5 Magnetic Resonance Imaging

DR MUHAMMAD BIN ZULFIQARPGR III FCPS Services institute of Medical Sciences/ Services

Hospital LahoreGRAINGER & ALLISON’S DIAGNOSTIC RADIOLOGY

• FIGURE 5-1 Schematic depiction of signal ■formation. (A) Application of a 90° RF pulse leads to the production of the free induction decay (FID); (B) a spin echo (SE) is produced at the time to echo (TE) by the application of a 180° RF pulse at TE/2; (C) a gradient echo (GE) is produced by the application of a magnetic field gradient (G).

• FIGURE 5-2 (A) Two-■dimensional (2D) ‘k-space’ magnitude data array and (B) corresponding sagittal image formed by 2D Fourier transformation of the k-space data.

• FIGURE 5-3 Basis of the ■gradient-echo (spin-warp) pulse sequence. Three orthogonal gradients are used for: slice selection (SS) (while the RF excitation pulse is applied, producing a flip of α0); phase encoding (PE) (the amplitude of which is iterated every repeat time (TR)); and frequency encoding (FE) (leads to echo formation at the echo time (TE)). The process is repeated after time TR for the total number of required phase-encoding steps.

• FIGURE 5-4 An example 8-channel brain RF ■coil showing low-resolution images of a homogeneous phantom depicting the sensitivity of the 8 coil elements. This sensitivity information is used for SENSE-based parallel imaging reconstruction.

• FIGURE 5-5 Principles of parallel imaging. In (A) the ■ k-space acquisition is subsampled, i.e. every other phase-encoding step is missed. This halves the MR data acquisition time but also means that the field of view (FOV) in the phase-encoding direction is halved and the image appears ‘wrapped’ in the phase-encoding (left-right) direction (B). In the SENSE reconstruction the coil sensitivity information as depicted in Fig. 5-4 is used to unwrap the image, effectively resulting in a full FOV (D). In the SMASH reconstruction the missing phase-encoding steps are synthesised in k-space to produce full k-space data (C). Standard reconstruction is then used to create the final full FOV image (D).

• FIGURE 5-5 Principles of parallel imaging. In (A) the ■ k-space acquisition is subsampled, i.e. every other phase-encoding step is missed. This halves the MR data acquisition time but also means that the field of view (FOV) in the phase-encoding direction is halved and the image appears ‘wrapped’ in the phase-encoding (left-right) direction (B). In the SENSE reconstruction the coil sensitivity information as depicted in Fig. 5-4 is used to unwrap the image, effectively resulting in a full FOV (D). In the SMASH reconstruction the missing phase-encoding steps are synthesised in k-space to produce full k-space data (C). Standard reconstruction is then used to create the final full FOV image (D).

• FIGURE 5-6 The spin-echo pulse sequence ■where a spin ech is refocused at time TE by a 180° RF pulse. Slice selection (SS) gradients are applied in conjunction with the RF pulses. One amplitude of a phase-encoding (PE) gradient is transiently applied within each TR period and frequency-encoding (FE) is performed during echo formation and readout.

• FIGURE 5-7 A multi-echo spin-echo pulse sequence ■applies multiple 180° RF pulses to produce multiple echoes that are individually frequency-encoded (FE) from the same slice having the same phase encoding (PE). Each of the echoes is acquired at different echo times (TEs) and hence will have different T2- weighted signal components.

• FIGURE 5-8 Train of 180° refocusing RF pulses that form the ■basis of the fast spin-echo (FSE) sequence. Each 180° RF pulse has a different magnitude phase-encoding (PE) gradient. A whole train of 2, 4, 8, etc., echoes are produced per TR, with each echo being frequency-encoded (FE), reducing the number of TR values required for a given total number of phase-encoding steps.

• FIGURE 5-9 Imaging of the fetus in utero ■using single-shot fast spin echo. This fetus has agenesis of the corpus callosum depicted in the (A) transverse and (B) midline sagittal planes.

• FIGURE 5-9 Imaging of the fetus in utero ■using single-shot fast spin echo. This fetus has agenesis of the corpus callosum depicted in the (A) transverse and (B) midline sagittal planes.

• FIGURE 5-10 Patient with multiple ■sclerosis with plaques of demyelination. (A) Fast spin-echo (FSE) proton density; (B) FSE T2; and (C) FSE FLAIR. There is no discernible abnormality on T1-weighted images (D).

