Diffusion MRI

In a normal state random diffusion of water molecules occurs across cell membranes. Tissue damage and ischemia begin to occur at a cerebral blood flow value of approximately 30 mL per 100 g per minute. Severe ischemia of brain tissue induces the rapid dysfunction of cellular metabolism and ion exchange pumps responsible for adenosine triphosphate (ATP) metabolism. This dysfunction causes a massive shift of water from the extracellular to the intracellular compartment, leading to cytotoxic edema that is associated with restricted diffusion of water molecules. This reduction in water diffusion can be quantified by a decrease in the apparent diffusion coefficient (ADC) and can be visualized as an increased signal on DWI. In contrast to CT, in which brain ischemia is typically not visualized for several hours, diffusion signal increases within minutes of stroke onset. Several studies have confirmed the greater sensitivity of DWI for detecting stroke compared with CT scanning or conventional MRI during the acute period ( Fig. 1 ). Thus, in the evaluation of acute stroke, DWI increases diagnostic confidence and may help identify candidates for reperfusion therapies.

A 53-year-old right-handed man presented with a sudden onset of aphasia. Serial axial CT scans ( top row ) at 1 hour after the symptom onset showed no obvious abnormalities. The patient received intravenous alteplase. Serial diffusion-weighted imaging scans at the same levels as the CT ( bottom row ) obtained approximately 1 hour later show no definite ischemia in the left middle cerebral artery territory.

DWI may also yield important prognostic information in the evaluation of acute stroke. Large-volume hemispheric ischemic strokes that cause severe space-occupying edema with the risk of resultant brain herniation are called malignant infarctions ( Fig. 2 ). Early DWI lesion volumes larger than 145 mL predict malignant infarction. Despite medical treatments to reduce intracranial pressure, their prognosis is poor, with fatality rates of nearly 80%. A life-saving emergent surgical decompression (hemicraniectomy), if performed early, reduces mortality and increases the chance for favorable functional outcome in patients with malignant infarction. DWI lesion volume larger than 145 mL is now a validated criterion for emergent hemicraniectomy.

A 62-year-old man presenting 5 hours after the onset of left-sided weakness and hemianopia. Diffusion-weighted imaging ( top row ) and apparent diffusion coefficient ( bottom row ) maps show a malignant right middle cerebral artery/posterior cerebral artery territory infarct (approximately 200 mL).

In addition to providing information regarding which patients may benefit from emergent surgical decompression, DWI lesions may also predict the risk of hemorrhagic transformation after thrombolysis. The risk for symptomatic intracerebral hemorrhage after thrombolysis has been associated with pretreatment DWI lesion size. Additionally, areas of severe ADC reduction have been reported to predict subsequent hemorrhagic transformation. In the Diffusion and Perfusion Imaging Evaluation For Understanding Stroke Evolution (DEFUSE) study, patients with a “malignant MRI profile” characterized by a DWI lesion larger than 100 mL and/or a PWI lesion larger than 100 mL with longer than 8 seconds of T _max delay had a high risk of symptomatic hemorrhagic transformation and death after rt-PA therapy. Furthermore, in a more recent study, patients with a DWI lesion larger than 70 mL before intra-arterial thrombolysis had a poor prognosis and high mortality rate despite a 50% recanalization rate. The large volume of severe ischemic brain and microvascular injury in these malignant infarcts may increase the risk for reperfusion-related hemorrhage. Excluding patients with unfavorable imaging profiles from recanalization therapies may improve the safety and efficacy of these therapies.

The degree of reduction in ADC values may also predict DWI lesion reversal after recanalization/reperfusion. Several studies have shown that, for patients who do not have a malignant stroke profile, acute DWI lesions can be partially reversed through recanalization/reperfusion. These findings indicate that an acute DWI lesion does not necessarily represent irreversible injury, and may contain a portion of the ischemic penumbra. ADC values typically decrease sharply soon after stroke onset. In animal studies, the reduction in ADC in acute stroke has been shown to occur before irreversible failure of energy metabolism. Other supportive data have emerged from voxel-based comparisons between MRI and PET. Portions of acute DWI lesions show PET scan characteristics of penumbra. Higher ADC values and successful early reperfusion are factors associated with DWI reversal. In the DEFUSE study, an ADC threshold of 615 × 10 ⁻⁶ mm ² /s or less provided a sensitivity of 80% and specificity of 77% for identifying the infarct core. Thus, ADC thresholds may be useful to help distinguish ischemic core from penumbra.

Perfusion MRI

Bolus tracking perfusion imaging using a paramagnetic contrast agent (eg, gadolinium) is the most commonly used MRI technique for assessing cerebral perfusion. The delay in gadolinium arrival to the vascular bed and the intensity of the signal change induced by its passage within capillaries can be used to generate a time/concentration curve. Various maps and parameters can be generated from these data, including cerebral blood flow (CBF), cerebral blood volume (CBV), and time-based factors such as T _max , mean transit time (MTT), and time to peak (TTP) ( Fig. 3 ).

A 55-year-old man approximately 4 hours after the onset of expressive aphasia. ( A ) Diffusion-weighted imaging (DWI) of the left frontal infarct. Hemodynamic maps from the MR perfusion study are obtained at the same time: ( B ) cerebral blood flow (CBF), ( C ) cerebral blood volume (CBV), ( D ) mean transit time (MTT), and ( E ) T _max . Within the infarct core, a markedly reduced CBV and CBF and markedly elevated MTT and T _max are seen. Posterior to the infarct core (the DWI lesion), in the left temporal parietal region, a region with less severely reduced CBF and less severely prolonged MTT and T _max is seen, consistent with the penumbra.

