[0:07]Welcome back, everyone. This is Dr. Rybinnik. Imaging is part of our daily professional life in neurology. So, let's give you a crash course in reading it.
[0:20]This video is rated MS for medical students, due to basic concepts, clear language, and case-based examples. Medical student participation is advised.
[0:31]Our objectives for this talk are to introduce a case that will act as the glue keeping the sections together, to introduce an approach to reading imaging, to review major anatomical landmarks so you can find your way around the scans, to assess the degree of symmetry or asymmetry to allow you to spot abnormalities, to review the causes of hyper density and hypodensity on CT, and since we're in the topic of hypodensity, we will introduce the concept of cytotoxic and vasogenic edema. Edema will transition nicely into a discussion of the causes of hyper and hypointensity on various MRI sequences. We will conclude with a review of patterns and locations of enhancement. And summarize. Let's start with a case. It just got real. You are asked to see the following patient. This is a 73-year-old previously healthy woman, with weeks of memory difficulties, 10 days of increasing lethargy, 3 days of urinary incontinence, and now, mild left-sided weakness and rigidity. Head CT was already conveniently done for you by the emergency department team. Radiology resident has passed out. Your neurology resident is running a stroke code. So, it's up to you to diagnose this patient. First things first. It's the clinical history and not imaging that will lead you to the final diagnosis. Imaging can at best help you with a reasonable differential. So, whenever facing the unknown, it's useful to have an approach. First step is to identify the scan, the imaging sequence and the slice, then, look for symmetry, or more appropriately, asymmetry. Next, we need to identify the lesion causing the asymmetry and its density and intensity, decide if contrast enhancement pattern may be useful, and locate the lesion in the extra-axial compartments, outside the brain, or intra-axial compartments, in the brain parenchyma. Put all this data and the blender, puree it, and you get the delicious smoothie that is the final differential diagnosis. Yum. First step, landmark review. And for our landmarks, we'll use the head CT. Head CT without contrast is the most common imaging modality that we get. Contrast can be added to look for breakdown of blood brain barrier, for example with neoplastic or infectious lesions, but that's better seen on MRI. CT can be used to obtain vascular imaging by rapidly tracking a bolus of contrast into the brain, this is called the CT angiogram. By the way, the difference between the CT with contrast and a CT angiogram is essentially the timing of the scan with respect to the contrast injection. CT angiograms are really power to identify vessels, not brain parenchyma. And finally, CT-based perfusion can be used to estimate brain blood flow. But for the purpose of this talk, the only sequence we will focus on is the plain old non-contrast head CT. Here's a typical normal head CT. Axial slice with a cut through the orbits. You can even make out the lenses inside the eyeballs. By convention, the patient is lying in front of you on a table, feet first, like in the anatomy lab or the operating room. So, the nose is pointing up, patient's left is on your right, and the back of the head is on the bottom of the slide. In this image, we're looking a bit higher in the brain. The slice is taken through the upper frontal and parietal lobes. How do I know that? Well, the slide is labeled. A better method is to locate the central sulcus. It makes a curve that looks like the Greek letter Omega. Anything anterior to that sulcus is the frontal lobe, and posterior is the parietal lobe. Fun fact, the Omega-looking bump in the precentral gyrus is also called the hand knob. This is where the motor control of the hand resides. Since we're on the topic of anatomical localization, we probably should quickly introduce the T1 sequence of the MRI. T1 is called the anatomical sequence because it shows white matter as white and gray matter is gray, as they would look in gross pathology. Can you still identify the Omega sign? Yep, here it is on the scan. And if you need extra help, Netter to the rescue.
[5:11]Scanning lower down, now we're at the level of the basal ganglia. Lateral ventricle is in the middle, head of caudate is next to the frontal horn of the lateral ventricle. Thalami are separated by the third ventricle. Head of caudate and thalamus are separated from the lentiform nucleus by the internal capsule. More laterally, you will find the insular cortex covered by the anterior temporal lobe. Now, can you identify those structures on a T1 image? Head of caudate, thalamus, internal capsule, lentiform nucleus, insula, and anterior temporal lobe. Here's an illustration.
