T. Benner et al. Animated 3D visualization of ischemic lesions with diffusion-weighted MR imaging


In the recent years many papers have been presented reviewing the principles of diffusion-weighted imaging (DWI) (1-7) and its application to cerebral ischemia (8-10). DWI provides information about the average diffusion and displacement profiles of particles in a sample in the micrometer range by using diffusion-sensitizing gradients which can either be applied in one gradient direction yielding anisotropic diffusion-weighting or in a more complex scheme in all three gradient directions yielding isotropic diffusion-weighting. The strength of the diffusion-weighting is usually given by the so-called b-value, calculated from the parameters of the diffusion gradients (gradient strength, time between commencement of the two gradient pulses, duration of the gradient pulses). The apparent diffusion coefficient (ADC) - "apparent" because the measured voxel contains a mixture of different diffusion coefficients, e.g. intra- and extracellular - may even be quantified using a series of images with different b-values and the time course of the ADC may reflect the severity of the ischemic damage (1, 6, 7, 9-13). Because diffusion MR imaging is very sensitive to bulk motion - i.e. not only head motion in the millimeter range but also to small micrometer shifts of cerebral structures due to e.g. liquor pulsation - it is most easily done with echo planar imaging (EPI) (14, 15). However, it can also be performed on conventional scanners with sequences which are typically less sensitive to image distortions than single-shot methods (16-18). The motion artifacts often seen at conventional scanners can be corrected using the navigator echo technique (19-21).

Diffusion-weighted imaging was shown to be sensitive shortly after the onset of cerebral ischemia even before conventional PD-, T1- or T2-weighted MR sequences or other imaging modalities like CT show any signs (8-10, 22-28). The ischemic area shows a hyperintense signal compared to healthy brain tissue due to slower diffusional motion most likely related to the development of ischemia induced cytotoxic edema. With increasing b-value the signal intensity of the ischemic area strongly increases compared to the surrounding tissue leading to the so-called "light bulb" effect.

Diffusion MR images are usually viewed like other MR images in a multi-planar or multi-slice manner. From these two-dimensional (2D) images the three-dimensional (3D) spatial extension of the lesion cannot be identified exactly. This is worsened by the strong contrast between ischemic and healthy tissue and the low signal-to-noise ratio of healthy tissue at high b-values. Conventional T1- or T2-weighted images at identical slice positions may be helpful in the localization of the lesion. Nevertheless, the assessment of volume, shape and connectivity of the lesion depends on the experience and the 3D imagination of the physician.

The maximum intensity projection (MIP) is a method which is routinely used for the display of MR angiography images (29-31). From a parallel projection of the original image data to the view point the maximum intensity of all signal intensities on a beam is selected to form the new image. Therefore, the applicability of the MIP depends on the strong contrast between bright vessels and dark surrounding tissue and low signal intensities of other brain structures. Diffusion-weighted images from acute stroke patients exhibit the same properties: strong signal intensities in the ischemic area and low signal intensities in the surrounding tissue.

In this work we apply the maximum intensity projection method to strongly diffusion-weighted images of acute stroke patients. The resulting images are compared in steady state and in cine mode to conventional planar views. The degree to which it is possible to define position and extent of the lesion is assessed. Advantages and disadvantages of the method are discussed and further visualization possibilities stated.

[ Title | Abstract | Introduction | Materials & Methods | Results | Conclusion | References ]