A protocol for stroke onset time estimation in a rat model of stroke exploiting quantitative magnetic resonance imaging (qMRI) parameters is described. The procedure exploits diffusion MRI for delineation of the acute stroke lesion and quantitative T1 and T2 (qT1 and qT2) relaxation times for timing of stroke.
Magnetic resonance imaging provides a sensitive and specific tool to detect acute ischemic stroke by determining the Apparent Diffusion Coefficient of brain tissue termed ADC. In a rat model of ischemic stroke, hemispheric differences in quantitative T1 and T2 relaxation times within the ischemic lesion increased with time from stroke onset. The time dependency of these differences are heuristically described by a linear function and so it provides simple estimates of stroke onset time.
The volumes of tissue with elevated T1 and T2 relaxation times within the ischemic lesion also increases linearly, providing a complimentary method for stroke timing. Here, we present a semi-automated computer routine that uses ADC to delineate acute ischemic stroke tissue in the hyperacute phase of permanent focal ischemia. Elevated relaxation times also identified within the ischemic stroke tissue are used to determine stroke onset time.
Male Wistar rats weighing 300 to 400 grams were used and given permanent middle cerebral artery occlusion to induce focal ischemic stroke. For the duration of the operation and MRI experiments, the rat is anesthetized with isoflurane through a face mask. After making an incision in the neck, the trachea is exposed and muscles are separated from the trachea.
The wound is expanded to clear the path to the Common Carotid Artery or CCA. The CCA is exposed whilst avoiding the nerve bundle next to it. A micro hook is used to catch the CCA and expose the External Carotid Artery, the ECA, and the Internal Carotid Artery, the ICA.
The circulation of the CCA is blocked with surgical thread. The ECA is sutured. The carotid body is separated from the ICA and the bifurcation of the ICA is located to ensure that when the occluding thread is inserted, it reaches and blocks the left middle cerebral artery.
A loose suture is introduced around the ICA. In the ICA, a clip is inserted so that it is placed deeper than the suture. The ECA is kept close to the double knot that is tied round the external carotid artery.
The excess part of the ECA is burned to prevent hemorrhage from the ECA. The occluding thread is inserted into the ECA and the ICA is aligned with the ECA. The thread is then advanced up to the clip.
The loose suture is then tightened so that the thread can still slide inside the ECA. The clip is then removed and the occluding thread is further advanced until approximately two centimeters has been inserted. Another knot is added so the loose knot becomes a double knot.
The sutures are tied up and retractors removed. The wound is then closed. Immediately after occlusion, the rat is secured in a cradle at the center of the 9.4 Tesla magnet bar using ear buds and a bite block.
During the imaging, isoflurane levels are maintained at around 2%and temperature is maintained close to 37 degrees Celsius using a water heating pad placed under the rat torso. During imaging, the breathing rate and rectal temperature are monitored. MRI data is acquired for up to five hours post occlusion and at hourly intervals, 12 congruently sampled axial slices of the trace of diffusion tensor, CPMG T2, and Flash T1 are acquired.
Details of sequence parameters are listed in the manuscript. At the end of the experiment, rats are sacrificed and Triphenyl Tetrazolium Chloride or TTC staining is performed to confirm the presence of ischemia in rat brains. Details for the TTC staining are given in the manuscript.
Having obtained MRI data, we next sought to develop a semi-automated method to detect the ischemic tissue and characterize its relaxometric signature. T2 relaxation was modeled as a mono exponential decay. T2 maps were fitted by taking the natural algorithm along the echo time dimension and fitting a linear function at each voxel.
T1 relaxation was modeled as a mono exponential returned to equilibrium magnetization and fitted using the binary chalk method. ADC maps were computed assuming mono exponential loss of signal with respect to B value. With all quantitative images in place, ischemic tissue could be identified using the reciprocal ADC maps.
These maps were thresholded such that the tissue was labeled as ischemic if the reciprocal ADC exceeded one median absolute deviation above the median reciprocal ADC of nonischemic tissue. This process could be carried out automatically and objectively. With ischemic tissue demarcated, the distribution of T1 and T2 in each volume of interest was calculated.
To obtain the distribution of T1 and T2 in the contralateral hemisphere, the volume of interest for the ischemic tissue was reflected. In this example, the T1 and T2 distributions can be seen to right shift with time after onset. To describe the changes in the distributions of T1 and T2 with time, we use two different parameters.
The parameter f is defined as shown. This provides a heuristic parameterization of the fraction of voxels in the ischemic lesion which have unusually high or low T1 or T2 values. It is required that voxel in a T1 or T2 map may be categorized as having a high, low, or normal value.
A high T1 or T2 value in the ischemic lesion was defined as one exceeding 1/2 maximum above the median T1 or T2 of the nonischemic volume of interest. F1 is the fraction of ischemic voxels with high T1 values, F2 the fraction of voxels with high T2 values. Secondly, the volume of overlap within the regions of ADC identified ischemic tissue with both high T1 and high T2 is calculated, normalized by the total brain volume.
The F1, F2, and volume of overlap were calculated for all rats at all times of measurement. They were then separately used to back calculate the time of onset of ischemia by fitting a linear model. By doing so, the uncertainty in onset times could be calculated.
These uncertainties were parameterized by the root-mean-square error for the linear fit. In the diffusion positive stroke lesion, all parameters increased linearly with time. And according to the root-mean-square error bars, our protocol provides stroke onset time estimates with an accuracy of half an hour.
It is a general trend that the T2 data outperforms T1 and the best accuracy for onset time determination is obtained from the volume of overlapping elevated T1 and T2.Regions with abnormal T1 and T2 are heterogeneously scattered within the ischemic lesion likely due to different sensitivities of these quantitative MRI parameters to pathophysiological changes caused by ischemia. This suggests that quantitative MRI parameters may also be informative of tissue status. Here, we presented an MRI protocol that enables stroke onset time to be estimated with considerable accuracy and by doing so is informative of brain tissue status during permanent ischemia.
The benefit of quantifying relaxation times for estimation of stroke onset time is that they're insensitive to inherent variations caused by technical factors such as magnetic field inhomogeneities and proton density including the expected magnetic field variation within the ischemic lesion. It should be realized that the current protocol applies only to permanent ischemia and the MRI data acquisition parameters given in the paper are specific to the 9.4 Tesla MRI scanner used. Our previous studies in the attached paper show that the procedure is also accurate at 4.7 Tesla.
Overall, we believe that our protocol has application in preclinical stroke research including studies investigating anti-ischemic therapies.