[5:57]Moving right along. This is a slice through the Sylvian fissure. Incidentally, the orbits and eyeballs are back. What lives in the Sylvian fissure? Yeah, the middle cerebral artery. Here it is, isodense to the brain parenchyma. It's generally very difficult to distinguish from the nearby brain. But when you can make it out because it's brighter, that's when you know things are bad. More on that later. Here is the Sylvian fissure on T1. As you can see on the localizing image in the upper left corner, T1 and CT slices are not necessarily at the same angle. So, on this T1, you can actually make out the Mickey Mouse-looking midbrain. And once again, the Netter illustration. Speaking of the midbrain, here is the CT axial cut at the level of the midbrain. The midbrain is central here with the CSF-filled basal cistern right behind it. The medial temporal lobe, or uncus, is lateral to the midbrain. Uncus as in uncle herniation. The uncus may herniate onto the midbrain, so, needless to say, it's important to be able to locate it. More on that in the coma talk. Next, temporal horn of the lateral ventricle. Normally, as in this example, it's slit-like. If it becomes dilated, it's one of the earlier signs of hydrocephalus. Moving lower down, now we are at the mid-pontine level. Here's the pons with its brachium pontus, arms of the pons, or cerebellar peduncles. It's as if the pons is reaching out to the cerebellum for a hug. While the pons and the cerebellum are hugging, the structure in the middle of the hug that's feeling all the love is the fourth ventricle. The place where brachium pontus meets the cerebellum is called the... wait for it... cerebellopontine angle. Later in the talk, we'll take a look at some mass lesions in this area. And here's what the pons looks like on T1. Where's the fourth ventricle? Getting hugged, right in the middle. And here is the illustration.
[8:16]Now, this slice is even more caudal, or lower. Here's the medulla with the fourth ventricle adjacent to the cerebellum.
[8:25]Look at how much better the resolution of MRI is, especially in the posterior fossa. Here's the illustration of the same thing. And finally, the foramen magnum with the cervical spinal cord bathed in CSF.
[8:49]So, now that you're an expert, let's make things a little more challenging. Can you identify the major landmarks on our patient's head CT? Pause the video now.
[9:01]First, let's identify the ventricles. Frontal horn of the lateral ventricle, occipital horn of the lateral ventricle, and the third ventricle. Now, you should be able to locate the basal ganglia structures. Caudate head, thalamus, separated from the lentiform nucleus by the internal capsule. Then, more laterally, the insula, covered by the anterior temporal lobe. Well done, grasshopper. Progress to the next level. Let's talk about symmetry. Here is a non-contrast head CT, axial slice, through the basal ganglia. This image should haunt you in your nightmares by now. Let's locate the midline by drawing a straight line from the falx cerebri to the confluence of venous sinuses, right through the septum pellucidum in the middle of the lateral ventricle. There's no asymmetry here, and the midline is where it needs to be. Now, what about the slice through the pons? Also symmetric. By the way, whenever you see overlapping slices during this talk, both slices come from the same study and the same patient. What about in this picture? Is there asymmetry? Absolutely! Once you spot asymmetry, look for the lesion causing it. It has to be on the patient's right side, since the shift is towards the left. And here it is, a concave subacute subdural hematoma. What about this image? Again, there is significant mass effect and midline shift. But this time, the lesion is a massive acute right middle cerebral artery stroke. Don't worry about how I figured out what these lesions are, that's for the next section. Who's up for another challenge? Let's analyze a posterior fossa slice. Who says there's a lesion on the patient's right? The patient's left? No lesion? Let me highlight the pons, the fourth ventricle, and the cerebellum. Does that help? Yes, there's definitely a midline shift. Now, what's causing the mass effect? There's a mass at the right cerebellopontine angle. You see how difficult it is to pick up on CT? This happens to be a vestibular schwannoma.
[11:30]And finally, not all asymmetry is just about the midline shift. Does this image look symmetric to you? No, there are two subcortical hypodensities that are present on the patient's right side, but not on the left. This happens to be a patient with toxoplasmosis. Again, don't worry about how I knew that, all shall be revealed in its own good time. Let's return to our case once again. Is our patient's scan symmetric or asymmetric? Asymmetric. Now, look for a lesion causing the asymmetry. And here it is.