The concept of the perfusion-diffusion (PWI/DWI) mismatch for selecting patients for thrombolysis was introduced with several small case series approximately 10 years ago, followed by larger series by several international groups over the past few years. Studies have used multiple different perfusion parameters and thresholds to define critical hypoperfusion as the PWI lesion that will predict the extent of final infarction if the tissue does not experience early reperfusion. T _max , the time to peak of the residue function, indicating a delay in bolus arrival between the site of selection of the arterial input function and the tissue, was used to define critical hypoperfusion in both the DEFUSE trial and the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET). These studies selected a T _max value of greater than 2 seconds to threshold the PWI lesions in an attempt to exclude benign oligemia. However, retrospective analyses of these studies showed that higher T _max thresholds provide a more accurate assessment of critical hypoperfusion. For instance, a post hoc analysis of DEFUSE showed that a PWI lesion with a T _max delay between 4 and 6 seconds provides a better estimation of final infarct volume in patients who did not reperfuse compared with the threshold of greater than 2 second threshold. These findings are supported by the results of direct voxel-to-voxel comparisons between MRI PWI and either PET or Xe-CT scans that suggest that 4- to 6-second T _max delays correspond most closely with penumbral cerebral blood flow values on PET or Xe-CT. A pooled analysis of both DEFUSE and EPITHET datasets presented at the International Stroke Conference in February 2010 showed that PWI lesions larger than 125 mL defined by a T _max threshold of greater than 8 seconds were associated with significantly worse clinical outcomes in patients treated with rt-PA versus placebo.

Overall, the refinement of PWI techniques for acute stroke is ongoing. PWI does not provide an accurate quantitative estimate of cerebral blood flow in individual stroke patients compared with Xe-CT or PET gold standards. Corrections for contrast delay and dispersion seem to improve PWI performance. Although penumbral imaging algorithms are part of several guidelines, including those of the American Heart Association and European Stroke Organization, the ability of DWI/PWI mismatch to detect penumbra requires further validation using quantitative methods.

Perfusion MRI

Bolus tracking perfusion imaging using a paramagnetic contrast agent (eg, gadolinium) is the most commonly used MRI technique for assessing cerebral perfusion. The delay in gadolinium arrival to the vascular bed and the intensity of the signal change induced by its passage within capillaries can be used to generate a time/concentration curve. Various maps and parameters can be generated from these data, including cerebral blood flow (CBF), cerebral blood volume (CBV), and time-based factors such as T _max , mean transit time (MTT), and time to peak (TTP) ( Fig. 3 ).

A 55-year-old man approximately 4 hours after the onset of expressive aphasia. ( A ) Diffusion-weighted imaging (DWI) of the left frontal infarct. Hemodynamic maps from the MR perfusion study are obtained at the same time: ( B ) cerebral blood flow (CBF), ( C ) cerebral blood volume (CBV), ( D ) mean transit time (MTT), and ( E ) T _max . Within the infarct core, a markedly reduced CBV and CBF and markedly elevated MTT and T _max are seen. Posterior to the infarct core (the DWI lesion), in the left temporal parietal region, a region with less severely reduced CBF and less severely prolonged MTT and T _max is seen, consistent with the penumbra.

The concept of the perfusion-diffusion (PWI/DWI) mismatch for selecting patients for thrombolysis was introduced with several small case series approximately 10 years ago, followed by larger series by several international groups over the past few years. Studies have used multiple different perfusion parameters and thresholds to define critical hypoperfusion as the PWI lesion that will predict the extent of final infarction if the tissue does not experience early reperfusion. T _max , the time to peak of the residue function, indicating a delay in bolus arrival between the site of selection of the arterial input function and the tissue, was used to define critical hypoperfusion in both the DEFUSE trial and the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET). These studies selected a T _max value of greater than 2 seconds to threshold the PWI lesions in an attempt to exclude benign oligemia. However, retrospective analyses of these studies showed that higher T _max thresholds provide a more accurate assessment of critical hypoperfusion. For instance, a post hoc analysis of DEFUSE showed that a PWI lesion with a T _max delay between 4 and 6 seconds provides a better estimation of final infarct volume in patients who did not reperfuse compared with the threshold of greater than 2 second threshold. These findings are supported by the results of direct voxel-to-voxel comparisons between MRI PWI and either PET or Xe-CT scans that suggest that 4- to 6-second T _max delays correspond most closely with penumbral cerebral blood flow values on PET or Xe-CT. A pooled analysis of both DEFUSE and EPITHET datasets presented at the International Stroke Conference in February 2010 showed that PWI lesions larger than 125 mL defined by a T _max threshold of greater than 8 seconds were associated with significantly worse clinical outcomes in patients treated with rt-PA versus placebo.

Overall, the refinement of PWI techniques for acute stroke is ongoing. PWI does not provide an accurate quantitative estimate of cerebral blood flow in individual stroke patients compared with Xe-CT or PET gold standards. Corrections for contrast delay and dispersion seem to improve PWI performance. Although penumbral imaging algorithms are part of several guidelines, including those of the American Heart Association and European Stroke Organization, the ability of DWI/PWI mismatch to detect penumbra requires further validation using quantitative methods.