**3. MRI principles**

MRI is based on the concept of nuclear magnetic resonance. Atomic nuclei with an odd number of protons or neutrons exhibit a net spin entailing a charge circulation that forms an individual magnetic field surrounding the atom, giving the protons a

magnetic dipole [50–55]. As the hydrogen atoms exhibit ±½ spin, and the nuclear spins exist in two states that are randomly oriented, in absence of a net magnetic field, there is no overall net magnetization. When placed in an external magnetic field, the spins orient parallel and anti-parallel to the direction of the B0 magnetic field, with a slight propensity for the spins to align in the parallel direction, causing the tissue to express a net equilibrium magnetization [50, 53]. The magnetic moment of the atom rotates like a spinning top, predominately in the direction of the applied magnetic field. This magnetic moment rotates at an angular frequency unique to individual atoms, termed the Larmor frequency.

When a perpendicular radiofrequency field (B1) is applied at the hydrogen Larmor resonance frequency, only the protons absorb energy, and are tipped from the direction of the main magnetic field, with the flip angle denoting the degree that the spins are displaced from the equilibrium B0 direction [55]. This excites the protons to precess in a rotational motion around the B0 field vector. The excited proton magnetization vector then relaxes in the direction of the main B0 magnetic field, generating a longitudinal and transverse time-varying magnetization signal that is detected by the MRI receiver coils.

The rate at which this magnetization vector relaxes towards the main magnetic field direction is measured in terms of spin-lattice relaxation (T1) in the direction of the B0 magnetic field, and the spin-spin relaxation rate (T2) trans-verse to the B0 magnetic field direction [50, 53]. The T1 and T2 decay rates result from random static magnetic field variations. However, the relaxation rates are also influenced by time varying factors, such as magnetic field inhomogeneities, that combine with tissue static magnetic field to affect the relaxation rate.

The net magnetism applied to each proton results from both the field generated from the MRI system, in addition to the fields generated by the surrounding protons and bulk susceptibility [55, 56]. A chemical shift in the precession frequency results from the magnetic fields generated from these surrounding protons. This can allow identification of specific molecules present in the tissue, that introduce a distinctive chemical shift in the MR signal [55]. The degree of this shift also has a temperature dependence. Using the principle of proton resonance frequency shift (PRFS) thermometry, the individual temperature of each voxel can be quantified from the resulting temperature-dependent phase change due to this chemical shift [56].

### **4. MRI liver imaging**

Radiological imaging is used in a variety of manners in treating CRLM: including, to diagnose a condition, stage the disease, to locate extra-hepatic metastases, for treatment planning, for interventional image-guided procedures, and for post-treatment evaluation [57]. MRgHIFU requires additional MRI sequence protocols, compared to general diagnostic MRI.

### **4.1 Diagnostic MRI for CRLM**

Although CRLM is usually confirmed with computed tomography, MRI is an acceptable and common alternative, and is advantageous at identifying small lesions [1]. Some studies have shown MRI to provide the best results among all diagnostic imaging modalities, though more expensive [58]. The primary objectives for MRI liver tumor diagnosis are to verify the neoplasm presence, staging the lesion, and

### *Magnetic Resonance-Guided Focused Ultrasound in the Treatment of Colorectal Cancer Liver… DOI: http://dx.doi.org/10.5772/intechopen.105906*

classifying the type of neoplasm [22]. Accurate assessment of these techniques is crucial to guiding subsequent treatment such as resection, biopsy, and chemotherapy [22]. National Comprehensive Cancer Network (NCCN) guidelines recommend CT be used for initial workup and staging; with MRI recommended for potentially resectable cases, prior to locoregional treatment, and for inadequate imaging with CT [59].

Metastatic liver tumors have been reported as a factor of 18–40 more frequent than primary tumors [17]. The presence of both benign and malignant liver lesions are common. The challenge is often distinguishing the benign liver lesions from malignant lesions, as misdiagnosis can greatly impact staging and treatment planning. CRLM lesions exhibit T1 signal hypointensity, higher FATSAT-T2W signal intensity, and higher diffusion-weighted imaging (DWI) signal intensity. On T2W, the tumor resembles a target; with coagulative necrosis causing a relatively higher signal intensity in the tumor center, followed by a reduced signal exterior due to bulk desmoplasia, and an even lower intensity thin edge from desmoplasia growing at the periphery. This thin edge resembles a ring in the arterial phase when gadolinium is administered. These features can change due to fatty liver infiltration and edema [60].

Standard liver tumor protocols are concerned with imaging the parenchyma, vascular supply, and biliary tract [61]. Basic liver protocols often include: T2 half acquisition single-shot turbo spin echo (HASTE) localizer, in-op phase T1 Gradient Recall Echo (GRE), T2 fast spin echo (FSE) with fat saturation (FATSAT), and gadoliniumenhanced 3D FATSAT T1 GRE [61, 62]. The HASTE localizer uses a motion insensitive T2 single-shot spin echo sequence in combination with half-Fourier to acquire a multislice image in about 2 seconds during a single breath hold [63]. The in-op phase Dixon technique, is a spectroscopic technique used to suppress fat signal, quantify the hepatic fat content of the liver, and estimate iron content [63]. The spectroscopic method distinguishes an image at the ∙CH2 fat chemical shift from an image at the water chemical shift [64]. In-phase and op-phase sequences are often spin-echo or GRE sequences with equal repetition times, but different echo times. It acquires a normal in-phase image containing the water and fat, an opposed-phase image containing the water phase signal lessened by the fat phase contribution. Combining in-phase and op-phase images generates the water only image, and subtraction of the op-phase image from the in-phase image allows isolation of the fat signal [64, 65]. Additionally, the Dixon technique allows the generation of a T2 \* map, from which the local iron content (mg g−1) can be formulated [65, 66].