[12:16]Now that you have an understanding of the basic approach to the head CT, it's time to talk about density.
[12:40]Hyperdensities on CT are easy to spot, and by design, which makes them an excellent diagnostic tool. Bright signal on CT is caused by mineralized structures, such as calcium-rich bone and chronic calcified lesions. But also, head CT is wonderful at detecting acute blood with a sensitivity of above 90%. But before we get bloody, I want to remind you that not all hyperdensities are abnormal. Here is a non-contrast head CT axial cut through the Sylvian fissure. Calcified pineal, calcified choroid plexus, and bone are some examples of normal hyperdensities. What about this image?
[13:26]There are multiple calcified lesions in the left hemisphere and a small one in the right thalamus. I'm about to ruin your appetite. This is a calcified scolex of the tapeworm taenia solium. You probably know this disease by another name. Neurocysticercosis. It's a common cause of seizures worldwide. On this image, the lesion is bright, but less dense and less bright than calcium and bone. This basal ganglia hyperdensity is an acute hemorrhage with some intraventricular extension. Where is the hyperdensity on this scan?
[14:07]It looks like all the sulci were outlined by chalk. There's diffuse hyperdensity in the subarachnoid space, which is extra-axial, outside of the brain parenchyma. This is a great example of diffuse subarachnoid hemorrhage with some intraventricular extension of blood. Once again, identified the midline shift, then look for the lesion causing the shift, and identify the lesion location. We know this lesion is located in the subdural space, because it's concave and crossing suture lines, and therefore extra-axial. You can actually adjust the contrast setting on the CT to bone window. This will help you to identify bony abnormalities like fractures and landmarks like sutures. At this point, it's probably painfully obvious that this is an example of epidural hemorrhage. If the last image was an example of an epidural, this is a subdural. Identified midline shift, look for a lesion causing the shift, and identify the lesion location. We know this lesion is located in the subdural space because it's concave and crossing suture lines, and therefore extra-axial. So, this game should be getting pretty boring by now. Time for another challenge. Pause the video and identify abnormal hyperdensities on these two scans from two different patients. The image on the left shows a serpentine hyperdensity outside of the brain in the Sylvian fissure. What lives in the Sylvian fissure? That's right, middle cerebral artery. The image on the right also shows a hyperdensity, but this time it's between the cerebral peduncles, in the interpeduncular cistern. This is the top of the basilar artery. You know that all acutely clotted blood looks hyperdense or bright on CT. So, these dense vessel signs represent acute clots. Pretty important information to have when treating a patient with suspected acute ischemic stroke.
[16:40]Hypodensities on CT are generally much harder to see, especially in the posterior fossa.
[16:47]Fluid is less dense and appears dark on CT. So, hypodensities usually reflect chronic lesions, cysts, or cerebral edema. Let's look at some examples. First up, chronic damage. This hypodensity is very well-defined, wedge-shaped, same density as the CSF, and conforms to a vascular territory. The brain on the patient's right side looks shrunken, and the frontal horn of the lateral ventricle is expanded as a result. The proper name for this loss of brain tissue is encephalomalacia. So, you probably guessed by now that this is an example of a chronic right MCA stroke. Where is the encephalomalacia on this image? You're right, it's bifrontal. More so on the left, but the right is abnormal too. This does not conform to a vascular distribution. Bilateral frontal lobes are common side of trauma, so this is an example of chronic traumatic brain injury.
[17:57]Now, how is this hypodensity different? Is it involving the brain parenchyma? Is it encephalomalacia? It looks well-defined, but also smooth, unlike the previous examples. The location of this lesion is extra-axial with similar density to CSF, because it's a CSF-filled structure. This is an example of a frontotemporal arachnoid cyst, which is fairly benign, but may become large enough to exert mass effect and cause seizures. Get ready for your next challenge. Is there a hypodensity? If so, which side?
[18:45]Difficult, isn't it?
