A Fast Protocol for Multiparametric Characterisation of Diffusion in the Brain and Brain Tumours

Ricardo Loução, Ana‐Maria Oros‐Peusquens, Karl‐Josef Langen, Hugo Alexandre Ferreira, N. Jon ShahFrontiers in Oncology 2021 · DOI

TL;DR

This paper introduces a rapid (4-minute) MRI protocol that simultaneously measures multiple diffusion parameters (IVIM, Gaussian, and non-Gaussian) in the brain. This approach aims to provide a more comprehensive characterization of brain tumors than standard methods, offering a potential for improved diagnosis and treatment planning.

What Problem Does This Paper Address?

Accurately characterizing brain tumors is crucial for diagnosis, grading, and treatment planning. Traditional MRI relies on T1 and T2-weighted images, which are often qualitative. Quantitative MRI (qMRI) offers more objective measurements, but existing multiparametric approaches are often time-consuming or require separate acquisitions. Specifically, understanding water diffusion in brain tissue is key, as it reflects microscopic environments. However, diffusion can behave in different ways: fast movement within capillaries (IVIM), hindered movement (Gaussian diffusion), and restricted movement due to complex tissue microstructure (non-Gaussian diffusion). Existing methods struggle to capture all these regimes efficiently, especially in the context of rapidly growing and heterogeneous brain tumors. This paper aims to bridge this gap by developing a single, fast protocol that can simultaneously assess these different diffusion behaviors.

Methods

The core of this work is a novel, fast diffusion-weighted MRI protocol that utilizes a "trace-based design." This means it acquires diffusion-weighted images along three orthogonal directions and combines them to approximate the diffusion trace. The protocol employs a wide range of b-values (0 to 3000 s/mm²) to capture three diffusion regimes:

  • Intravoxel Incoherent Motion (IVIM): Characterized by a fast-decaying signal component at low b-values, reflecting water movement in capillaries. This is modeled using a bi-exponential decay.
  • Gaussian Diffusion: Modeled by a mono-exponential signal decay, representing hindered water diffusion.
  • Non-Gaussian Diffusion (NG-diff): Characterized by deviations from mono-exponential decay at higher b-values, reflecting complex tissue microstructure. This is modeled using diffusion kurtosis imaging (DKI) formalism.

The protocol uses a simultaneous fitting approach to estimate parameters like perfusion fraction (f), pseudo-diffusivity (D*), apparent diffusivity (Dapp), and apparent kurtosis (Kapp) from a single acquisition. The researchers also implemented a denoising algorithm using Principal Component Analysis (PCA) to improve the signal-to-noise ratio (SNR), especially at higher b-values.

Datasets

The study utilized data from a cohort of 17 brain tumor patients (mean age ± std: 46.2 ± 12.4 years). These patients underwent simultaneous PET and MRI measurements on a hybrid Siemens scanner. The MRI dataset included standard clinical protocols (T1-weighted pre/post contrast, T2-weighted SPACE, FLAIR, and DSC) alongside the proposed research protocol. For comparison, a standard clinical Diffusion Kurtosis Imaging (DKI) protocol and Dynamic Susceptibility Contrast (DSC) perfusion measurements were also acquired. Amino acid (¹⁸F-FET) PET data was used to identify active tumor tissue.

Key Insights and Ideas

  • Simultaneous Multi-Regime Diffusion: The most significant insight is the successful development of a single, rapid MRI acquisition that can simultaneously probe and quantify parameters from IVIM, Gaussian diffusion, and non-Gaussian diffusion regimes. This offers a more holistic view of tissue microstructure and microcirculation.
  • Clinical Applicability of "Trace-Based" Design: The use of a "trace-based" acquisition, while not a strict diffusion tensor trace, is shown to be clinically relevant and a good approximation for characterizing diffusion properties, especially in regions with low anisotropy like tumors.
  • Denoising is Crucial: The study highlights the critical role of denoising, particularly PCA-based methods, in improving the stability and accuracy of fitting parameters from diffusion data, especially at higher b-values where SNR is lower.
  • IVIM vs. DSC Perfusion: The research suggests that IVIM-derived perfusion parameters (f and f.D*) are not direct surrogates for DSC-derived CBV and CBF. While correlations exist, they are weak, indicating that these methods capture different aspects of vascularity. This implies that combining both might offer a more complete picture.
  • DKI Metrics Correlate Well: The diffusion and kurtosis metrics derived from the proposed protocol show strong correlations with those from a dedicated DKI protocol, suggesting the proposed method can reliably estimate these parameters.

