Progressive Brain Atrophy on Serial Mri in Dementia with Lewy Bodies, Ad and Vascular Dementia

1

J.T. O'Brien

Methods

All subjects who had all taken part in a cross-sectional MRI study 4 and who were willing and able to attend for a repeat scan 12 months later were included. In all, 28 subjects over the age of 60 years who fulfilled DSM-IV criteria for dementia were recruited from a community-dwelling population of patients with an informant in regular contact. Twenty age-matched controls were recruited from among spouses and friends of dementia subjects. Cognitive function was measured using the Mini Mental State Examination (MMSE) and stage of dementia with the clinical dementia rating scale (CDR).

Diagnoses of AD, VaD, and DLB were made in accordance with NINCDS/ADRDA, NINDS/AIREN, and DLB Consensus criteria respectively by consensus agreement between three experienced raters blind to all MRI scan findings (CT findings were used to assess the presence or absence of infarcts and other vascular changes necessary for the application of each of the diagnostic criteria). We included 10 subjects with consensus criteria probable DLB (M:W 8:2, age 74.46.1, MMSE 16.86.0), nine subjects with NINCDS/ADRDA AD (probable n = eight, possible n = 1; M:W 3:6, age 74.35.3, MMSE 16.26.7), and nine subjects with NINDS/AIREN VaD (probable n = six, possible n = three, M:W 8:1, age 76.46.7, MMSE 17.44.0). Twenty control subjects (M:W 10:10, Age 75.84.7, MMSE 28.01.6) were also recruited.

Baseline and follow-up scans were performed on the same 1.0 Tesla Siemens Magnetom Impact Expert MRI Scanner (Siemens Medical, Germany) using an identical set-up and imaging protocol and the same experienced radiographer. T1-weighted 3D MPRAGE (magnetization prepared rapid acquisition gradient echo) turbo flash sagittal sequence was used to acquire whole brain images (TR = 11.4ms, TE = 4.4ms, TI = 400ms, TD = 50ms, matrix 256x256, slice thickness = 1mm). Standard head positioning was used. Images were transferred to a Sun Ultra 30 workstation (Sun Microsystems inc. California, USA) and analysed using MIDAS software 6. All analysis was performed by the same operator who was not aware of the patient's name or diagnosis. Following segmentation and registration, quantification of atrophy rates was performed by measuring the change in the brain/CSF boundary shift between scans as previously described 6, 7.

MRI Analysis Technique

A three-stage procedure was adopted used to obtain rates of cerebral atrophy from serial MRI scans:

i)Segmentation:

For both baseline and repeat scans a semi-automated seed intensity-threshold based processing routine was used to delineate regions that were “brain” and “non-brain” in each image.

ii)Registration:

3D rotations, translations and linear scaling were applied to the repeat scan to eliminate any differences in the patient position and image size from the first scan to the second. Registration was performed using automated image registration routines comparing only voxels delineated as “brain” in both images.

iii)Quantification:

A “difference map” was created by subtracting the registered repeat image from the baseline image, allowing small changes in brain structure over time to be clearly seen. A precise, robust and well-validated method “the brain boundary shift integral” (BBSI) was then applied to quantify the global atrophy that has occurred. The change in brain volume is determined from the intensity changes of registered pairs of voxels in the vicinity of brain-CSF boundaries. This is equivalent to integrating the positional shifts of all such boundaries in the brain. Applied in three dimensions, the output of this process was expressed as value of brain tissue loss, expressed both as ml or, by dividing by total baseline brain volume, as per cent of total brain volume lost per year. Reproducibility of the technique, assessed by resegmenting and reprocessing scan pairs, produced repeat measurements with a standard deviation of 0.3ml (0.03%). Overall accuracy was assessed by rescanning a single subject 6 times during one day, with repositioning between scans. The standard deviation for measurement of brain volume between the 15 possible scan pairs was 2ml or 0.17%. This indicates the lower limit of measurable change in brain volume.