Evaluation of Acute Cerebral Infarction Using a Fast Kurtosis Diffusion imaging Protocol
Chengxu Li1, Tianyi Qian2, Jinsuh Kim3, Philip Zhe Sun4, Jie Lu1, and Kuncheng Li1

1Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China, People's Republic of, 2MR Collaborations NE Asia, Siemens Healthcare, Beijing, China, People's Republic of, 3Department of Radiology, University of Illinois at Chicago, Chicago, IL, United States, 4Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, United States


Conventional DKI is limited in use for detecting acute stroke because of its relatively long scan time and current need of offline processing. Here we used a 13 directions protocol combined with simultaneous multi-slice technique with inline reconstruction to test the diffusion pattern at different time points. The results shows the lesion size observed within 24hr on MK matched with the lesion size on T2-FLAIR after one month, which was better than lesion size observed within 24hr on DWI. Early application of the fast DKI to detect the anomalous range of MK was beneficial to predict the range of the eventual infarction for clinical treatment.


Stroke was the second most common cause of death and the third leading cause of disability in 2010. Diffusion-weighted imaging (DWI) has become one of the most widely used acute stroke imaging techniques, even though non-Gaussian diffusion is observed throughout the brain. Diffusion kurtosis imaging (DKI) is an advanced technique to quantify information on the non-Gaussian movement of water. However, the time of routine DKI acquisition is relatively long and image data needs to be post-processed offline, limiting its use in an acute stroke setting. Hansen et al.1proposed an approach of fast DKI acquisition and post-processing to substantially reduce the scanning time, and P.Z. SUN et al.2demonstrated that fast DKI protocols correlate with conventional DKI protocols in a rodent stroke model. The aim of our study is to investigate the consistency of lesion volumes detected by MK in acute stage and eventual infarction volumes, and the trend of MK value from 12 hours to 7 days after stroke onset by means of the fast DKI protocols combined with simultaneous multi-slice (SMS) technique.


We performed a retrospective review of DKI data from ten acute ischemic stroke patients (6 males, average age 61.7years) at three time points: 12hr, 7days, and 1month after stroke onset. Moreover, data from another forty ischemic stroke patients (28 males, average age 62.8 years) were acquired after onset at different times in the range of 12 hours to 7 days. All data were collected on a MAGNETOM Tim Trio 3.0 T MR scanner (Siemens Healthcare, Erlangen, Germany) with 32-channel head coil. The parameters are as follows: axial T1-weighted, T2-weighted images, fluid-attenuated inversion recovery (FLAIR) were acquired. DWI: TR/TE=3000/80 ms, 23 slices, distance factor=24%, FOV=230 ×230 mm2, b-values of 0, 1000 s/mm2. For the fast DKI protocols with three b-values: 0, 1000, and 2500 s/mm2. One image was obtained with b = 0, three images were obtained with b = 1000 s/mm2, 9 images were obtained with b = 2500 s/mm2. TR/TE=3000/113 ms, 23 slices, distance factor=24%, FOV=210 ×210 mm2, three averages, total acquisition time: 1min57s. Lesion volumes of MK, MD, ADC, DWI images and FLAIR images in one month after stroke were measured by two radiologists using manually drawn ROIs. The volume mismatch percentage was defined as (ROI_diffusion-ROI_T2FLAIR)/ROI_T2FLAIR. A paired t-test was used to test if the MK map matched better with the T2-FLAIR than others. We used rMK, rMD, rADC represent the time-variant of the MK, MD and ADC values which was defined as (ROI_affected side-ROI_ contralateral side)/ROI_ contralateral side.


Fig.1 shows the DWI1000, FLAIR, MD and MK maps of an acute patient within six hours of stroke onset. Notice the distinct ischemic lesion which displays hypo-intensity in the DWI1000 and MK maps. MD maps shows hypo-intensity while the signal on FLAIR maps remains normal. Table.1 shows the voxel size of each lesion detected on MK and routine DWI in acute phase. The lesion size detected by FLAIR was measured on the image acquired one month after stroke onset. Using the FLAIR images as reference, the percentage of lesion size differences of MK is 8.96%±12.17 while DWI is 25.57%±17.30 compared to FLAIR. The lesion volume measured on MK are consistent with the lesion volume measured on FLAIR (p=0.003). Fig.2 shows the time-variant of the rMK, rMD and rADC values from 12 hours to 7 days after stroke onset. We can recognize that the value of rMK is markedly increased 24 hours after stroke, and does not further increase. The tendency of both rMD and rADC value is much the same (p>0.05). The value of both rMD and rADC slightly increases during the 24 hours after stroke, and decreases slightly after 24 hours.


The routine DKI protocol for brain scan needs at least 3 b-values (e.g. 0, 1000, 2000 s/mm2) and more than 20 directions for high b-values, in total 41 directions. The fast DKI protocol only needs 13 directions and can shorter the scan time to 1/3 of the standard protocol length. When combined with SMS technique, the acquisition time is less than 2min (with three averages). A single average protocol could finish the scan in 1min. This protocol shows the same performance in detecting eventual infarction volume in acute stage compared with previous results.


The lesion size observed on MK is consistent with eventual volume. The most significant difference between rMK and rMD occurred 24 hours after stroke onset. Early application of the fast DKI to detect anomalous range of MK is beneficial to predict the range of the eventual infarction and provide reliable clinical imageology.


No acknowledgement found.


1. Hansen B, Lund TE, Sangill R, Jespersen SN. Experimentally and computationally fast method for estimation of a mean kurtosis. Magn. Reson. Med. 2013; 69(6): 1754–1760.

2. Sun, P.Z., et al., Validation of fast diffusion kurtosis MRI for imaging acute ischemia in a rodent model of stroke. NMR in Biomedicine, 2014. 27(11): p. 1413-1418.


Figure 1. The MK(a), MD(b), DWI1000(c), and FLAIR(d) maps of an acute patient within six hours of stroke onset.

Table1. The voxel volume of MK, DWI ROI lesions in acute phase, and the FLAIR ROI lesions one month after each patients’ stroke.

Figure 2. The time variants of rMK, rMD and rADC value from 12 hours to 7 days after stroke onset.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)