Can CT perfusion accurately assess infarct core?

Cerebral perfusion imaging predicts final infarct volume after basilar artery thrombectomy

OBJECTIVES

Cerebral perfusion imaging may be used to identify the ischemic core in acute ischemic stroke (AIS) patients with a large vessel occlusion of the anterior circulation; however, perfusion parameters that predict the ischemic core in AIS patients with a basilar artery occlusion (BAO) are poorly described. We determined which cerebral perfusion parameters best predict the ischemic core after successful endovascular thrombectomy (EVT) in BAO patients.

MATERIALS AND METHODS

We performed multicenter retrospective study of BAO patients with perfusion imaging before EVT and a DWI after successful EVT. The ischemic core was defined as regions on CTP, which were co-registered to the final DWI infarct. Various time-to-maximum (Tmax) and cerebral blood flow (CBF) thresholds were compared to final infarct volume to determine the best predictor of the final infarct.

RESULTS

28 patients were included in the analysis for this study. Tmax >8s (r²: 0.56; median absolute error, 16.0 mL) and Tmax >10s (r²: 0.73; median absolute error, 11.3 mL) showed the strongest agreement between the pre-EVT CTP study and the final DWI. CBF <38% (r²: 0.76; median absolute error, 8.2 mL) and CBF <34% (r²: 0.76; median absolute error, 9.1 mL) also correlated well with final infarct volume on DWI.

CONCLUSIONS

Pre-EVT CT perfusion imaging is useful to predict the final ischemic infarct volume in BAO patients. Tmax >8s and Tmax >10s were the strongest predictors of the post-EVT final infarct volume.

ISP-Net: Fusing features to predict ischemic stroke infarct core on CT perfusion maps

BACKGROUND

Acute ischemic stroke is one of the leading death causes. Delineating stoke infarct core in medical images plays a critical role in optimal stroke treatment selection. However, accurate estimation of infarct core still remains challenging because of 1) the large shape and location variation of infarct cores; 2) the complex relationships between perfusion parameters and final tissue outcome.

METHODS

We develop an encoder-decoder based semantic model, i.e., Ischemic Stroke Prediction Network (ISP-Net), to predict infarct core after thrombolysis treatment on CT perfusion (CTP) maps. Features of native CTP, CBF (Cerebral Blood Flow), CBV (Cerebral Blood Volume), MTT (Mean Transit Time), Tmax are generated and fused with five-path convolutions for comprehensive analysis. A multi-scale atrous convolution (MSAC) block is firstly put forward as the enriched high-level feature extractor in ISP-Net to improve prediction accuracy. A retrospective dataset which is collected from multiple stroke centers is used to evaluate the performance of ISP-Net. The gold standard infarct cores are delineated on the follow-up scans, i.e., non-contrast CT (NCCT) or MRI diffusion-weighted image (DWI).

RESULTS

In clinical dataset cross-validation, we achieve mean Dice Similarity Coefficient (DSC) of 0.801, precision of 81.3%, sensitivity of 79.5%, specificity of 99.5%, Area Under Curve (AUC) of 0.721. Our approach yields better outcomes than several advanced deep learning methods, i.e., Deeplab V3, U-Net++, CE-Net, X-Net and Non-local U-Net, demonstrating the promising performance in infarct core prediction. No significant difference of the prediction error is shown for the patients with follow-up NCCT and follow-up DWI (P >0.05).

CONCLUSION

This study provides an approach for fast and accurate stroke infarct core estimation. We anticipate the prediction results of ISP-Net could offer assistance to the physicians in the thrombolysis or thrombectomy therapy selection.

