Over time, CT has evolved from a “slice-oriented” imaging modality to an “organ-oriented” modality. Helical CT, also known as spiral CT, has several advantages over older imaging techniques, such as eliminating interscan delay, its ability to produce overlapped images without overlapped scans, and having a more uniform sampling density.

CT scanners are differentiated by the number of tubes and detectors, classifying them as single- or dual-source scanners, and also by the number of slices that can be acquired per rotation. Currently, many centers utilize 64-slice scanners that generate 64 slices per rotation and acquire conventional axial and helical acquisitions with collimation ranges of 1 to 32 mm and 16 to 32 mm, respectively. In higher level applications, CTs may generate up to 320 slices per rotation, which could cover up to a 16-cm columnar length of a subject patient during a gantry rotation (ie, cardiac CT). Different gradations of tissue and substance density are acquired by CT imaging. This is represented by the Hounsfield unit (HU) CT density scale that ranges from -1000 for air
to +1000 HU for bone, with 0 HU representing the density of water. The upper limit of the scale may be further differentiated (up to 4000 HU) for bone density and even different types of metals.

When an image is taken, the CT gantry, a circular rotating frame, rotates around the patient. Within the gantry, an x-ray tube is mounted on one side with a detector on the opposite side. The gantry rotates around a center point called the isocenter. This generates raw data representing the digitalized x-ray signal collected by the detector. Contrast resolution distinguishes between different densities and can generate sharp edges between small heterogeneously dense objects. Image processing utilizes pixels and voxels. Pixels are two-dimensional elements utilized to generate images on a display monitor. Voxels are three-dimensional volume representations derived from pixels. Pixels are the building blocks of an image matrix. Volume averaging describes when different objects are represented in the same voxel, thus only partially filling an individual voxel. Reformatted images are created from axial images and arrange axial data into other planes (ie, coronal, sagittal). Reconstruction algorithms are mathematical volumetric rendering techniques utilized to generate multidimensional views. Currently, there are a multitude of software packages available to generate three-dimensional reconstruction images for different pathologic conditions across the spectrum of clinical care (ie, diagnosis, surgical planning).

CTA utilizes intravenous (IV) iodinated contrast media to visualize the vasculature. Of note, because contrast media consist of small and highly diffusible molecules, they also give better interpretive resolution to certain organs with high perfusion. For example, in the non-neurologic realm, the liver, spleen, and kidneys may undergo contrast enhancement, thus improving the sensitivity of CT imaging to rule out traumatic injuries or pathologic conditions.

Angiography involves rapid contrast injection using mechanical injectors and standardized multiphase injection protocols. Contrast media travel from the right heart through the lungs to the left heart, and then distribute through the central arterial vasculature before entering different vascular beds and organs. The resultant attenuation in the vessels is influenced by the contrast injection rate, injection duration, cardiac output, tube voltage, and venous return. Thus, the timing of image acquisition relative to the contrast injection is critical and varies from patient to patient. One of the minimum prerequisites for more advanced CTA applications is a 64-slice scanner. Some applications, such as coronary CTA, are optimized further with higher slice scanners.

CTA has undergone recent evolution and may be categorized into different detection modes—single-phase and multiphase modes. Multiphase CTA provides angiographic images in three distinct phases after contrast injection: peak arterial phase, peak venous phase, and late venous phase. Traditional single-phase CTA only acquires images in the arterial phase and appears to give less accurate information about pial collateral arterial filling.

