This is the first study to investigate the association between the presence and severity of CCSVI and venous vasculature in the brain parenchyma on SWI venography in patients with MS and in HC. There are three main findings in this study: 1) first, by examining only brain parenchyma VVV, we found that MS patients are compromised compared to HC. The novelty of our SWI approach using pre- and post-contrast SWI venography experiment is that its suggests that the reduced venous visibility in brain parenchyma of MS patients is not only the result of reduced metabolism, but may be affected by possible morphological changes in venous vasculature. This was not shown previously by using post-contrast SWI venography and only one previous study  showed similar qualitative (but not quantitative) pre-contrast SWI venography findings; 2) second, by quantitatively associating VVV with the presence or absence of CCSVI, we found significantly increased pre-contrast DFV, decreased ATVV and ATVV of veins with a diameter < .3 mm, and a trend for decreased VIF in subjects presenting with CCSVI; 3) third, MS patients with higher number of venous stenoses, indicative of CCSVI severity, showed significantly decreased VVV in the brain parenchyma on both pre- and post-contrast SWI. The pathogenesis of these findings has to be further investigated, but they suggest that decreased venous vasculature in brain parenchyma of MS patients is strongly related to presence and severity of CCSVI. These findings are important for better understanding of the MS pathogenesis, and the current findings support results of two previous pilot hemodynamic imaging studies (using PWI and CSF flow measurement) that showed that drainage problems in patients with MS are associated with presence and severity of CCSVI [18, 19]. By quantitatively measuring the most important drainage pathway from brain parenchyma to the periphery, which consists of the veins themselves, we are confirming and extending the validity of these previous preliminary findings.
When MS patients' subtypes were compared, the association between the severity of CCSVI and altered pre- and post-contrast SWI VVV was stronger in RR than in SPMS patients, and RRMS patients also presented with a trend for more altered pre- and post-contrast SWI VVV compared to the SP group. The significance of these findings is unclear and should be further explored; however, it could indicate that VVV changes may occur very early on in the MS disease process.
This study did not explore the relationship between SWI-VVV and markers of inflammation and demyelination in patients with MS from the earliest clinical stages. It could be that the reduced venous visibility in brain parenchyma of MS patients, and consequently decreased venous vasculature, is secondary to the presence and severity of inflammatory and demyelinating lesions that are accumulating over time in MS patients. Further studies using conventional and non-conventional MRI measures are needed to determine how presence and severity of intrinsic lesion damage relates to changes in venous vasculature. Therefore, the pathogenetic role of our VVV findings should be interpreted with caution.
The automated vein extraction segmentation approach we have developed greatly diminishes the impact of individual operators. The algorithm was based on the vesselness filter described by Sato et al., , as this was proven by the authors to be stable and highly reproducible. The intrinsic advantage of this approach lies in its multi-dimensionality, allowing for the segmentation of vessels of different sizes, the ability to recover continuity of the structures despite higher local noise levels, and most of all the fact that it is not based on absolute signal intensity (zero order information) but rather on intensity gradients surrounding each voxel (second order information), giving sensitivity to the shape of the structures. The vesselness filter is integrated into the ITK library (Insight Segmentation and Registration Toolkit, http://www.itk.org/) and has been applied to vessel segmentation on various imaging techniques, including MRI, computerized tomography and optical computerized tomography.
The scan-rescan experiment data showed that SWI venography VVV indices we measured are highly reproducible (Table 2). As expected, post-contrast SWI VVV measures showed slightly better reproducibility results than pre-contrast measures. Of all different vein diameter ATVV measurements we performed, the veins with larger diameter (> .3 mm) showed better reproducibility than those < .3 mm. This is expected, as the segmentation outcome of veins < .3 mm will be most influenced by image quality, movement-related partial volume effects and image processing steps. These may be related to different parameters, including image contrast-to-noise ratio, unwrapping of phase images, filter characteristics (tuning parameters, threshold) of the Hessian filter, motion artifacts, and change in SNR. However, we inspected all segmentation output of the images visually. In addition, our reproducibility analysis did not show that the potential change of these imaging parameters affected the reproducibility of results over one week. On the other hand, VIF and DFV displayed a reduced reproducibility. For the VIF, the reason most likely lies in the intrinsic uncertainty in the estimation of the intra-cranial volume, which is used to normalize the VVV across subjects. For the DFV instead, reduced reproducibility is the effect of the enhanced vein visibility which characterizes the mIP maps.