[18:49]There's a large, faintly hypodense lesion within the right MCA territory. You may remember that acutely ischemic neurons swell, but that edema takes several hours to show up on the head CT. Let's just say 6 to 8 hours on average. So, it's very challenging to detect edema related to acute ischemic stroke on a head CT in the first 6 hours after symptom onset. In fact, most early head CTs in acute stroke patients are normal. This scan was actually done 8 hours after symptom onset. Here is another example. What's unique about the abnormality on this scan? Hopefully, you noticed the finger-like subcortical hypodensities. Looks like a giant hand gripping a small ball. Well, that ball is a brain metastasis, and the hypodense hand is another type of cerebral edema. So, you can see how the hypodensity on the CT can only take us so far. Time to introduce the MRI. As the famous fictional detective Sherlock Holmes would say, It's worth noting that when placing a standard order for a non-contrast MRI of the brain, you typically get the following four-course meal. T1-weighted imaging for appetizer. Your main course will be a T2-weighted imaging with a side of T2 derivative called fluid attenuated inversion recovery, or FLAIR, arguably the most filling and useful sequences. The cheese and fruit plate would be diffusion-weighted imaging and its evil twin, the apparent diffusion coefficient, which are essentially designed to diagnose energy failure and ischemia. And for dessert, we have the gradient recalled echo sequence, or GRE, to detect hemorrhages. If you want to finish your meal with a cup of delicious coffee, may I suggest T1 with gadolinium contrast. But this must be ordered separately and you'll have to pay extra. MRI can also image vasculature, which requires an order for MR angiogram. Finally, there are numerous specialized sequences including fat suppression, perfusion, spectroscopy, tractography, et cetera. There are more fancy MRI sequences than Ben and Jerry's ice cream flavors. MR angiogram and special sequences are outside of the scope of this talk. You can't digest everything in one meal. So, take a normal head CT, zoom in, and adjust contrast. You can plainly make out two different tissue densities. The lighter or less dense is the gray matter, and the darker is the subcortical white matter. The junction where the two meet is called... wait for it... the gray-white junction. Now, let's bring in the T2. Fluid is bright on T2, so identifying fluid-filled cell bodies of the cortical gray matter and distinguishing them from the darker subcortical white matter and the subcortical gray matter is much easier. What is actually happening on a cellular level in all of those crevices? I know, I know, you're allergic to basic science, but stay with me.
[22:10]Have you ever heard of the neurovascular unit? This unit consists of astrocyte foot processes, wrapped around capillary endothelial cells, which are connected by tight junctions. Brain parenchyma filled with neurons is outside, and blood is inside the capillary lumen. You may call it a barrier between blood and brain. Oh, neurovascular unit is a central component of the blood brain barrier, and blood brain barrier surrounds the entirety of the brain parenchyma. What happens when the neurovascular unit is deprived of blood flow? Astrocytes, capillary endothelial cells, and neurons become ischemic. They can't maintain the sodium potassium gradient and take on fluid. Despite that, the blood brain barrier remains essentially intact early on. This is called cytotoxic, or cell body, edema, because it's toxic to the cells. Here is a patient with acute right middle cerebral artery stroke. Edema is everywhere, gray matter, white matter, the cortical ribbon is completely lost, and the loss of this gray-white junction is a clue that we're dealing with cytotoxic edema of ischemia.
[23:30]T2-weighted imaging is much better at highlighting that fluid. Get it? Highlighting because it's bright? No, all right. The edema is bright, but CSF is also bright. So, interpretation of this image is somewhat hampered by the CSF-filled sulci. Here's an idea, what if we subtract the CSF signal? Introducing FLAIR, fluid attenuated inversion recovery sequence. It's unmistakably related to T2, since gray matter is lighter than white matter, and fluid appears bright, with one important exception, the bright signal from CSF has been subtracted. And now, you can quickly identify the abnormality from across the room. Now, imagine that instead of energy failure, the integrity of the neurovascular unit is compromised. The tight junctions become leaky, allowing the plasma to escape into the interstitial space. This type of edema is called vasogenic, as in poor vessel integrity. This happens with neoplasms and cerebral abscesses, which secrete substances that increase blood brain permeability, brain trauma, which can cause mechanical disruption, and even very high blood pressure that can overwhelm the neurovascular unit's ability to regulate fluid flow across the capillary endothelium.
