Rapid processing and quantitative evaluation of structural brain scans for adaptive multimodal imaging

Abstract

Current neuroimaging acquisition and processing approaches tend to be optimised for quality rather than speed. However, rapid acquisition and processing of neuroimaging data can lead to novel neuroimaging paradigms, such as adaptive acquisition, where rapidly processed data is used to inform subsequent image acquisition steps. Here we first evaluate the impact of several processing steps on the processing time and quality of registration of manually labelled T1‐weighted MRI scans. Subsequently, we apply the selected rapid processing pipeline both to rapidly acquired multicontrast EPImix scans of 95 participants (which include T1‐FLAIR, T2, T2*, T2‐FLAIR, DWI and ADC contrasts, acquired in ~1 min), as well as to slower, more standard single‐contrast T1‐weighted scans of a subset of 66 participants. We quantify the correspondence between EPImix T1‐FLAIR and single‐contrast T1‐weighted scans, using correlations between voxels and regions of interest across participants, measures of within‐ and between‐participant identifiability as well as regional structural covariance networks. Furthermore, we explore the use of EPImix for the rapid construction of morphometric similarity networks. Finally, we quantify the reliability of EPImix‐derived data using test–retest scans of 10 participants. Our results demonstrate that quantitative information can be derived from a neuroimaging scan acquired and processed within minutes, which could further be used to implement adaptive multimodal imaging and tailor neuroimaging examinations to individual patients.

Tissue volume estimation and age prediction using rapid structural brain scans

“…Both global and voxelwise analyses showed that grey matter, white matter and cerebrospinal fluid volumes are comparable between T 1 -w and EPImix T 1 -w scans. This extends our previous findings of correspondence between tissue intensities and Jacobian determinants (derived from registration of T 1 -w scans to MNI standard space), between EPImix and standard T 1 -w scans 7 . While Jacobian determinants can be derived relatively rapidly from T 1 -w scans, the primary constraint being the speed of non-linear registration, they are both less interpretable and less widely used than estimates of tissue volume.…”

“…In this way, active acquisition holds the potential for reduced scan time and improved accuracy, reliability and individualisation of (neuro)imaging 16 . The main feasibility obstacles to active acquisition are the speed of image collection as well as data processing and analysis; to realise the benefits of this multimodal adaptive approach, analyses need to be carried out in near real-time 7,16 . Due to the relative speed of the new EPImix sequence, it could contribute considerably to this process.…”

“…However, both sequences are T 1 -weighted and are referred to as such [as well as simply "T 1 -w"].) We further demonstrated the utility of the multicontrast EPImix sequence to construct morphometric similarity networks (MSNs), which provide individual estimates of anatomical connectivity 8 , in minutes following participants entering the MRI scanner 7 . However, EPImix scans have not previously been compared to standard sequences using established quantitative structural neuroimaging measures such as estimates of tissue volume 9 or predicted brain-age, a putative biomarker of brain health 10,11 .…”

“…Still, as both sequences are T 1 -weighted, we refer to both as such (as well as simply "T 1 -w").) using simple rapidly-derived quantitative measurements, including image intensity and Jacobian determinants (obtained from the registration to Montreal Neurological Institute (MNI) standard space), and found high similarity in these rudimentary quantitative measures as well as their potential to identify inter-individual differences 7 . (Note that the EPImix sequence includes a T 1 -FLAIR contrast, while the "conventional" single-contrast scans were acquired using an IR-FSPGR sequence.…”

Network level enrichment provides a framework for biological interpretation of machine learning results

“…It is worth noting that many investigators have studied the proper ways to draw a null distribution. See, for example, Xia et al (2018) shuffled the networks in addition to permuting the labels to generate the null distribution; Zamani Esfahlani et al (2020) summarized different ways to create a null model that could preserve neuroanatomical realism; and more recently, Váša et al (2021) provided a comprehensive review on the logic, implementation and interpretation of null models for functional connectomes. Here, we have not shuffled the functional networks of the brain in order to preserve the underlying covariance structure of real biological data (Nichols & Hayasaka, 2016).…”