in model training.

Conclusion

The implications of this research are far-reaching for clinical practice in the diagnosis and monitoring of MS. The QCNN-LSTM models have demonstrated their worth by offering performance on par with classical models while also promising reduced computation times. Such gains in efficiency could translate into faster clinical decision-making and improved patient management strategies. However, as with any groundbreaking work, further validation on geographically and demographically diverse datasets, as well as real-world quantum hardware, is essential. If these models can maintain their efficacy and efficiency outside controlled test environments, they could become critical tools in predictive healthcare and the broader adoption of quantum-informed machine learning models across various domains.