CTP provides a functional snapshot image of the ischemic penumbra, differentiating the region of the unsalvageable core from the salvageable region of the penumbra. Brain tissue flow may be described by several variables, including cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT) (Table 5.1). Currently, a widespread version of CTP automatic processing using artificial intelligence (AI) is RAPID AI.¹ It provides a fast automated quantitative interpretation of the CTP, and was the software utilized in extended window thrombectomy trials.² RAPID AI is utilized following an initial CTA. A second, smaller volume contrast bolus (˜40 mL) is administered as multiple sequential CT scans (from skull base to vertex) are obtained to capture the entire contrast passage (arterial inflow through venous outflow). Pixel data is interpreted by the RAPID software generating perfusion maps from CBV, MTT, time to peak concentration, and CBF. The final display generates an image (see Figure 5.1), with a nonsalvageable core (milliliter) and a penumbral volume of hypoperfusion (milliliter) amenable to reperfusion interventions. Target thresholds for core and mismatch volume are used to estimate benefit from thrombectomy in the extended window of stroke presentation (Table 5.1).

MRI, and its many sequencing options, is an alternative to CT imaging, and free of ionizing radiation. It has several applications in the hyperacute phase of care. MRI can be utilized to confirm the onset of an acute ischemic stroke in the hyperacute phase. It gives information about stroke onset; enables arterial and venous imaging; identifies hemorrhages that may not be seen on CT; and it may also be utilized for assessing the etiology and severity of spinal cord injuries or syndromes in patients with fractures or new neurologic deficits.

The major components of an MRI include a superconducting magnet, gradient coils, and radiofrequency coils. The superconducting magnet ranges from the milli-Tesla range to 7 Tesla, with most conventional MRIs ranging between 1.5 and 3 Tesla. It defines the sequence time and imaging resolution. The superconducting magnet may, in some units, be cooled to almost absolute zero by liquid helium and mounted with a hollow bore to accommodate the patient. Gradient coils are used to provide intentionally measured variations in the magnetic field in the x, y, and z planes. Radiofrequency coils emit low-energy electromagnetic waves that emit a signal and receive a return signal.

Principles of quantum mechanics, namely, the atomic property of angular nuclear spin momentum, underlies the kinetics of magnetic resonance. Unlike some modalities of CT that focus on electrons, MRI focuses on protons. Essentially, the protons of hydrogen atoms, present in the total body water of a subject undergoing MRI, are influenced by the application of a strong external magnetic field. Without a magnet, there is a random alignment of protons. With the application of the magnetic field, alterations in angular spin cause an alignment to the magnetic field. In this magnetic-nuclear interaction, hydrogen nuclei may align either in parallel or perpendicular to the magnetic field depending on whether there is a low- or high-energy state. This is a reversible, modifiable, and measurable event. The term describing the change in orientation of the rotation axis of the spinning nucleus is precession.

The fraction of protons aligned with the magnetic field may be influenced by other modifiable forces, namely, radiofrequency energy pulse sequences. Application of a radiofrequency pulse can alter the net magnetic movement of protons. Upon cessation of the radiofrequency, relaxation phases occur; and echo waves are emitted by the protons, which can be measured. Time constant T1 refers to longitudinal relaxation time along the axis of the magnetic field, and is also known as spin-lattice relaxation time. It is the time constant by which anatomic spins align themselves to the magnetic field. Time constant T2 refers to the transverse relaxation time perpendicular to the magnetic field, and is also known as spin-spin relaxation time. It is the time constant for the loss of phase coherence among spins positioned at an angle to the magnetic field due to interactions between the spins. Thus, three distinct properties of materials can be determined by MRI: concentration of protons, T1, and T2. The utilization of radiofrequency pulse sequences allows the emphasis of one property over another, thus enabling contrast differentiation between different tissues dependent on the pulse sequence utilized.

