First described by Hakim and Adams in 1965, normal pressure hydrocephalus (NPH) refers to a clinical entity consisting of the triad of gait disturbance, dementia, and incontinence, coupled with the laboratory findings of normal cerebrospinal fluid (CSF) pressures and radiographic findings of ventriculomegaly.  Although NPH is a relatively rare cause of dementia, identifying NPH is important because it is one of the few treatable entities. NPH is one of the reasons that all dementia patients should have neuroimaging with either CT scanning or MRI as part of their workup. [2, 3]
The images below are examples of normal pressure hydrocephalus (NPH) from computed tomography (CT) scanning and magnetic resonance imaging (MRI), respectively.
MRI of the brain is the preferred radiologic examination for the diagnosis of NPH, especially with T2-weighted images. CT scanning of the brain is useful if MRI is unavailable. Both radiologic techniques require clinical correlation. The primary role of CT scanning and MRI is to assess for hydrocephalus with ventriculosulcal disproportion. This observation is a subjective assessment, and in patients with some sulcal widening or only minimal ventriculomegaly, the studies may not be sensitive or specific.
Plain radiographs, in the form of pneumoencephalographs, have been replaced by CT and MRI scans for the diagnosis of hydrocephalus and are now of only historical interest. Pneumoencephalography was used to demonstrate nonobstructive hydrocephalus. Intrathecally introduced air (via lumbar puncture) was found, on radiographs, within the enlarged lateral ventricles and not in the subarachnoid convexities.
In patients with NPH, CT scans demonstrate hydrocephalus, with ventriculomegaly that is out of proportion to sulcal atrophy. This so-called ventriculosulcal disproportion differentiates NPH from ex vacuo ventriculomegaly, in which sulcal atrophy should also be present. CT scans depicting NPH are presented below.
In NPH, ventriculomegaly is prominent in all 3 horns of the lateral ventricles and in the third ventricle, with relative sparing of the fourth ventricle.
Frontal and occipital periventricular hypoattenuating areas, which may represent transependymal CSF flow, may be noted in NPH, but this sign is infrequent and often may represent periventricular leukoencephalopathy of microangiopathic disease.
Another finding possibly associated with NPH is corpus callosal thinning, although this finding is nonspecific and can be associated with many other conditions.
Degree of confidence
CT scanning alone cannot be used to make a diagnosis of NPH, since the clinical picture and CSF pressures also are necessary in diagnosis. With an appropriate clinical picture and ventriculosulcal disproportion demonstrated on either CT or MRI scans, 50-70% of patients are likely to respond favorably to a CSF-shunting procedure.
In the diagnosis of NPH, the exact percentage of false-positive and false-negative CT-scan findings is unknown. This is partially because NPH remains an incompletely understood entity, and no criterion standard test exists with which to make an unequivocal diagnosis. Assessing for the ability to predict response to surgery seems more appropriate. Unfortunately, individual patient response to CSF shunting in NPH is variable, and the exact percentage of false-positive and false-negative findings of suggestive CT scans is unclear. Disease entities that may mimic the CT-scan findings of NPH include obstructive hydrocephalus, ex vacuo dilatation secondary to cerebral atrophy, and idiopathic arrested hydrocephalus.
Magnetic Resonance Imaging
As in CT scanning, the first abnormality that should be noted on MRI views is ventriculomegaly out of proportion with sulcal atrophy. More specifically, the temporal horns of the lateral ventricles may show dilatation out of proportion with hippocampal atrophy. MRI scans depicting NPH are presented below.
Tsunoda and colleagues used 3-dimensional MRI volume-acquisition techniques to objectively assess ventriculosulcal disproportion.  They measured ventricular volume (VV) and intracranial CSF space volume (ICV) and then calculated the VV/ICV ratio. They found that patients with NPH (n = 16) had significantly higher VV/ICV ratios than did the young control subjects (n = 14), the elderly control subjects (n = 13), and patients with cerebrovascular disease (n = 16). The authors found that 13 of the 16 patients with NPH had a VV/ICV ratio greater than 30%, while no patients in the other groups had ratios higher than 30%. Although the neuroimaging hallmark in NPH is ventriculomegaly out of proportion with sulcal atrophy, volumetric analysis via MRI does not seem to help predict patient response to CSF shunting. 
MRI provides additional physiologic information on NPH compared with CT scanning, because an estimate of CSF flow often can be made by using T2-weighted images.
On T2-weighted images, regions of moving CSF demonstrate no signal instead of the increased signal observed in slow-moving CSF, similar to the flow effects seen with vascular flow voids.
In patients with NPH, the cerebral aqueduct may demonstrate a pulsatile flow void.
A jet of turbulent CSF flow may be observed distal to the aqueduct in the fourth ventricle. This finding appears as a hypointense or absent signal in the proximal fourth ventricle on proton density– and T2-weighted images, with surrounding CSF appearing isointense on proton density–weighted images or hyperintense on T2-weighted images.
MRI may show transependymal CSF flow in the form of a periventricular high signal on T2-weighted images, primarily anterior to the frontal horns or posterior to the occipital horns of the lateral ventricles. However, as with CT imaging, these periventricular abnormalities may be confused with leukoencephalopathy resulting from microvascular ischemia.
An ongoing issue in the management of NPH is that clinical features and even MRI features in patients with NPH can overlap with patients who have much more common diagnoses, such as Alzheimer disease with ex vacuo dilatation of the ventricles. Further, the treatment of NPH is quite invasive, requiring intracranial procedures such as ventriculoperitoneal shunting. Thus, much research has been devoted to trying to identify factors that can predict response to shunting.
