Early Identification of Potentially Salvageable Tissue with MRI-Based Predictive Algorithms after Experimental Ischemic Stroke

Abstract

Individualized stroke treatment decisions can be improved by accurate identification of the extent of salvageable tissue. Magnetic resonance imaging (MRI)-based approaches, including measurement of a 'perfusion-diffusion mismatch' and calculation of infarction probability, allow assessment of tissue-at-risk; however, the ability to explicitly depict potentially salvageable tissue remains uncertain. In this study, five predictive algorithms (generalized linear model (GLM), generalized additive model, support vector machine, adaptive boosting, and random forest) were tested in their potency to depict acute cerebral ischemic tissue that can recover after reperfusion. Acute T 2 -, diffusion-, and perfusion-weighted MRI, and follow-up T 2 maps were collected from rats subjected to right-sided middle cerebral artery occlusion without subsequent reperfusion, for training of algorithms (Group I), and with spontaneous (Group II) or thrombolysis-induced reperfusion (Group III), to determine infarction probability-based viability thresholds and prediction accuracies. The infarction probability difference between irreversible-i.e., infarcted after reperfusionand salvageable tissue injury-i.e., noninfarcted after reperfusion-was largest for GLM (20±7%) with highest accuracy of riskbased identification of acutely ischemic tissue that could recover on subsequent reperfusion (Dice's similarity index ¼ 0.79±0.14). Our study shows that assessment of the heterogeneity of infarction probability with MRI-based algorithms enables estimation of the extent of potentially salvageable tissue after acute ischemic stroke.

The precision–recall curve overcame the optimism of the receiver operating characteristic curve in rare diseases

“…This has been recently done in the field of medical imaging where the prevalence of diseased voxels after stroke is usually low. The AUC gave a biased view of the model predictive performance, whereas the AUPRC reflected better the quality of the prediction of the diseased volumes [9].…”

“…However, other authors have shown that the area under the ROC curve (AUC) and the area under the PR curve (AUPRC) are not equivalent but show very large differences [9]. In the case of rare events, some authors have recommended the use of the AUPRC instead of the AUC [7]; however, up to now, no comparisons are readily available.…”

Lesion Development and Reperfusion Benefit in Relation to Vascular Occlusion Patterns after Embolic Stroke in Rats

“…10 Perfusion-diffusion mismatch volume was calculated from the difference between the total volumes of ADC and MTT abnormality. The presence of a significant perfusion-diffusion mismatch was confirmed when the MTT-based perfusion lesion volume was 20% larger than the ADC-based tissue lesion volume.…”

“…Benefit of reperfusion was defined as the absence of lesion growth beyond 10% of the acute ADC-based tissue lesion volume, as measured from the follow-up T 2 lesion volume at poststroke day 7 (calculated as the ipsilateral tissue volume with T 2 values above 2 standard deviations from mean contralateral tissue values 10 ).…”

Infarct Volume Prediction by Early Magnetic Resonance Imaging in a Murine Stroke Model Depends on Ischemia Duration and Time of Imaging

“…38 Desai et al 43 showed that prediction could be improved by measuring ADC changes at multiple time points. Wu et al 37 and Bouts et al 44 found high prediction accuracy in rats using a general linear model combining perfusion and diffusion information. Christensen et al 45 have demonstrated increased prediction accuracy of perfusion imaging by more sophisticated CBF parameter analysis.…”