To investigate the apparent venous calibre changes in susceptibility weighted images (SWI) during the first hour of hypoxia, and to propose a physical cause for these changes.In previous studies the apparent increase in venous calibre seen on SWI between normoxia and at 3 hours of hypoxia, have been interpreted as a marked increase in venous diameter [1]. In this study the changes in vessel volume during the first hour of hypoxia are quantified.

Five subjects were imaged using SWI (TR/TE/α/FoV/Matrix/sw=42.1ms/24.7ms/15o/220x198mm/ 320x224/1.3mm) and a 3D anatomical sequence (TR/TE/TI/α/FoV/Matrix/sw=7.8ms/3.0ms/ 400ms/12o/260x195mm/256x256/1.3mm) at 3T (GE Signa HDxt, Milwaukee USA), at normoxia and then during hypoxia. The arterial oxygen saturation (SpO₂) was measured throughout using pulse oximetry. A hypoxic generator (Everest Summit Hypoxic Generator, Hypoxic Systems, New York, NY, USA) was connected to extended MRI compatible tubing and a tight fitting mask that enabled the subjects to remain hypoxic during image acquisition. Once saturation levels of 70% were induced, usually after 8 to 10 minutes of hypoxia, subjects were maintained at 70% SpO₂ .

Two subjects were scanned at 0, 30 and 60 minutes. The remaining 3 subjects were imaged at normoxia, continuously during the first 12 minutes of hypoxia and at 30 minutes and 60 minutes of hypoxia.

The following steps were carried out to numerically quantify the deep venous volume as the number of voxels within a defined low signal intensity range in the SWI images. First the SWI images were registered with the 3D anatomical images and warped into standard brain space using SPM8 (Wellcome Trust Centre for Neuroimaging, University College, London, UK). A mask was then applied to the warped and registered SWI images to remove the cerebral spinal fluid and skull. Finally, an in-house MatLab program (Mathworks Inc., Cambridge, UK) applied a linear intensity normalisation correction, and summed the number of dark voxels within a range identified as representing the signal intensity of the venous vessels by a radiology specialist registrar (RS).

All 3 subjects imaged during the first 12 minutes of hypoxia showed an immediate visual increase in the apparent venous calibre on the hypoxic SWI images (figure 1).

Figure 2 shows the number of ‘dark’ voxels (within the identified intensity range) for each volunteer at each time point. For all subjects there was a step increase in the number of dark voxels with hypoxia. This effect occurred within the first few minutes of hypoxia, and was maintained up to the 60 minutes.

The increase in the number of dark voxels occurs immediately after the induction of hypoxia. In the first few minutes of hypoxia as the blood oxygen saturation level decreased, the number of dark voxels increased.

We believe the increase in the number of dark voxels occured too rapidly after the induction of hypoxia to be entirely due to an increase in vessel volume. We believe the vessels appear dilated because of the greater magnetic susceptibility of deoxyhaemoglobin compared to oxyhaemoglobin; known as the BOLD effect [2].

So, for future studies, changes in venous volumes should not be assessed by comparing baseline (normoxia) with subsequent hypoxia time points. Rather the first SWI, after at least 10 minutes of hypoxia, should be regarded as representing the baseline venous volume.