Tullberg and colleagues differentiated between periventricular and deep white matter hyperintensity as seen on T2-weighted images and found that neither was predictive of the outcome of CSF shunting.  Thus, the authors caution that findings compatible with microvascular white matter disease do not predict a poor outcome of CSF shunting.
Jack and coworkers assessed the predictive value of 3 MRI findings with respect to positive response to CSF shunting.  These included CSF flow void sign, periventricular increase signal on T2-weighted images, and corpus callosal thinning. The authors found that only the CSF flow void sign may be predictive of shunt responsiveness and that periventricular signal hyperintensity and corpus callosal morphology are not predictive of positive treatment results.
Bradley and colleagues assessed the predictive value of the presence of a CSF void for shunt responsiveness and found a significant correlation.  However, in a later study, the researchers did not find a statistically significant relationship between responsiveness to CSF shunting and aqueductal flow void score, but they did find that MRI assessment of CSF flow stroke volume was predictive of shunt responsiveness.  Marmarou and colleagues concluded that "neither MRI CSF flow void sign nor quantitative CSF flow velocity seems to have significant diagnostic value," and they questioned whether stroke volume may have some benefit.  However, Kahlon suggested that cine phase-contrast MRI measurements of stroke volume in the cerebral aqueduct are not useful in predicting patient response to CSF shunt surgery. 
Tullberg and coworkers found that the presence of periventricular hyperintensity on T2-weighted images, which usually is considered to be evidence of transependymal CSF flow, is not predictive of a good outcome to shunt surgery. 
Studies by Kizu and colleagues using proton chemical shift imaging have suggested that intraventricular lactate measurements may be useful in discriminating patients with NPH from those with other forms of dementia.  In the study, all 9 patients with clinically diagnosed NPH exhibited ventricular lactate peaks by way of proton chemical shift imaging. No lactate peaks were found in the 5 control subjects or in the 6 patients with other diagnosed dementias, including Alzheimer disease (4), Pick disease (1), and frontotemporal dementia (1).
Kazui and colleagues reviewed 71 NPH patients who had undergone surgery.  They did not find that any neuroimaging studies predicted the disappearance of symptoms after shunt surgery. McGirt et al found that corpus callosum distension had some predictive value.  Though not imaging parameters, the former group found that younger age was a predictor of gait improvement and the later group found that gait as the primary symptom was predictive.
A CSF hydrodynamics study,  which is not a standard MRI tool used in clinical practice, looked at a cohort of 20 patients deemed to be shunt responsive (n=14) versus not (n=6), using a threshold mean velocity of CSF through the aqueduct of Sylvius greater than 26 mm/second. To predict responsiveness, they found a sensitivity of 50%, specificity of 83.3%, positive predictive value of 87.5%, and accuracy 70%. Thus, this study of patients treated from 2006-2011 also stresses the ongoing difficulty in identifying patients who might benefit from surgery; 6 of 20 patients did not benefit. Vanneste’s article from 1994  asks: Three decades of normal pressure hydrocephalus: are we wiser now? This study adds another decade without much improvement in response rates.
In a study of 108 patients with idiopathic normal pressure hydrocephalus who underwent preoperative MRI, clinical evaluation 12 months after surgery showed that a small callosal angle, wide temporal horns, and occurrence of disproportionately enlarged subarachnoid space hydrocephalus were significant predictors of a positive shunt outcome. 
In a study of CSF pressure gradients in patients with idiopathic normal hydrocephalus versus gradients in healthy controls, 4-dimensional phase-contrast MRI showed that the pressure gradients of patients with normal pressure hydrocephalus was 3.2 times greater than that in controls. 
Degree of confidence
Degree of confidence in MRI in helping to diagnose NPH or, more importantly, in helping to predict a positive result with neurosurgical CSF shunting is unknown. Positive surgical results are demonstrated in 50-70% of patients with a strong clinical history of NPH and classic NPH findings on magnetic resonance images or CT scans. 
Similar to CT scanning, MRI contributes to the diagnosis of NPH, but no criterion standard test exists with which to accurately assess the occurrence of false-positive and false-negative findings of MRI alone.
Traditionally, isotope cisternography and CT cisternography have been used in NPH to assess for disturbances in CSF dynamics, such as reversal of flow. This investigation is likely to be an unreliable predictor of NPH, despite its historical popularity. [21, 22]
Using single-photon emission computed tomography (SPECT) and statistical brain mapping, Sasaki found regional cerebral blood flow reduction with frontal dominance and severe hypoperfusion around the corpus callosum.  This was consistent with some of the regions of brain dysfunction that clinical assessment has indicated are involved.
Degree of confidence
However, one of the likely difficulties in identifying responsiveness to shunting is that many individuals in the age group can have overlapping diagnoses such as Alzheimer disease and subcortical microangiopathic disease, of which the latter can cause both dementia and gait disturbance identical to that seen in NPH. Incontinence, the remaining features of the NPH triad, is common in both men and women. Thus, recent advances in neuroimaging for other conditions, such as Alzheimer disease, might help establish the probability of these other confounding conditions.
There have been advances in positron emission tomography (PET) scanning in this regard. For example, although clinical outcomes were not a focus of the study, Leinonen et al (2013)  looked at [(18)F]flutemetamol PET scanning in patients with possible NPH. In concept, this could help establish the total burden of amyloid-beta pathology and, hence, the burden of Alzheimer disease that would not be corrected by invasive surgical treatment of NPH such as ventriculoperitoneal shunting.