MRI Reconstruction Artefacts: Simulation & Mitigation Strategies

This project was developed as part of the Medical and Surgical Imaging at Imperial College London. It focuses on simulating and analysing typical artefacts in MRI reconstruction — caused by limitations in k-space sampling and patient-induced motion—using T1-weighted high-resolution head-neck MRI data.

🧠 Project Objectives

• Simulate real-world artefacts arising from limited k-space acquisition, partial sampling, and physiological motion.
• Analyse the image degradation both qualitatively and quantitatively.
• Investigate mitigation strategies used in clinical practice, grounded in imaging physics and signal processing.
• Connect theoretical findings to clinical scenarios, improving practical understanding of image quality challenges.

🔍 Key Artefacts Analysed

🔹 1. Inadequate Sampling

• Cause: Sparse k-space lines in the phase-encoding direction (Δk too large).
• Effect: Aliasing and spatial mislocalisation, including wrap-around of structures outside the field of view.
• Mitigation:
  • Increase FOV or oversample in the phase-encoding direction.
  • Use RF pre-saturation bands or surface coils to suppress external signals.

🔹 2. Partial k-space Sampling

• Cause: Truncated high-frequency k-space lines.
• Effect: Gibbs ringing artefacts near sharp tissue boundaries.
• Mitigation:
  • Apply spatial smoothing filters (e.g., exponential or Gaussian).
  • Increase matrix size or decrease FOV to capture finer detail.
  • Use fat suppression if one edge is due to fatty tissue.

🔹 3. Patient Motion

• Periodic Motion: Creates discrete ghosting artefacts, especially aligned with TR cycles (e.g., from breathing or cardiac pulses).
• Non-Periodic Motion: Introduces global blurring and smear in the phase-encoding direction.
• Mitigation:
  • Use gated sequences or increase the number of signal averages (NSA).
  • Apply radial sampling or parallel imaging techniques.
  • Apply fast imaging sequences and instruct patients for stability.

💡 Technical Highlights

• Transformed anatomical MRI slices into k-space using 2D Fourier Transforms.
• Simulated degraded acquisitions by masking or altering k-space lines.
• Reconstructed images using inverse FFT to reveal artefact patterns.
• Designed and evaluated clinical mitigation methods using synthetic motion and under-sampling scenarios.

🏫 Project Info

• Module: Medical and Surgical Imaging
• Challenge: MRI Image Reconstruction Artefacts
• Institution: Imperial College London
• Supervisor: Prof Daniel Elson
• Student: Gavin Yue

This project deepened understanding of signal-domain behaviour in medical imaging, bridging the gap between MRI physics and practical reconstruction challenges. The analysis supports safe and accurate clinical diagnosis by improving image quality assessment and correction strategies.