[25:00]In this example, a brain metastasis caused these finger-like projections of vasogenic edema. Which is much better seen on T2-weighted imaging. The edema essentially stops at the gray-white junction and does not affect the gray matter. The cortical ribbon is preserved, and you can make out the angry heterogeneous mass in the middle. And of course, you can see the full extent of the edema better on FLAIR. So, one more time, vasogenic edema because of failing neurovascular unit integrity, which can be caused by tumors, abscesses, trauma, and severe hypertension, and cytotoxic edema or cell body edema caused by energy failure and ischemia. Got it? Look at the hypodensities on CT. How much more inferior they are to FLAIR and T2. T2 and FLAIR are superior by design. And just when you thought the dark side was winning,
[26:05]The light strikes back.
[26:09]With CT, it's all about density. With MRI, it's all about intensity. Bright signal is easiest to see, so we've been tackling edema first. If MRI were a book, the first chapter would be called, look in the bright side, it's swollen. So, we discussed non-contrast head CT, T1, T2, and FLAIR. But when it comes to cytotoxic edema caused by ischemia, T2 and FLAIR kind of behave like CT. It takes hours for a lesion to build up enough cerebral edema to show a bright signal. Diffusion-weighted imaging is designed to detect ischemia earlier, within the first 30 minutes from onset. Apparent diffusion coefficient is DWI's evil twin, it's the anti-DWI. Ischemia will cause bright signal on DWI and dark signal on ADC. Well, all of this seems overly complicated. You're telling me that we need another sequence of MRI to be confirmed by yet another sequence of MRI? Unfortunately, yeah. And here's why. DWI is related to T2 and FLAIR. So, whatever is bright on DWI will also be bright on those sequences. Or it's probably more accurate to say that whatever is bright on T2 and FLAIR will also be bright on DWI. This phenomenon is called T2 shine through. T2-FLAIR abnormalities shine through to the DWI. Now, look at the ADC. Is this an example of an acute stroke? No, ADC is bright, not dark, like in the last example. This is T2 shine through. Bottom line, always check the DWI against ADC to confirm acute ischemia. Our brightness motive continues. Here's another T2 and FLAIR pair of sequences from the same study. What kind of edema is this? Vasogenic or cytotoxic? Finger-like projections, gray matter unaffected, has to be vasogenic. This is an example of a pyogenic brain abscess in right parietal lobe. Incidentally, abscess-forming organisms usually spread through the bloodstream from a primary site. And of course, more of these emboli will end up in the vascular territory of the artery that supplies the majority of the hemisphere. So, it's not surprising that the most common locations for abscesses are in the frontal and parietal lobes in the distribution of the middle cerebral artery. Again, we're really focusing on the bright signal here. From these sequences alone, you cannot unequivocally distinguish an abscess from another mass. You need other sequences of the MRI, but more on that later. Ah, speaking of masses. What's the most common neoplasm in the brain? That's right, metastasis. What is the type of edema pictured on this T2 image? Vasogenic or cytotoxic? Right again. These are finger-like projections of vasogenic edema, which spares the cortex.
[29:24]Remember that these metastatic lesions are near the gray-white junction.
[29:31]That's because, like bacteria, metastasis generally spread through the bloodstream. And arterioles at the gray-white junction are just small enough to trap traveling neoplastic cells. Lung and breast cancer metastases are the most common. This example happens to be lung cancer, but again, the imaging is never that specific. Tissue biopsy is generally necessary to confirm the tumor type. Now, say one of those metastases moves into the posterior fossa, exerting tremendous mass effect on the fourth ventricle. What is the abnormality on this FLAIR sequence? Notice the symmetric smooth hyperintensity that is periventricular. Under pressure, CSF is being pushed out of the ventricle across its walls or ependyma, into the brain parenchyma. So, this transependymal flow of CSF is seen in obstructive hydrocephalus. Lateral ventricles are also certainly dilated. In this case, the hydrocephalus was caused by a posterior fossa mass, obstructing the fourth ventricle, below the level of the slice. Let's move on to the next example of FLAIR hyperintensity. This is a sagittal image, just lateral of midline, with the patient looking to your left. How would you describe the abnormalities?
