Question 20 from the second paper of 2022 asked the candidates to list CT and MRI features of severe hypoxic ischaemic encephalopathy. This is probably a topic large enough to require its own chapter. The best peer-reviewed resources for this have been Lopez Soto et al (2020) for prognostication using imaging, Rajajee et al (2023) and Sandroni et al (2020) for neuroprognostication in a broader sense, and Huang and Castillo (2008) as well as Gutierrez et al. (2010) for specific findings.
In summary:
CT features of hypoxic ischaemic encephalopathy:
MRI features of hypoxic ischaemic encephalopathy:
Value of these features in neuroprognostication:
Bilaterality: the damage is global, and therefore bilateral and symmetrical.
Predominantly grey matter injury: Grey matter is more metabolically active and is therefore more readily injured by the loss of metabolic substrate. Apart from the cortical grey matter one can also expect deep structures to be injured, for example the basal ganglia. Gutierrez et al (2010) refer to this as "selective neuronal necrosis" and "selective vulnerability" implying that glial cells are relatively spared in this condition (whereas in stroke the white matter also dies). However these terms do not seem common and exam candidates are advised against using them too liberally.
"Phylogenetic susceptibility" is another infrequently mentioned sign from Gutierrez et al (2010); the name refers to the observation that more "ancient" areas of the brain are less vulnerable, whereas the newer and more sophisticated structures (neocortex, etc) are more easily damaged. The upshot of this is that one can often see severe damage to the caudate and putamen with relatively spared thalami. Borrowing some excellent Radiopedia teaching images, the reader can have little a thalami as a reminder for them as to where these structures are placed, in case neuroanatomy remains elusive for them even at the later stages of their ICU training:
Distribution of injury along watershed areas is sometimes mentioned, as some parts of the brain are more distal to the larger vessels than others, and are supposedly more damaged by the loss of perfusion, whereas theoretically more proximal areas are somehow perfused better even if blood flow completely stops. The reader may rightly complain that this defies logic (because surely the whole brain is equally underperfused in cardiac arrest), and this would be largely correct - but most of the time some kind of poor perfusion is maintained through CPR, and it would be expected to deliver at least some blood to the better-plumbed regions of the brain. Ergo, you may see watershed areas affected more than centrally supplied regions. These watersheds occur at the boundaries between cerebral vascular territories. Unlike the wedge-shaped cortical infarcts that you would expect from n hypoxic ischaemic encephalopathy the infarcted border zones are also often internal, and are generally described as "parasagittal" or "parafalcine" because of this specific appearance:
The CT appearance is not as spectacular as MRI, which lights up impressively, but sometimes the decreased attenuation is clearly visible. This image from Radiopedia is a case with a unilateral watershed infarct between the ACA and MCA territories, which is included here mainly because it is valuable to see the affected side next to the unaffected (and also it was too hard to find an example with bilateral lesions).
Loss of grey-white differentiation, a loss of the distinction between grey and white matter densities, is a finding that suggests diffuse cerebral oedema. Specifically this is a finding associated with cytotoxic oedema, like the sort that you typically get with hypoxia and ischaemia (whereas vasogenic oedema usually results in the opposite effect, an exaggeration of grey-white differentiation). Here is an example from Radiopedia, with a normal CT next to it for contrast
Yes, it is in fact possible to quantify this objectively. Torbey et al (2000) determined that anything less than about 18-20% difference in density (measured in Houndsfield units at the basal ganglia level) was enough to predict a poor outcome. For Scheel et al (2013), a difference of around 16% predicted poor outcome with 100% specificity and 38% sensitivity. However, in practice, nobody does this (for one each scanner produces a slightly different set of numbers and Houndsfield units cannot be compared easily from one machine to another). Fortunately "eyeballing" the scan also seems to have a specificity of close to 100% (Lang et al, 2022).
