 |
|
||||
| Home | Login | |||||
| About | Editorial board | Articles
|
NSI Publications
|
Search | Instructions | Online Submission | Subscribe | Videos | Etcetera | Contact |
|
||||||||||||||||||||
|
|
|
Magnetic resonance sequences: Practical neurological applications
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.156293
Magnetic resonance imaging (MRI) plays an important role in the diagnostic work-up of almost all patients with neurological complaints. Various complex MRI sequences are available, for detection and characterization of lesions, functional imaging, and evaluation of therapeutic response. Each MRI sequence is carefully designed to highlight a particular tissue or parameter. It is important to understand the utility and limitations of each sequence, so that an appropriate set of sequences may be chosen and the scan tailored to the patient's requirement. [1] The purpose of this article is not to discuss the detailed physics of MR sequences but to familiarize readers with the practical applications of commonly used MR sequences in neuro-imaging. Magnetic resonance sequences are basically a series of different radio-frequency (RF) pulses, applied at particular times in a specified way to obtain an image. [2] Two key parameters that influence the image contrast are TR (repetition time) and TE (echo time). TR is the time interval between application of an RF excitation pulse and start of the next RF pulse, and TE is the time elapsed between the RF pulse and the peak echo which is detected. [1] T1-weighted (T1W) sequences have both a short TR and TE. On T1W images fat, subacute hematomas, slow moving blood and gadolinium based MR contrast appear bright, and tissues with fluid (water, cerebrospinal fluid [CSF] etc.) appear dark. T2W sequences have both a long TR and long TE. Free fluid and tissues with high free water appear bright. Proton density (PD) weighted sequences have long TR and short TE, and produce contrast due to the difference in PD of the tissues. Tissues having a higher density of protons give higher signals. [2] With the advent of fluid-attenuated inversion recovery (FLAIR) sequence, PD sequences are rarely used. There are two types of MR pulse sequences, spin echo (SE) and gradient recalled echo (GRE); all other sequences are variations of these, with different parameters added on. [1] The four basic MR sequences used in almost all patients are T1W, T2W, T2 FLAIR and diffusion weighted imaging (DWI). Apart from these, MR angiography (MRA), MR venography (MRV), susceptibility weighted imaging (SWI), MR spectroscopy (MRS) and postcontrast T1W sequences are frequently used. Advanced work-up with diffusion tensor imaging (DTI), perfusion and functional imaging is being explored in many conditions due to their availability in the latest MR scanners. [3] The T1-weighted images are good for showing anatomical details, and in combination with T2W images, they are used for characterization of lesions as iso-, hypo -or hyper-intense relative to the grey matter, white matter, CSF or fat. T1W images are also required for assessing the degree of contrast enhancement on postcontrast scans. [4] Conventional SE T2W sequences have been largely replaced by fast spin echo (FSE) T2W sequences for imaging of brain and spine [Figure 1], due to their shorter acquisition times. [4] Motion artifact-free T2W sequences are available for restless patients that compensate for head movement (called periodically rotated overlapping parallel lines with enhanced reconstruction "(PROPELLER)" by GE Healthcare [5] and "BLADE" by Siemens).
Fast spin echo T2W sequences have less CSF pulsation artifacts and fat appears relatively bright in signal intensity. [4] There is improved conspicuity of spinal nerve roots, which is helpful in the setting of arachnoiditis and other nerve root lesions. [4] However, FSE T2W sequences are relatively insensitive for detecting hemorrhage; hence, they have to be supplemented with GRE sequences. [6] A relative weakness of SE techniques, and particularly of newer FSE sequences, is the diminished conspicuity of lesions located in the peripheral cortex or the periventricular region, due to adjacent bright CSF signal. [7] In T2 FLAIR sequences, the CSF signal is suppressed and periventricular lesions are better visualized [Figure 2]. T1 FLAIR sequences are also useful and compared with the conventional T1W image, the T1-FLAIR sequences produce better contrast between gray and white matter [Figure 1]. In addition, a double inversion recovery (DIR) pulse sequence is also available, which uses a combination of two different inversion pulses, to suppress the signal of CSF as well as the normal white matter. DIR sequence is not widely used because of its relatively long acquisition time and propensity for CSF pulsation artifacts. [8]
Echo-planar imaging (EPI) sequences have very short acquisition time as multiple lines of imaging data are acquired in a single TR. These sequences are the basis for advanced MRI applications such as diffusion, perfusion and functional imaging. They use either the SE or GRE format. EPI sequences provide greater tissue contrast than standard GRE sequences; therefore, these are widely used to assess cerebral perfusion. [1] Problems with EPI sequences are their vulnerability to magnetic susceptibility effects and their limited spatial resolution. In DWI sequence, gradients are applied in such a manner that tissues where diffusion is restricted, appear bright. [9],[10] An apparent diffusion coefficient (ADC) image is a parametric map, which is derived from the diffusion images, and areas having true restricted diffusion appear dark (show low ADC values). Tissues with high T2 relaxation times may also appear bright on DWI, even if there is no true restricted diffusion. This is called "T2 shine through" effect. However, such