The use of magnetic resonance imaging (MRI) in veterinary medicine allows the ante-mortem assessment of the brain at a level of detail not previously possible. However, a prerequisite for accurate interpretation is an understanding of what represents normal. Nowhere is this more pertinent than the canine brain ventricular system where a spectrum of normal size variations occur.

Accurate determination of brain and ventricle volumes based on MR images is possible using commercial software; however this process is laborious and not suited to clinical settings. In order to address this, a variety of methods have been proposed to estimate brain and lateral ventricle volumes, but these have not been validated within and between different breeds. The study aim was therefore to development a reliable technique for rapid accurate determination of canine brain and lateral ventricle volume, irrespective of breed or brain disease.

Twenty dogs representing a range of breeds, with no history of intracranial neurological disease and with complete MRI brain studies were included. Dogs with brain disease were represented by 8 Yorkshire Terriers with presumptive granulomatous meningoencephalitis. MR imaging was performed using a 1.5 Tesla scanner. True brain volume and left and right lateral ventricle volumes were determined by computer software analysis. The following additional measurements were obtained for the brain and ventricles: height, width and transverse area at the level of the inter-thalamic adhesion, length and (for the brain only) midline sagittal area. Formulae representing previously published techniques and novel formulae were devised to predict brain volume and lateral ventricle volume.

Regression lines were fitted using true volume as outcome and various estimates as explanatory variables. The results for brain volume determination revealed an adjusted r2 >90% for two variables: midline sagittal brain area (adjusted r2=93.4%) and midline sagittal brain area * brain width (adjusted r2=92.6%). The results for lateral ventricle volume determination revealed an adjusted r2 >95% for two variables: lateral ventricle width * height * length (adjusted r2=95.8%) and transverse area at the level of the inter-thalamic adhesion * length (adjusted r2=95.8%).

In conclusion rapid and accurate clinical determination of brain and ventricle volume is possible by use of the regression equations mentioned here. The correlation coefficients associated with some previously used techniques, in particular for brain volume calculation, were substantially lower than the techniques reported here.