Reversal of grey and white attenuation: you read that right, this is a reversal of the normal attenuation which increases the apparent"brightness" of the white matter as compared to the (normally brighter) grey. Han et al reported this way back in 1990, which means the pictures in their article are pretty terrible, but here they are anyway:
To be precise, this is an inversion of the normal gray/white matter density ratio in Hounsfield units, and it was found in patients who failed to awaken after cardiac resuscitation. Is this all just due to the relative decrease in the density of the white matter, i.e. is cerebral oedema all there is to it? Probably not, because this is not clever "windowing": the white matter genuinely becomes more electron-dense. According to Radiopedia, "this finding is due to the distension of deep medullary veins secondary to partial obstruction of venous outflow", i.e. the extra density comes from the increased blood content of the white matter structures.
White cerebellum sign is presumably the extension of the same venous drainage related principle, though many authors seem to describe it in terms of relative normodensity when compared to the infarcted supratentorial structures (i.e. "it looks white because the rest of the brain looks so oedematous"). Here is an example from Radiopedia.
Cortical laminar necrosis is the appearance of a linear hyperdensity separating the laminae of the cerebral cortex. According to Gutierrez et al (2010), this is because cortical laminar Layer 3 (external pyramidal neurons) is the most vulnerable, and HIE can produce a selective injury there, creating a band of necrosis between the laminae.
Parada-Duarte et al (2019) presents some excellent images of what this is supposed to look like on CT, which have been reproduced here with no permission whatsoever:
The reason we have resorted to combing through case reports here is that, though this is listed here as one of the CT findings, in fact most of the work on this radiological sign comes from MRI literature.
Effacement of CSF spaces (sulci, ventricles) is the result of cerebral oedema, and is usually seen along with the loss of grey-white differentiation.
Tonsillar herniation is the natural extension of cerebral oedema and can be listed here as one of the late radiological findings in severe hypoxic ischaemic brain injury. Speaking of late findings:
Pseudosubarachnoid haemorrhage is an end-stage manifestation of severe global cerebral oedema, resulting from the congestion and dilatation of superficial venous structures around the brain. These engorged vessels become thrombosed and the high density clots fill the CSF spaces and give the appearance of blood products in the subarachnoid space:
Unlike "real" subarachnoid haemorrhage, these dense areas are usually less dense (30-40 Houndsfield units, as compared to around 60 units in "proper" aneurysmal subarachnoid). Yuzawa et al (2008) and Given et al (2003) are available if the reader would like to find out more about the meaning of this horrible finding. In short, it's much worse prognostically than a real SAH.
In summary, a "poor CT" can be used to identify people who are going to have a poor outcome, but a "good CT" cannot be used to identify people who are going to do well neurologically. Consider:
Early hyperintensity on diffusion-weighted imaging (DWI): The earliest MRI changes are hyperintensity of signal in basal ganglia and cortex, particularly in the watershed areas. Oedema on DWI looks "bright" because the diffusion of water molecules is restricted in oedematous tissue (they are stuck inside cells and their movement is impeded). This can be happening even while the cells themselves are reasonably intact, i.e. DWI can be abnormal even when T1-weighted and T2-weighted imaging looks completely fine. Here's an example of severe hypoxic ischaemic encephalopathy on DWI next to a normal specimen; the watershed areas and thalami are lit up because of acute cerebral oedema.
The abnormal image here has come from White et al (2013), where the middle-aged female patient had experienced some hypotension during liver surgery; a classical reason to develop watershed infarcts. In fact DWI patterns can vary significantly from this "textbook" appearance, as illustrated in these excellent images from Muttikkal et al (2013). Everything is possible, ranging from a diffuse pattern of involvement in (a), or where in (b) there is grey matter damage sparing the "perirolandic" region (paracentral, surrounding the Rolandic fissure), or diffuse white matter involvement in (g) and isolated basal ganglia damage in (h).
These changes on DWI are an early feature, and disappear within a few days. The exact timeframe is difficult to predict and probably very individual but six days seems to be the accepted window for maximum accuracy. In a meta-analysis by Wei et al (2019), the accuracy of DWI improved significantly when they removed the data of that one study that scanned the patients on days 7-18 following their hypoxic brain injury.