Ischemic lesion growth in acute stroke: Water uptake quantification distinguishes between edema and tissue infarct

Infarct growth from the early ischemic core to the total infarct lesion volume (LV) is often used as an outcome variable of treatment effects, but can be overestimated due to vasogenic edema. The purpose of this study was (1) to assess two components of early lesion growth by distinguishing between water uptake and true net infarct growth and (2) to investigate potential treatment effects on edema-corrected net lesion growth. Sixty-two M1-MCA-stroke patients with acute multimodal and follow-up CT (FCT) were included. Ischemic lesion growth was calculated by subtracting the initial CTP-derived ischemic core volume from the LV in the FCT. To determine edema-corrected net lesion growth, net water uptake of the ischemic lesion on FCT was quantified and subtracted from the volume of uncorrected lesion growth. The mean lesion growth without edema correction was 20.4 mL (95% CI: 8.2-32.5 mL). The mean net lesion growth after edema correction was 7.3 mL (95% CI: -2.1-16.7 mL; p < 0.0001). Lesion growth was significantly overestimated due to ischemic edema when determined in early-FCT imaging. In 18 patients, LV was lower than the initial ischemic core volume by CTP. These apparently "reversible" core lesions were more likely in patients with shorter times from symptom onset to imaging and higher recanalization rates.

Incorporation of relative cerebral blood flow into CT perfusion maps reduces false ’at risk' penumbra

Purpose

The region defined as ‘at risk’ penumbra by current CT perfusion (CTP) maps is largely overestimated. We aimed to quantitate the portion of true ‘at risk’ tissue within CTP penumbra and to determine the parameter and threshold that would optimally distinguish it from false ‘at risk’ tissue, that is, benign oligaemia.

Methods

Among acute stroke patients evaluated by multimodal CT (NCCT/CTA/CTP) we identified those that had not undergone endovascular/thrombolytic treatment and had follow-up NCCT. Maps of absolute and relative CBF, CBV, MTT, TTP and Tmax as well as summary maps depicting infarcted and penumbral regions were generated using the Intellispace Portal (Philips Healthcare, Best, Netherlands). Follow-up CT was automatically co-registered to the CTP scan and the final infarct region was manually outlined. Perfusion parameters were systematically analysed – the parameter that resulted in the highest true-negative-rate (ie, proportion of benign oligaemia correctly identified) at a fixed, clinically relevant false-negative-rate (ie, proportion of ‘missed’ infarct) of 15%, was chosen as optimal. It was then re-applied to the CTP data to produce corrected perfusion maps.

Results

Forty seven acute stroke patients met selection criteria. Average portion of infarcted tissue within CTP penumbra was 15%±2.2%. Relative CBF at a threshold of 0.65 yielded the highest average true-negative-rate (48%), enabling reduction of the false ‘at risk’ penumbral region by ~half.

Conclusions

Applying a relative CBF threshold on relative MTT-based CTP maps can significantly reduce false ‘at risk’ penumbra. This step may help to avoid unnecessary endovascular interventions.

Deep learning guided stroke management: a review of clinical applications

Stroke is a leading cause of long-term disability, and outcome is directly related to timely intervention. Not all patients benefit from rapid intervention, however. Thus a significant amount of attention has been paid to using neuroimaging to assess potential benefit by identifying areas of ischemia that have not yet experienced cellular death. The perfusion-diffusion mismatch, is used as a simple metric for potential benefit with timely intervention, yet penumbral patterns provide an inaccurate predictor of clinical outcome. Machine learning research in the form of deep learning (artificial intelligence) techniques using deep neural networks (DNNs) excel at working with complex inputs. The key areas where deep learning may be imminently applied to stroke management are image segmentation, automated featurization (radiomics), and multimodal prognostication. The application of convolutional neural networks, the family of DNN architectures designed to work with images, to stroke imaging data is a perfect match between a mature deep learning technique and a data type that is naturally suited to benefit from deep learning's strengths. These powerful tools have opened up exciting opportunities for data-driven stroke management for acute intervention and for guiding prognosis. Deep learning techniques are useful for the speed and power of results they can deliver and will become an increasingly standard tool in the modern stroke specialist's arsenal for delivering personalized medicine to patients with ischemic stroke.