Key Results

  • The proposed 4-minute protocol successfully generated maps for D*, f, Dapp, and Kapp.
  • Diffusion and kurtosis metrics from the proposed protocol showed strong to very strong correlations with those from a standard DKI protocol (Spearman's r ranging from 0.48 to 0.89 for diffusivity and 0.45 to 0.84 for kurtosis).
  • A bias towards lower values was observed for the proposed protocol's diffusion and kurtosis metrics compared to the dedicated DKI protocol.
  • Pseudo-perfusion parameters (f and f.D*) showed very weak to weak correlations with their perfusion counterparts (CBV and CBF) from DSC (Spearman's r around 0.15 for simultaneous fit, even weaker for sequential fit).
  • The simultaneous fitting approach generally produced more precise results than sequential fitting, except for Kapp.
  • The proposed protocol's acquisition time (4 min 19 sec) is significantly shorter than the combined time for standard DKI (6 min 01 sec) and DSC (1 min).

Limitations and Future Work

  • IVIM Parameter Reliability: The study acknowledges that IVIM parameters, particularly D*, are not very reliable even at high SNR, and Dapp shows some bias. This is partly attributed to the limited number of b-values sampled in the IVIM regime and the small IVIM fraction in brain tissue.
  • Trace-Based Limitations: The "trace-based" design, while efficient, is not strictly the diffusion tensor trace and may not be ideal for regions with high anisotropy. The use of only three diffusion directions also limits the characterization of directional information.
  • Partial Volume Effects: The study did not explicitly account for partial volume effects, which can bias parameter estimates in voxels containing multiple tissue types.
  • Limited Patient Cohort: The small number of patients and the heterogeneity of the cohort limited the ability to perform detailed grading analyses.
  • Future Work: The authors suggest that including higher b-values could further characterize the slow diffusion component. They also propose that combining information from multiple diffusion regimes (IVIM, Gaussian, kurtosis) could lead to a more comprehensive tumor characterization and potentially a "tumor signature." Further validation with larger datasets and histological correlation is also recommended.

Summary updated

Abstract

Multi-parametric tissue characterisation is demonstrated using a 4-minute protocol based on diffusion trace acquisitions. Three diffusion regimes are covered simultaneously: pseudo-perfusion, Gaussian, and non-Gaussian diffusion. The clinical utility of this method for fast multi-parametric mapping for brain tumours is explored. A cohort of 17 brain tumour patients was measured on a 3T hybrid MR-PET scanner with a standard clinical MRI protocol, to which the proposed multi-parametric diffusion protocol was subsequently added. For comparison purposes, standard perfusion and a full diffusion kurtosis protocol were acquired. Simultaneous amino-acid (<sup>18</sup>F-FET) PET enabled the identification of active tumour tissue. The metrics derived from the proposed protocol included perfusion fraction, pseudo-diffusivity, apparent diffusivity, and apparent kurtosis. These metrics were compared to the corresponding metrics from the dedicated acquisitions: cerebral blood volume and flow, mean diffusivity and mean kurtosis. Simulations were carried out to assess the influence of fitting methods and noise levels on the estimation of the parameters. The diffusion and kurtosis metrics obtained from the proposed protocol show strong to very strong correlations with those derived from the conventional protocol. However, a bias towards lower values was observed. The pseudo-perfusion parameters showed very weak to weak correlations compared to their perfusion counterparts. In conclusion, we introduce a clinically applicable protocol for measuring multiple parameters and demonstrate its relevance to pathological tissue characterisation.
BibTeX
@inproceedings{loucao2021fast,
  author = {Ricardo Loução and Ana‐Maria Oros‐Peusquens and Karl‐Josef Langen and Hugo Alexandre Ferreira and N. Jon Shah},
  title = {{A Fast Protocol for Multiparametric Characterisation of Diffusion in the Brain and Brain Tumours}},
  booktitle = {Frontiers in Oncology},
  year = {2021},
  doi = {10.3389/fonc.2021.554205},
}
Figure