Reduced apparent diffusion coefficient (ADC) map: resisting the urge to lead the reader away into the dark woods of MRI physics, it will suffice to remind them that ADC is a map of automatically calculated diffusion restriction which is derived from DWI data, and so could be viewed basically as just another way of reading DWI. Contrary to this, it can demonstrate abnormality even when raw unprocessed DWI may not (Tchofo et al, 2005), and can be used to quantify the lesions using measured ADC values. As a measure of diffusion, ADC represents restriction as decreased signal (i.e. less diffusion = darker pixels), which is the opposite of DWI. Here is an irrelevant but highly illustrative series of images from Schlaug et al (1997), showing the evolution of a lacunar stroke over time. The top row is ADC and the bottom row is DWI:
So, damaged tissue is dark; and so one should therefore expect a lot of dark looking brain on ADC images of hypoxic ischaemic encephalopathy, mirroring the increased bright signal intensity seen on DWI. This is certainly the pattern observed by Oren et al (2019), who have an excellent series of images at different stages of severity. For the purposes of learning, they were remixed below to demonstrate what a normal ADC looks like next to an ADC from someone with severe hypoxic brain injury:
So, another MRI modality that looks at the same thing as DWI, but from a different angle. Is this really necessary, would be a fair question from the reader confused by their imaging options. Does it really make a difference? Why not just eyeball the MRI and report it as "severe HIE"? Well, in fact there is considerable advantage from measuring the ADC values, as it allows thresholds to be determined, making the process of prognostication more scientific. Consider, when Hirsch et al (2020) set thresholds at an ADC of less than 650 ×10−6 mm2/s over 10% of the imaged brain tissue, the positive predictive value ended up being 0.95 (sensitivity 63%, specificity 96%). This is sufficiently consistent across studies to merit some weak recommendations from the 2023 NSC guidelines (Rajajee et al, 2023).
T2-weighted and fluid-attenuated inversion–recovery (FLAIR) are late to change. FLAIR is a special sequence that eliminates the signal from the CSF in a T2-weighted image. As the result, CSF spaces appear dark, but other sorts of brain tissue oedema appears white, as in a normal T2 image. However, grey matter is also lighter in T2 and FLAIR, making it harder to spot oedema (especially considering that it is usually widespread and bilateral, making it harder to compare it to normal tissue). Here is a normal and a HIE example from Radiopedia, where the brain injury occurred about five days before the scan:
Prominent increase in signal intensity in affected regions (especially the basal ganglia in the image above) and the cortical thickening are main clues that something has happened, but otherwise the findings are a lot more subtle than in DWI.
Changes on T1-weighted images are very late, taking as long as two weeks to develop. These changes are usually related to cortical laminar necrosis and are thought to represent some kind of protein deposition in the brain tissue. Here, some very early MRI images from Takahashi et al (1993) are presented - the helpful arrows point to the areas where increased cortical laminar signal intensity identifies necrotic brain tissue.
Lopez Soto, Carmen, et al. "Imaging for neuroprognostication after cardiac arrest: systematic review and meta-analysis." Neurocritical Care 32 (2020): 206-216.
Rajajee, Venkatakrishna, et al. "Guidelines for Neuroprognostication in Comatose Adult Survivors of Cardiac Arrest." Neurocritical Care (2023): 1-31.
Bertoni, M., et al. "Neuroimaging assessment of hypoxic ischemic brain injury of the adult with perfusion computed tomography." Journal of the Neurological Sciences 381 (2017): 71-72.
Sandroni, Claudio, et al. "Prognostication in comatose survivors of cardiac arrest: an advisory statement from the European Resuscitation Council and the European Society of Intensive Care Medicine." Intensive care medicine 40 (2014): 1816-1831.
Huang, Benjamin Y., and Mauricio Castillo. "Hypoxic-ischemic brain injury: imaging findings from birth to adulthood." Radiographics 28.2 (2008): 417-439.