[31:01]You probably said that these are ovoid periventricular hyperintense lesions. These are Dawson's fingers in patients with multiple sclerosis. And on the axial FLAIR image, you can see them as well. Ovoid lesions. Now, that's a strangely specific appearance. Do you know why that happens? Well, there are so-called central veins running perpendicular to the ventricles. This is one of those fancy MRI sequences that we're not going to cover. In MS, immune cells exit these veins and penetrate the brain parenchyma here, causing this ovoid appearing inflammation. Now, how is this image different? Is this another example of a patient with multiple sclerosis? The hyperintensities are still periventricular and subcortical, but they are more confluent, not ovoid. Also, the sulci are quite deep and prominent because of significant brain atrophy. Confluent subcortical lesions with brain atrophy in a patient with vascular risk factors are likely caused by small vessel ischemic disease.
[32:17]Small vessel ischemic changes are so common that I'm showing them to you twice.
[32:26]Now that we discussed bright signals on T2 and FLAIR, let's quickly mention bright signals on T1. Generally, bright signal on T1 is caused by paramagnetic substances, like iron, copper, melanin, calcium, fat, protein-rich lesions, and subacute bleed. Here is an example of metastatic melanoma in the right eyeball. It's showing up as a T1 hyperintensity. A note that the orbital fat is also bright on T1. That's normal. How about this example? The bright signal in the basal ganglia corresponds to pathological copper accumulation. This is Wilson's disease, an autosomal recessive disorder where gene mutations causes impaired trafficking of copper in and through the hepatocytes, so it deposits in many organs and tissues including the brain. And one more example. This well-circumscribed T1 hyperintensity corresponds to a protein-rich mass at the foramen of Monroe, which is protein-rich. This mass is called a colloid cyst. Its claim to infamy is that it can act like a ball valve occluding the foramen of Monroe and causing hydrocephalus. This can happen so acutely and so severely that it can kill.
[33:55]Blood in the brain doesn't quite play by the rules, and its intensity on T1 and T2 actually evolves over time. In the first several days, the hematoma containing deoxyhemoglobin looks isointense on T1 and darkens on T2. Beyond 3 days, methemoglobin creates a bright signal on T1 and T2.
[34:21]Ultimately, the site of hemorrhage is essentially replaced by a CSF-filled slit-like hole, which looks dark on T1 and bright on T2, as all fluid does. It sort of looks as if the brain has been stabbed. And there's usually a rim of hypointense signal on T2, representing the rim of hemosiderin, which is like a fingerprint left by the chronic hemorrhage. So, what's the stage of this hematoma? Right, it's late subacute stage, about 3 to 14 days. Both T1 and T2 are bright.
[34:59]Hoo, that was a long section, but it had to be. MRI is really designed to highlight lesions, so most abnormalities will be hyperintense. So, a brief summary. T2 and FLAIR brightness will help you diagnose ischemic stroke, DWI is also indispensable for stroke, vasogenic edema due to metastatic disease, brain abscess, transependymal flow due to hydrocephalus, inflammatory lesions, due to MS.
[35:36]T1 brightness will help you diagnose metal deposition, like copper in Wilson's disease, other paramagnetic substances like melanin in ocular melanoma, protein-rich substances like a colloid cyst, fat, and subacute bleeding, 3 to 14 days from onset.
[36:29]Dark signal on T2, can be caused by paramagnetic substances, protein-rich lesions, flowing blood, and acute bleed, 1 to 3 days from onset. And chronic lesions on T1.
[36:54]Dark signal on T1, basically corresponds to chronic lesions which are filled with fluid.
[37:39]And that is all. Thank you for joining me. You should feel incredibly exhausted, but hopefully wiser. Go forth and read imaging studies. Compare your impression with a radiologist's report, that's how you learn. Practice makes perfect. Until next time, bye.