SWI is very sensitive in detecting signals from substances with magnetic susceptibilities that are different from that of their neighbors. Consequently, SWI is able to detect tissue iron in the form of ferritin, hemosiderin and deoxyhemoglobin, [25, 33, 34] and is sensitive to the visualization of small veins in the brain . SWI venography allows detailed visualization of cerebral veins in the brain parenchyma without the use of an exogenous contrast agent . This is possible by exploiting the difference in magnetic susceptibility properties between oxygenated and deoxygenated hemoglobin. The abundance of the paramagnetic deoxyhemoglobin molecule in the venous blood results in increased local magnetic field inhomogeneity, which in turn leads to spin dephasing and signal loss on SWI venography, resulting in decreased VVV . One of the key aims of this study was to address whether the SWI venography VVV differences between MS patients and HC are only a result of decreased oxygen utilization in MS patients (with correspondent decreased levels of venous deoxyhemoglobin), as previously proposed,  or if there were also morphological changes in the small veins that became atrophic and disappeared due to the MS disease process, possibly leading to permanent decrease of signal on SWI. Contrast-enhanced SWI significantly increases the visualization of number and volume of signal hypointensities on SWI venography [28, 29] in T2 lesions and in normal appearing WM (Figure 4),  and may be an additional means of investigating whether SWI venography differences between MS patients and HC are due only to hypometabolic status or whether morphological changes of veins may be taking place in MS patients. 81.4% of the MS patients and 21.2% of HC included in this study underwent both pre- and post-contrast SWI sequence in order to further elucidate this important question. We demonstrated a very similar decrease in brain parenchyma VVV on pre- and post-contrast SWI parameters we examined in MS patients, but significantly increased ATVV and ATVV of veins with a diameter < .3 mm in HC was found, as expected (Table 4). The reduction of vascular visibility on pre-contrast SWI between MS patients and HC was previously observed and attributed to hypometabolic status in brain parenchyma of MS patients . However, the pre- and post-contrast SWI venography experiment performed in the present study further extends understanding of this phenomenon and suggests that the reduced VVV in MS may be a combination of two main effects - reduced metabolism and morphological changes of the venous vasculature. Further cross-sectional and longitudinal studies are needed to better elucidate this phenomenon. Subtraction technique approaches may be useful for detecting whether the loss of signal occurs in the same regions between pre- and post-contrast SWI venography. In addition, region-specific analysis may shed light regarding key areas that are involved. Visual analysis performed in this study suggests that cortical and deep WM/GM regions may be the most affected (Figure 3).
A hypoxia-like condition has been evidenced in patients with MS [35, 36]. It has been shown that hypoperfusion of the brain parenchyma in MS patients may precede disease onset [37, 38]. Abnormal perfusion patterns within normal appearing WM and GM were demonstrated in MS patients [38–41]. Chronic inflammatory events related to local blood congestion and secondary hyperemia of the brain parenchyma are proposed as a cause of these hemodynamic abnormalities detected on perfusion MRI in patients with MS [39, 42, 43]. Whether reduced perfusion of the WM and GM in MS patients is a sign of vascular pathology, decreased metabolic demand  or precipitated hemodynamic changes in the extra-cranial venous pathways  is not clear at this time. We recently reported in a pilot study a significant relationship between the severity of CCSVI and hypoperfusion in the brain parenchyma of 16 MS patients, but not in 8 healthy controls .
One of the key findings in this study is that MS patients presenting with CCSVI (and with increased severity of CCSVI, as measured by VHISS) showed decreased pre- and post-contrast VVV in brain parenchyma on SWI venography. These findings confirm results of earlier studies, [38–42] all of which observed significantly lower perfusion in the brain parenchyma of MS patients compared with controls. It could be hypothesized that decreased venous outflow from the brain parenchyma to the periphery would lead to increased intracranial pressure and subsequent venous stasis, especially of the small vein vasculature. Venous pressure was not measured in the current study; however, a recent study showed no increased intracranial venous pressure in MS patients . Nevertheless, increased venous pressure in the subarachnoid space of MS patients cannot be excluded at this time. Several authors have hypothesized that reduced venous drainage outflow in MS may increase intra-capillary oncotic pressure, which would lead to decreased capillary permeability toward the extra cellular compartment and consequent intra-tissue accumulation of toxic metabolites [12, 45]. In the present study, the association between presence and severity of CCSVI was particularly strong with VVV indices of small veins. We showed that there was a trend for differences between patients and controls for veins with a diameter < .6 mm. Hypoxia arising from stasis in the veins might therefore induce morphological changes which could result in occlusion and atrophy of these veins. Therefore, the most plausible explanation of our findings would be that reduced outflow from small vessels to the periphery--independent of whether it is primary or secondary to CCSVI or inflammatory and demyelinating lesions--leads to destruction of smaller veins and consequent loss of signal visibility on SWI venography. The pathogenesis of this process remains unanswered at this time and should be explored together with dynamic of regional lesion accumulation over time.
This loss of vasculature suggests that rerouting of the intracranial venous blood flow is probably taking place. If CCSVI is secondary to various vascular, infective and inflammatory processes (this hypothesis could explain the presence of CCSVI in HC), then the tendency to be chronic in its development may help explain the temporal dissociation between the loss of the VVV and no development of intracranial hypertension. In that context, hemodynamic compensatory mechanisms may play a key role. One such mechanism could relate to the development of extra-cranial collateral circulation [14, 16] or altered CSF dynamics  that would compensate for altered primary outflow pathways.
Some recent reports have presented evidence against the CCSVI hypothesis in MS using Doppler and MRV assessments [46–51]. The conflicting reports of CCSVI-related venous abnormality prevalence findings between different studies using non-invasive and invasive imaging techniques emphasize the urgent need for better understanding of these anomalies. It is also important to place CCSVI in the context of other known associations in MS, with the most well established MS risk factors and clinical and MRI outcomes. The current study provides important evidence that extra-cranial venous abnormalities may be related to intra-cranial brain parenchyma reduced venous visibility. The dynamic of these findings should be further explored.
There are a number of limitations to the present study. One relates to the use of the proposed CCSVI criteria,  which could be insufficient to adequately describe the cerebral venous outflow due to the lack of assessment of functional data on blood flow velocity and blood volume flow . Evaluation of blood flow velocity and blood volume flow may offer more complete status of the cerebral venous outflow in relation to decreased brain parenchyma SWI VVV in patients with MS. Another limit of the study is related to its cross-sectional design and to the relatively small number of subjects studied, especially of HC and MS patients at the early stage of the disease. Due to the nature of SWI phase imaging and associated "blooming" effects, our measurements of venous volume are inherently relative rather than absolute. Finally, the validation of our VVV method should be performed against either simulations or phantom measurements.