Data-Driven multi-Contrast spectral microstructure imaging with InSpect: INtegrated SPECTral component estimation and mapping.

Paddy J Slator,Mary Rutherford,Alison Ho,Laurence H Jackson,Lucy C Chappell,Joseph V Hajnal,Marco Palombo,Razvan V Marinescu,Jana Hutter,Daniel C Alexander

doi:10.1016/j.media.2021.102045

Paddy J Slator, Mary Rutherford + Show 8 more

Open Access

https://doi.org/10.1016/j.media.2021.102045

Copy DOI

Abstract

We introduce and demonstrate an unsupervised machine learning technique for spectroscopic analysis of quantitative MRI experiments. Our algorithm supports estimation of one-dimensional spectra from single-contrast data, and multidimensional correlation spectra from simultaneous multi-contrast data. These spectrum-based approaches allow model-free investigation of tissue properties, but require regularised inversion of a Laplace transform or Fredholm integral, which is an ill-posed calculation. Here we present a method that addresses this limitation in a data-driven way. The algorithm simultaneously estimates a canonical basis of spectral components and voxelwise maps of their weightings, thereby pooling information across whole images to regularise the ill-posed problem. We show in simulations that our algorithm substantially outperforms current voxelwise spectral approaches. We demonstrate the method on multi-contrast diffusion-relaxometry placental MRI scans, revealing anatomically-relevant sub-structures, and identifying dysfunctional placentas. Our algorithm vastly reduces the data required to reliably estimate spectra, opening up the possibility of quantitative MRI spectroscopy in a wide range of new applications. Our InSpect code is available at github.com/paddyslator/inspect.

Highlights

Quantitative MRI can measure and map physical and chemical quantities that strongly relate to underlying tissue structure and function
Quantitative MRI spectroscopy has been demonstrated for single-contrast approaches, including T2 component analysis of multi-echo relaxometry data, as has been used to image myelin
We show that InSpect spectral components appear to be associated with distinct anatomical features in the placenta

Summary

Introduction

Quantitative MRI can measure and map physical and chemical quantities that strongly relate to underlying tissue structure and function. A convenient, and data-driven, way to analyse quantitative MRI data is to assume a population of spins with a distribution of quantities (e.g. relaxivity, diffusivity) that are encoded in a one-dimensional, or multidimensional correlation, spectrum. By estimating such distributions, multiple microstructural components can be distinguished without making a-priori modelling assumptions, such as fixing the number of tissue compartments. Multiple microstructural components can be distinguished without making a-priori modelling assumptions, such as fixing the number of tissue compartments This approach, which we refer to as quantitative MRI spectroscopy in this paper, has the potential to provide novel biomarkers (Bai et al, 2014). Several papers have leveraged recent advances in scanner hardware to extend these ideas into imaging, in the T1-diffusion (De Santis et al, 2016b; 2016a), T2-diffusion (Veraart et al, 2018; Kim et al, 2017a; Melbourne et al, 2018; De Almeida Martins and Topgaard, 2018; Lampinen et al, 2020; De Almeida Martins et al, 2020; Reymbaut et al, 2020; Gong et al, 2020), T1-T2-diffusion (Benjamini and Basser, 2017), T2∗-diffusion (Slator et al, 2019b), and T1-T2∗diffusion (Hutter et al, 2018) domains

Objectives

Methods

Results

Discussion

Conclusion