Gutierrez, Leonardo Guilhermino, et al. "CT and MR in non-neonatal hypoxic–ischemic encephalopathy: radiological findings with pathophysiological correlations." Neuroradiology 52 (2010): 949-976.
Parada-Duarte, Óscar Andrés, et al. "Cortical Laminar Necrosis in Computed Tomography Scan: A Case Report." Rev. Colomb. Radiol. 2019; 30(4): 5239-41
Muttikkal, Thomas J. Eluvathingal, and Max Wintermark. "MRI patterns of global hypoxic-ischemic injury in adults." Journal of Neuroradiology 40.3 (2013): 164-171.
Han, B. Kim, et al. "Reversal sign on CT: effect of anoxic/ischemic cerebral injury in children." AJR. American journal of roentgenology 154.2 (1990): 361-368.
Krishnan, Prasad, and Siddhartha Roy Chowdhury. "“White cerebellum” sign-A dark prognosticator." Journal of Neurosciences in Rural Practice 5.04 (2014): 433-433.
Chakraborty, Santanu, et al. "Diffuse ischemia in noncontrast computed tomography predicts outcome in patients in intensive care unit." Canadian Association of Radiologists Journal 63.2 (2012): 129-134.
Torbey, Michel T., et al. "Quantitative analysis of the loss of distinction between gray and white matter in comatose patients after cardiac arrest." Stroke 31.9 (2000): 2163-2167.
Scheel, Michael, et al. "The prognostic value of gray-white-matter ratio in cardiac arrest patients treated with hypothermia." Scandinavian journal of trauma, resuscitation and emergency medicine 21 (2013): 1-7.
Lang, Margareta, et al. "A Pilot Study of Methods for Assessment of Hypoxic-Ischaemic Encephalopathy on Head Computed Tomography after Cardiac Arrest."
Yuzawa, Hironao, et al. "Pseudo-subarachnoid hemorrhage found in patients with postresuscitation encephalopathy: characteristics of CT findings and clinical importance." American journal of neuroradiology 29.8 (2008): 1544-1549.
Maurya, V. K., et al. "Hypoxic–ischemic brain injury in an adult: magnetic resonance imaging findings." Medical Journal, Armed Forces India 72.1 (2016): 75.
White, Matthew L., et al. "Anatomical patterns and correlated MRI findings of non-perinatal hypoxic–ischaemic encephalopathy." The British journal of radiology 86.1021 (2013): 20120464-20120464.
Wei, Ruili, et al. "Prediction of poor outcome after hypoxic-ischemic brain injury by diffusion-weighted imaging: A systematic review and meta-analysis." Plos one 14.12 (2019): e0226295.
Velly, Lionel, et al. "Use of brain diffusion tensor imaging for the prediction of long-term neurological outcomes in patients after cardiac arrest: a multicentre, international, prospective, observational, cohort study." The Lancet Neurology 17.4 (2018): 317-326.
Tchofo, P. Jissendi, et al. "Apparent diffusion coefficient (ADC) and magnetization transfer ratio (MTR) in pediatric hypoxic-ischemic brain injury." Journal of neuroradiology 32.1 (2005): 10-19.
Schlaug, G., et al. "Time course of the apparent diffusion coefficient (ADC) abnormality in human stroke." Neurology 49.1 (1997): 113-119.
Oren, Nisa Cem, et al. "Brain diffusion imaging findings may predict clinical outcome after cardiac arrest." Journal of Neuroimaging 29.4 (2019): 540-547.
Hirsch, Karen G., et al. "Prognostic value of diffusion-weighted MRI for post-cardiac arrest coma." Neurology 94.16 (2020): e1684-e1692.
Takahashi, Shoki, et al. "Hypoxic brain damage: cortical laminar necrosis and delayed changes in white matter at sequential MR imaging." Radiology 189.2 (1993): 449-456.
Youn, Chun Song, et al. "External validation of the 2020 ERC/ESICM prognostication strategy algorithm after cardiac arrest." Critical Care 26.1 (2022): 1-9.