Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hierarchical Quantum Control Gates for Functional MRI Understanding

Xuan-Bac Nguyen, Hoang-Quan Nguyen, Hugh Churchill, Samee U. Khan, Khoa Luu

TL;DR

This work tackles understanding extremely high-dimensional fMRI signals, where classical methods struggle with long sequences and voxel-level interactions. It proposes Hierarchical Quantum Control Gates (HQCG), consisting of Local Quantum Control Gates (LQCG) and Global Quantum Control Gates (GQCG), with amplitude encoding to process data on a quantum processor and a multi-class quantum state fidelity circuit for classification. Key contributions include the HQCG architecture, the amplitude-encoded data path, and empirical evidence of improved accuracy and reduced overfitting compared to classical baselines, especially when aggregating information across hemispheres. The results suggest that end-to-end quantum models can effectively learn local and global brain representations from long fMRI sequences and may extend to other domains with ultra-high-dimensional neural data.

Abstract

Quantum computing has emerged as a powerful tool for solving complex problems intractable for classical computers, particularly in popular fields such as cryptography, optimization, and neurocomputing. In this paper, we present a new quantum-based approach named the Hierarchical Quantum Control Gates (HQCG) method for efficient understanding of Functional Magnetic Resonance Imaging (fMRI) data. This approach includes two novel modules: the Local Quantum Control Gate (LQCG) and the Global Quantum Control Gate (GQCG), which are designed to extract local and global features of fMRI signals, respectively. Our method operates end-to-end on a quantum machine, leveraging quantum mechanics to learn patterns within extremely high-dimensional fMRI signals, such as 30,000 samples which is a challenge for classical computers. Empirical results demonstrate that our approach significantly outperforms classical methods. Additionally, we found that the proposed quantum model is more stable and less prone to overfitting than the classical methods.

Hierarchical Quantum Control Gates for Functional MRI Understanding

TL;DR

This work tackles understanding extremely high-dimensional fMRI signals, where classical methods struggle with long sequences and voxel-level interactions. It proposes Hierarchical Quantum Control Gates (HQCG), consisting of Local Quantum Control Gates (LQCG) and Global Quantum Control Gates (GQCG), with amplitude encoding to process data on a quantum processor and a multi-class quantum state fidelity circuit for classification. Key contributions include the HQCG architecture, the amplitude-encoded data path, and empirical evidence of improved accuracy and reduced overfitting compared to classical baselines, especially when aggregating information across hemispheres. The results suggest that end-to-end quantum models can effectively learn local and global brain representations from long fMRI sequences and may extend to other domains with ultra-high-dimensional neural data.

Abstract

Quantum computing has emerged as a powerful tool for solving complex problems intractable for classical computers, particularly in popular fields such as cryptography, optimization, and neurocomputing. In this paper, we present a new quantum-based approach named the Hierarchical Quantum Control Gates (HQCG) method for efficient understanding of Functional Magnetic Resonance Imaging (fMRI) data. This approach includes two novel modules: the Local Quantum Control Gate (LQCG) and the Global Quantum Control Gate (GQCG), which are designed to extract local and global features of fMRI signals, respectively. Our method operates end-to-end on a quantum machine, leveraging quantum mechanics to learn patterns within extremely high-dimensional fMRI signals, such as 30,000 samples which is a challenge for classical computers. Empirical results demonstrate that our approach significantly outperforms classical methods. Additionally, we found that the proposed quantum model is more stable and less prone to overfitting than the classical methods.
Paper Structure (18 sections, 1 equation, 6 figures, 1 table)

This paper contains 18 sections, 1 equation, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Brain Signal Encoding
  4. Quantum Basics
  5. Parameterized Quantum Circuit
  6. Proposed Approach
  7. Data Encoding
  8. Local Quantum Control Gate
  9. Global Quantum Control Gate
  10. Multi-Classification Quantum State Fidelity Circuit
  11. Dataset and Implementation Details
  12. Dataset
  13. Implementation Details
  14. Experiments and Results
  15. Multi-Objects Predictions From fMRI
  16. ...and 3 more sections

Figures (6)

  • Figure 1: An overall quantum system for the Functional MRI classification. The green box includes components that run on the quantum computer: Amplitude encoding, Local Quantum Control Gate (LQCG), Global Quantum Control Gate (GQCG), and Fidelity Circuit (FC).
  • Figure 2: Local Quantum Control Gate
  • Figure 3: Global Quantum Control Gate.
  • Figure 4: The training progresses of classical and quantum fMRI classification models using different brain hemispheres, i.e., left hemisphere, right hemisphere, and both.
  • Figure 5: The training progresses of classical and quantum fMRI classification models on different subjects.
  • ...and 1 more figures