Of high importance in clinical diagnosis of liver lesions are DWI and hepatocytespecific magnetic resonance contrast agent imaging, with MRI elastography to a lesser extent [67]. DWI is particularly useful for detection of small metastatic lesions [61]. Liver DWI consists of a T2 sequence with symmetric diffusion sensitizing gradients centered on the 180° refocusing pulse [67, 68]. Brownian motion of water molecules is more restricted in tumors and provides a noticeable degree of contrast compared to normal tissue [69]. The DWI sequence is generally used without the administration of a contrast agent, making it a completely non-invasive diagnostic sequence. The weighting factor in DWI is adjusted based on the b-value, that is a function of gradient strength and duration. The apparent diffusion coefficient (ADC) maps can be viewed by removing the T2-weighting from a series of diffusion-weighted images. Hyperintense regions generally correspond to regions of low fluid diffusion [63].

CRLM lesions are a solid liver lesion and a general protocol for identification and characterization can be described as follows. First, a highly T2-weighted SSTSE to identify benign fluid-filled lesions, such as cysts and hemangiomas. Next, a modestly T2-weighted FATSAT-TSE or DWI to identify metastatic tumor sites. Then, a Dixon

sequence might be used to observe the degree of fat infiltration into the tumor. Lastly, a contrast-enhanced image can be used for T1-weighted phase imaging to characterize the tumor [70].

Extracellular gadolinium agents are the most common contrast agents for general imaging throughout the body [71]. Two common hepatic specific contrast agents are gadoxetate disodium (Gd-EOB-DTPA, Primovist, Eovist, Bayer Healthcare Pharmaceuticals) and gadobenate dimeglumine (Gd-BOPTA, Bracco Diagnostics) [67]. In some studies, Gd-EOB-DTPA hepatocyte specific MRI contrast agents has shown improved sensitivity and specificity in diagnosis of liver metastasis compared to computed tomography, particularly to the improved ability to detect small metastases [67, 72]. The hepatic-specific contrast agents are specific to tumors originating from hepatocytes, and can help distinguish these lesions from cavernoma or metastatic lesions [71, 73]. Though, these are more expensive than extracellular analogues, have a lower recommended dose and signal, and can exhibit reduced uptake in patients with hepatocyte dysfunction [69]. A comparison of DWI and Gd-EOB-DTPA-T1W MRI for detecting small lesions from CRLM are shown in **Figure 1** [74].

### **4.2 MRgHIFU sequence aspects**

MRgHIFU requires additional MRI sequences that allow for temperature mapping. MR temperature mapping most commonly utilizes PRFS thermometry [75–77], though other possible techniques allow temperature measurements based on the temperaturedependence of relaxation rates, proton density, water diffusion coefficient, thermosensitive contrast agents, and magnetization transfer [78, 79]. Resonance frequency shift results from temperature differences in water molecules and aqueous tissues, due to varying degrees of hydrogen bonding. At increased temperatures, the amount of hydrogen bonding is reduced. This increases nuclear shielding of water protons from the incident magnetic field, generating a lower resonance frequency in the water molecules [78]. This results in a linear-dependence of the phase map values from the chemical shift due to temperature change, at a rate of about −0.01 ppm °C−1 [78, 79]. MRgHIFU sequences

### **Figure 1.**

*Comparison of diffusion-weighted MRI with contrast-enhanced T1W MRI. Left: diffusion-weighted MRI of liver metastases. The arrows indicate small metastatic tumors, less than 1 cm diameter. Right: CE-T1W image after applying Gd-EOB-DTPA, in the same patient. Reprinted with permission from Koh and Berry [74].*

### *Magnetic Resonance-Guided Focused Ultrasound in the Treatment of Colorectal Cancer Liver… DOI: http://dx.doi.org/10.5772/intechopen.105906*

are often based on GRE segmented echo-planar imaging (SEG EPI) sequences. A basic GRE sequence is the fast low-angle shot (FLASH) sequence that utilizes small flip angles to obtain a short echo time (TE) and repetition time (TR) [80]. The sequence further benefits from EPI to accelerate the acquisition rate.

Additional sequences are used to assess tissue peri-ablation and post-ablation. During the peri-ablation period, inflammation in the focal region results from edema, giving more contrast enhancement, and remains for some months. After ablation, T2 and peripheral T1 hyperintensity increases significantly due to the presence of hemorrhagic debris at the ablation region. Thickening or nodule formation in the peripheral hyperintense signal can also indicate recurrence or incomplete ablation, during the months following the procedure [67]. Alternative sequences are under study for other aspects of the modality. For example, magnetic resonance acoustic radiation force impulse (MR-ARFI) sequences allow simultaneous displacement and temperature measurements [81, 82], and is implemented in clinical research settings for tracking focal spot and assessing positioning errors [83, 84]. Additionally, MR-ARFI sequences are being studied for phase aberration correction that occurs in transcostal or transcranial procedures [85–87].

Also, thermal ablation needs temperature processing less than about one second. The faster sequences result in reduced signal to noise ratio and increased temperature uncertainty. Echo planar imaging, parallel imaging, alternate trajectories, and undersampling can increase the MRI frame rate [88–90]. In typical rectilinear sampling, the RF pulse frequency and slice-select gradient determine the slice to be imaged, the frequency encoding gradient amplitude controls the kx-dimension position, and the phase encoding gradient amplitude controls the ky-dimension position [54]. Alternate trajectories are useful, particularly for fast acquisition times and reducing motion artifacts. Radial trajectories are utilized in some of the fastest real-time MRI sequences [90]. Magnetic field inhomogeneities and magnetic susceptibility are also significant aspects to proper imaging and temperature mapping [91, 92].