Many different imaging sequences are derived from modulations of T1 and T2 relaxation times. T1-weighted images give the most anatomic representation of tissue planes. Fat rapidly aligns its longitudinal magnetization, appearing bright on image sequences. Water aligns at a slower rate, and thus has a dark appearance on sequences. This sequence can differentiate between central nervous system (CNS) gray and white matter. T2-weighted images alter with structural or metabolic changes in a tissue, making it able to identify pathologic changes in body tissue. The dominant signal intensities of different tissues are influenced by water and fat having a high signal intensity. Diffusion-weighted imaging (DWI) represents imaging sequences detecting the free diffusion of water and T2. In an acute ischemic stroke, signal abnormalities may be seen several minutes after onset due to swelling in the ischemic brain parenchyma that prevents water from diffusing freely from the extracellular to intracellular space. Fluid attenuation inversion recovery (FLAIR) sequences are fundamentally T2 sequences with subtraction of the cerebrospinal fluid (CSF). It is useful in evaluating many disorders of the CNS, including stroke and subarachnoid hemorrhage. It may also identify leptomeningeal diseases. Apparent diffusion coefficient (ADC) maps are images representing the actual diffusion values of water in tissue without T2 effects. Susceptibility-weighted imaging (SWI) is a sequence particularly adept at identifying compounds that distort the magnetic field (ie, deoxyhemoglobin, ferritin, hemosiderin), making it sensitive for identifying hemorrhage. Gadolinium contrast enhancement augments the T1 signal, and pathologic areas may have an increase in contrast accumulation. Fat suppression may be done for T1 and T2 sequences. Suppression can remove the high signal intensity from fat, enabling visualization of other underlying pathology (ie, the intramural thrombus of a craniocervical dissection). Finally, flow-sensitive sequences include magnetic resonance angiography (MRA), magnetic resonance venography (MRV), cine MRI utilized for CSF flow, and perfusion. Gadolinium contrast
enhancement improves the diagnostic accuracy of MRA and MRV over non-contrast-enhanced “time-of-flight” (TOF) imaging sequences.

Non-contrast CT of the head is the preliminary and most prevalent imaging modality utilized in the initial evaluation of ischemic and hemorrhagic stroke and traumatic brain injuries. Eligible patients with suspected acute ischemic stroke generally undergo a non-contrast CT of the head to exclude hemorrhage before treatment with IV tissue plasminogen activator (t-PA). However, the non-contrast CT may offer a wealth of other information in the acute phase for acute ischemic stroke presentations. It may differentiate acute (ie, loss of gray/white differentiation, sulcal effacement) from subacute changes (ie, well-defined hypodensity), may reveal a hyperdense vessel sign suggestive of a thrombus in a vessel causing a large-vessel occlusion (LVO), and may delineate the size of the infarct area (Figure 5.2). Of note, windowing, or gray-level mapping, is the process by which image brightness, image contrast, and the gray scale component of images can be manipulated to accentuate certain structures. Most viewing platforms have a “stroke” preset that will, for instance, highlight the differentiation between cerebral gray and white matter. There will also be a “brain” sequence preset that enables best identification of intracranial bleeding.

For acute ischemic stroke, a sample quantitative method validated to estimate the size of a penumbral core from the non-contrast CT is the ASPECTS score (Alberta Stroke Program Early CT Score).³ ASPECTS is a 10-point non-contrast CT scoring system evaluating for early ischemic changes indicative of infarct and cytotoxic edema. In the middle cerebral artery territory, seven cortical and three subcortical territories are assessed for ischemic changes (Figure 5.3). A score of 10 reflects a normal scan without any early ischemic changes. Scores may be stratified into three groups: small-volume infarcts (ASPECTS 8-10), large-volume infarcts (ASPECTS ≤ 7), and very large-volume infarcts (ASPECTS ≤ 5). A score ≥8 correlates with better clinical outcomes after reperfusion therapies. A score ≥6 was a selection criterion for identifying patients with LVO stroke eligible for endovascular thrombectomy within 6 hours of stroke onset. Conversely, poor ASPECTS scores (low scores) correspond with a higher risk of developing a malignant middle cerebral artery syndrome
from secondary edema, potentially necessitating a craniotomy. Non-contrast CT may also be utilized in the later phases of stroke syndromes to identify secondary complications of large territory strokes, including, but not limited to, hemorrhagic transformation, cerebral edema, mass effect, and herniation syndromes. Future investigation may identify and validate automated methods for calculating ASPECTS, as well as identifying other findings in the acute presentation (ie, hyperdense vessel signs).