• FIGURE 5-10 Patient with multiple ■sclerosis with plaques of demyelination. (A) Fast spin-echo (FSE) proton density; (B) FSE T2; and (C) FSE FLAIR. There is no discernible abnormality on T1-weighted images (D).

• FIGURE 5-11 Spin-echo EPI sequence where ■the train of echoes is produced by a 180° RF pulse and each of the echoes within the train occurs at a different ky due to the gradient ‘blips’ applied along the phase-encoding (PE) axis.

• FIGURE 5-12 Formation of gradient and Hahn echoes. Following ■an (α) RF pulse, a free induction decay (FID) occurs. The next RF pulse, as well as creating its own FID, can also refocus the first FID, creating a Hahn echo that occurs at the time of the third RF pulse. Appropriate use of RF pulse, gradients and timing of the data acquisitions means that (1) the Hahn echo can be destroyed, sampling only the FID; (2) the FID and Hahn echo signals can be combined and sampled; or (3) only the Hahn echo can be sampled. In practice only one of 1, 2 or 3 is actually collected, depending upon the actual gradient-echo pulse sequence used.

• FIGURE 5-13 Spoiled gradient-echo pulse sequence. ■Note the phase of the RF pulse is modulated so as to cancel the production of a Hahn echo by destroying the residual transverse magnetisation.

• FIGURE 5-14 Rewound gradient-echo pulse ■sequence. Note the phase-encoding gradient is rewound after the echo and there is no RF ‘spoiling’. The echo is a mixture of both gradient and Hahn echoes.

• FIGURE 5-15 Balanced steady-state free ■precession sequence A modification of the rewound gradient-echo sequence depicted in Fig. 5-14.

• FIGURE 5-16 Steady-state free precession ■sequence. This sequence is also known as a time-reversed PSIF, where the echo is a Hahn echo.

• FIGURE 5-17 Gradient-echo EPI pulse sequence. The ■train of echoes is produced by sampling the FID. In a similar manner to spin-echo EPI (Fig. 5-11), each of the echoes within the train occurs at a different ky due to the gradient ‘blips’ applied along the phase-encoding (PE) axis.

• FIGURE 5-18 A 70-cm-bore 3-T magnet ■installation. (Courtesy of Philips Healthcare.)

• FIGURE 5-19 Examples of ■dedicated RF receiver coils. The coil elements are placed close to the anatomical area under investigation to give relatively high signal-to-noise ratio: (A) a 15-channel head coil, (B) an anterior coil suitable for the torso/ legs and (C) an 8-channel semi-flexible wrap-around coil.

• FIGURE 5-20 Proton density ■(PD)-, T2- and T1-weighted contrast. Transverse (yellow and green) and longitudinal (blue and purple) magnetisation (Mxy and Mz, respectively) are shown graphically as a function of time for two tissues with different relaxivity and density characteristics. The vertical dashed lines indicate where the data are sampled: PD weighting is obtained when data are sampled at short TE where the TR is long; T2 weighting is obtained when data are sampled at a longer TE where the TR is still long and T1 weighting obtained when data are sampled at short TE with a short TR (before Mz has time for complete recovery).

• FIGURE 5-21 Typical ■appearance of a brain tumour, specifically a cerebellar medulloblastoma. (A) The tumour has a low signal on spin-echo T1-weighted image owing to a prolonged T1. (B) After the injection of Gd-DTPA, the tumour enhances avidly, depicting breakdown of the blood–brain barrier.

• FIGURE 5-22 Blood oxygen level-dependent functional MRI (fMRI). ■BOLD fMRI can be used to demonstrate magnetic susceptibility changes in close proximity to neuronal and synaptic activation associated with the vascular oxy-/deoxyhaemoglobin ratio. The subject rests, taps their fingers and repeats this pattern 5 times ((A) - blue line). Signal change within the right primary motor cortex is shown graphically as a function of time ((A) purple line). The anatomical regions where the haemodynamic response correlates with rest and movement are shown as a red overlay to the axial anatomical image (B).

• FIGURE 5-22 Blood oxygen level-dependent functional MRI (fMRI). BOLD ■fMRI can be used to demonstrate magnetic susceptibility changes in close proximity to neuronal and synaptic activation associated with the vascular oxy-/deoxyhaemoglobin ratio. The subject rests, taps their fingers and repeats this pattern 5 times ((A) - blue line). Signal change within the right primary motor cortex is shown graphically as a function of time ((A) purple line). The anatomical regions where the haemodynamic response correlates with rest and movement are shown as a red overlay to the axial anatomical image (B).

