I seem to be missing what possible use this is in getting survivors recovered.

Reclassifying stroke lesion anatomy

 

Author links open overlay panelAnna K.Bonkhoff^abTianboXu^bAmyNelson^bRobertGray^bAshwaniJha^bJorgeCardoso^cSebastienOurselin^cGeraintRees^dHans RolfJäger^b1ParashkevNachev^b

¹

https://doi.org/10.1016/j.cortex.2021.09.007Get rights and content

Under a Creative Commons license

open access

Abstract

Cognitive and behavioural outcomes in stroke reflect the interaction between two complex anatomically-distributed patterns: the functional organization of the brain and the structural distribution of ischaemic injury. Conventional outcome models—for individual prediction or population-level inference—commonly ignore this complexity, discarding anatomical variation beyond simple characteristics such as lesion volume. This sets a hard limit on the maximum fidelity such models can achieve. High-dimensional methods can overcome this problem, but only at prohibitively large data scales. Drawing on one of the largest published collections of anatomically-registered imaging of acute stroke—N=1333—here we use non-linear dimensionality reduction to derive a succinct latent representation of the anatomical patterns of ischaemic injury, agglomerated into 21 distinct intuitive categories. We compare the maximal predictive performance it enables against both simpler low-dimensional and more complex high-dimensional representations, employing multiple empirically-informed ground truth models of distributed structure-outcome relationships. We show our representation sets a substantially higher ceiling on predictive fidelity than conventional low-dimensional approaches, but lower than that achievable within a high-dimensional framework. Where descriptive simplicity is a necessity, such as within clinical care or research trials of modest size, the representation we propose arguably offers a favourable compromise of compactness and fidelity.

Keywords

Stroke

lesion anatomy

lesion–deficit prediction

dimensionality reduction

brain imaging

Abbreviations

DWI: diffusion-weighted imaging

t-SNE: t-stochastic neighbour embedding

NMF: non-negative matrix factorization

BA: Brodmann Area

Introduction

Stroke is remarkable in the wide diversity of its cognitive and behavioural manifestations and the difficulty of predicting them from the contemporaneous clinical picture alone(Boyd et al., 2017; Stinear, 2017; Ward, 2017). This cardinal aspect impedes the management of individual patients, the identification of protective or exacerbating factors in the population, and the quantification of treatment doses and effects. Were this heterogeneity biologically impossible to capture, we could do no more than to accept it as an unalterable fact of life. But it arises from the interaction of two biological characteristics that are, at least in theory, accessible even if complex enough to appear suffused with randomness. The first is the functional anatomy of the brain focal ischaemic injury definitionally disrupts, now comprehensively established to be not only highly complex but also remarkably consistent across individuals: meta-analytic imaging databases would otherwise be filled with noise, not generalisable clusters of coherent activation (Biswal et al., 2010; Eickhoff et al., 2018; Glasser et al., 2016). The second is the structural anatomy of stroke: the product of pathological and anatomical factors that are plausibly both highly complex and non-random (Adams Jr et al., 1993; Amarenco et al., 2009; Mah, Husain, et al., 2014). The topology of the vascular tree, the mechanisms of occlusion or rupture, and the symptomatic eloquence² of damaged brain will all combine to generate elaborate patterns of focal injury that will nonetheless conform to a potentially knowable spatial distribution (Figure 1). Since our knowledge of the functional anatomy of the brain depends to a great extent on the study of the functional consequences of stroke (Adolphs, 2016; Damasio & Damasio, 1989; Rorden & Karnath, 2004), the second of these characteristics is arguably of prior importance, and is our specific concern here.

1. Download : Download high-res image (159KB)
2. Download : Download full-size image