Tissue Is the Issue: A Systematic Review of Methods for the Determination of Infarct Volume in Acute Ischaemic Stroke

TL;DR

This paper reviews methods for measuring infarct volume in acute stroke, finding that automated and semi-automated tools with user refinement are most reliable.

Contribution

The study systematically evaluates the reproducibility and reliability of infarct volume measurement methods in acute ischaemic stroke.

Findings

The ABC/2 method is reliable for manual infarct volume estimation but can be inaccurate.

Semi-automated and automated methods with user refinement show high inter- and intra-rater correlation.

Standardization of image acquisition and reporting is needed to improve method performance and comparison.

Abstract

Background and aims: Recent advances in acute stroke interventions have highlighted the importance of accurate determination of infarct volume in the evaluation of acute stroke patients, carrying important prognostic and therapeutic implications for treatment planning, outcome prediction, and evaluation of the success of therapeutic interventions. However, there is no consensus on the methodologies employed to measure cerebral infarct volume. We aimed to assess the reproducibility and reliability of methods employed in the clinical determination of infarct volume in acute ischaemic stroke. Methods: We carried out a systematic review of studies assessing methodologies for the determination of infarct volume in the acute phase (<24 h). We searched Medline PubMed, Scopus, Cinahl, Cochrane Library, Web of Science, and Embase for studies examining image-based diagnosis of acute ischaemic stroke < 24 h by CT or MRI. Data on patient cohorts, imaging type, time from symptoms onset, methodologies and quantification strategies, rater reliability, accuracy, sensitivity, and specificity were compared. Results: We identified eighteen eligible studies with a total of 3298 ischaemic stroke patients assessing a variety of manual, semi-automated, and fully-automated methods. The ABC/2 method was found to be highly reliable, reproducible, and accurate, and provides the best manual estimate of infarction, but has a tendency to under- or overestimate infarct volume. Semi-automated and automated approaches with user refinement showed excellent inter-rater and intra-rater correlation. However, differences in operating algorithms and lack of standardisation of image acquisition parameters, quality, and format may impact performance and reproducibility. Conclusions: Of all methods, automated and semi-automated approaches utilising rater judgment and refinement represent the most robust approaches, with semi-automated tools demonstrating consistent and repeatable results. We recommend a standardised reporting of study methodologies for the accurate interpretation and comparison of efficacy of therapeutic interventions and patient outcomes, especially in a multi-centre setting. This may allow for more effective evaluation of stroke therapies and accelerate ischaemic stroke treatment decisions.

Figures