• FIGURE 5-23 Haematoma. A patient presents with a history ■suggestive of subarachnoid haemorrhage but MRI shows an extensive left frontal haematoma. The haemoglobin in the haematoma is in different stages of breakdown as shown on (A) the spin-echo T1- and (B) fast spin-echo T2-weighted images. (C) Note the high signal rim on FLAIR imaging indicative of oedema. (D) Gradient-echo sequences are very sensitive for acute haemorrhage and show prominent ‘blooming’ of reduced signal due to susceptibility effects.

• FIGURE 5-23 Haematoma. A patient presents with a ■history suggestive of subarachnoid haemorrhage but MRI shows an extensive left frontal haematoma. The haemoglobin in the haematoma is in different stages of breakdown as shown on (A) the spin-echo T1- and (B) fast spin-echo T2-weighted images. (C) Note the high signal rim on FLAIR imaging indicative of oedema. (D) Gradient-echo sequences are very sensitive for acute haemorrhage and show prominent ‘blooming’ of reduced signal due to susceptibility effects.

• FIGURE 5-24 Exogenous perfusion data. This data was ■obtained from a time series of T2*-weighted echo planar images in a 70-year-old woman who presented with amaurosis fugax and was found to have a 95% stenosis of the right internal carotid artery. (A) A base image shows two regions of interest within the middle cerebral artery territories. (B) A drop in signal intensity can be seen due to the first pass of the Gd chelate, from which (C) a concentration–time curve is calculated and a gamma-variate fit is performed (solid lines). (D) The resultant time-to-peak (TTP) map shows prolonged TTP in the affected hemisphere shown as high signal.

• FIGURE 5-25 Proton density-weighted proton magnetic ■resonance spectrum (A) obtained from a glioma within the brainstem of a patient with neurofibromatosis type I (C – hyperintense central lesion). Compared to a normal-appearing brainstem spectrum (B), increases in the relative levels of myo-inositol (mI) and choline (Cho) plus decreases in N-acetyl (NA) compounds are apparent. The doublet seen at 1.3 ppm is consistent with the presence of lactate, a product of anaerobic metabolism.

• FIGURE 5-26 Water (A) and fat (B) whole-■body images obtained using an in- and out-of-phase dual echo, 3D gradient-echo sequence.

• FIGURE 5-27 Three-dimensional time-of-■flight MR angiography projection of an intracranial, middle cerebral artery berry aneurysm.

• FIGURE 5-28 Non-contrast-enhanced angiogram of the iliac arteries obtained by ■subtracting two IR-prepared, ECGtriggered, 3D fast spin-echo (FSE) acquisitions. The two acquisitions are interleaved: one acquired in systole where arterial flow is dark; and one in diastole where arterial flow is bright. Venous signal and background tissue appears with the same signal intensity in both acquisitions which cancel out after subtraction.

• FIGURE 5-29 Non-contrast-enhanced MRA ■of the renal arteries. This image was obtained using a respiratory-triggered, inversion recovery prepared 3D balanced steady-state free precession (bSSFP) acquisition.

• FIGURE 5-30 Imaging in ■stroke. (A) Diffusion weighting (b = 1000 s/mm2) and (B) the ‘mean transit time’ image derived from a dynamic susceptibility contrast (DSC)-MRI acquisition in the same patient. The focal acute abnormality representing ischaemic change on (A) also shows focal abnormality on (B). However, the perfusion map also shows diffuse, extensive, longer transit time to the right hemisphere that may represent parenchyma at risk of vascular compromise.

• FIGURE 5-31 ■Multifunctional imaging in ovarian cancer. (A) The diffusion-weighted image (b = 500 s/mm2) as a colour overlay on a T2-weighted anatomical image and (B) the ‘perfusion’ image (Ktrans) derived from a dynamic contrast-enhanced (DCE) MRI acquisition in the same patient.

• FIGURE 5-32 Intracranial ■diffusion tensor imaging (DTI). A patient with an intra-axial brain glioma showing (A) directional tensor information overlaying a fast spin-echo T2-weighted transaxial image (red = right-left, blue = cranio-caudal and green = anterior-posterior) and (B) DTI-based tractography of modelled axonal tracts in the vicinity of the tumour.

• FIGURE 5-33 Image of a healthy lung obtained ■with a twodimensional, rapid, steady-state free precession sequence with 300 mL. Hyperpolarised 3He at 30% polarisation. (Courtesy of Professor J. M. Wild, University of Sheffield, UK.)