areas remain bright on ADC maps. The main clinical application of DWI is in the evaluation of acute stroke, for which it is the most sensitive sequence. It can differentiate between cytotoxic and vasogenic edema. It is useful for differentiating epidermoids from arachnoid cysts; epidermoids show restricted diffusion on DWI while arachnoid cysts show CSF signal intensity. [10] DWI sequence also reveals restricted diffusion in abscesses [Figure 1] and tumors of high cellularity (medulloblastoma, ependymoma, and lymphoma). After chemotherapy, ADC values may be used to assess response to therapy. A DWI sequence is relatively insensitive to patient motion and the imaging time is very short. The set of these four basic sequences are usually adequate to exclude or to establish a diagnosis of common entities like infarct (acute or chronic) or hemorrhage. In most other cases, at least an anatomical localization of the abnormality is possible. Further sequences are then required for a detailed analysis of the morphology, extent, enhancement characteristics of the lesion or for vascular imaging. The most common method of data collection for DWI is the single-shot (SS) EPI due to its short acquisition time and relative insensitivity to motion. However, a major draw-back of this technique is its sensitivity to field inhomogenities, which leads to geometric distortions in areas of susceptibility changes (e.g., air-tissue interface and bone-tissue interface). Image quality is poor due to low signal to noise ratio (SNR), low resolution and chemical shift artifacts. Several non-EPI diffusion techniques are being investigated for their clinical use. These include the SS-FSE methods, multi-shot noncartesan methods, simulated-echo based methods and steady-state methods. [11] SS-FSE and multi-shot methods have less image distortion and susceptibility artifacts; however, they are associated with more RF energy deposition in tissues and acquisition times are longer than the EPI methods. Multi-shot PROPELLAR DWI method is currently being used for evaluation of cholesteatomas. [12]
Some tissues or pathological manifestations have unique magnetic susceptibility differences with respect to background tissues, which are used to distinguish them from the background, for example, deoxyhemoglobin in veins, hemorrhage, iron laden tissues (in grey matter structures such as caudate and red nucleus and the substantia niagra) and calcium depositions. [13] SWI or GRE sequences are very useful if hemorrhage or calcification is suspected [Figure 3]. They are used in patients with acute head trauma; suspected hemorrhagic or melanotic metastasis; vascular malformations and hemorrhagic sequel of arteriovenous malformation (AVM). [13] The biggest disadvantage of GRE or SWI sequences is the loss of signal from static field inhomogeniety. The inherent internal signal characteristics of lesions get obscured. GRE and SWI sequences produce a lot of artifacts at air-brain interface and in postoperative patients.
Recent advances suggest that MRI may be used to differentiate calcification from hemorrhage based on tissue magnetic susceptibilities. Calcifications are diamagnetic relative to brain parenchyma, and blood-related products, such as deoxyhemoglobin and hemosiderin, are paramagnetic. [14] SWI generates magnitude and phase images from a high-resolution, three-dimensional (3D) fully velocity-compensated GRE sequence. The phase images are used to differentiate calcium from blood. [15] More recently; quantitative susceptibility mapping has been recommended for differentiating calcifications from hemorrhages. [14] In spin echo T2W sequences, the effect of magnetic field inhomogeneity is removed by a 180 degree pulse, whereas in GRE sequences with long TEs, the field inhomogeneities are not refocused and a T2* weighted image is produced rather than a T2 weighted image. Many newer sequences have been developed from the GRE sequences. The two main classes of GRE sequences are incoherent (or spoiled) and coherent sequences (or steady-state free precession [SSFP]). In incoherent sequences, GRE sequences with spoiled residual transverse magnetization produce T1W images. Common names for this type of GRE sequences by different vendors are spoiled gradient-recalled acquisition in the steady state (SPGR), Fast low-angle shot (FLASH) and T1 fast field echo (T1-FFE). T1W GRE sequences are fast and are particularly suited for monitoring gadolinium bolus arrival for chasing the arterial phase and for T1W high-resolution 3D imaging. In addition, ultra-fast spoiled gradient sequences are also available. Coherent sequences have several variants, each having a different contrast. [1],[2] Postexcitation refocused steady state sequences are mainly T2*-weighted. Common names for this type of GRE sequences by different vendors are Fast imaging with steady-state precession (FISP), gradient-recalled acquisition in the steady state (GRASS) and Fast field echo (FFE). These have high SNR and are suitable for 3D imaging. Preexcitation steady state refocused sequences are mainly T2 weighted. Common names for this type of GRE sequences by different vendors are Reversed FISP (PSIF), steady state free precession (SSFP) and T2- Fast field echo (T2-FFE).These have a lower signal-to-noise ratio and are used for CSF flow studies and inner ear imaging. Modified fully refocused steady state sequences yield high resolution 3D T2W images (constructive interference in steady state [CISS], fast imaging employing steady-state acquisition [FIESTA]). [1],[2]
Most volume sequences provide heavy T1 weighting with excellent grey-white matter differentiation. [16] Commonly used sequences are BRAVO from GE healthcare and MPRAGE from SIEMENS. These sequences are very useful for detection of cortical dysplasia and heterotopias [Figure 4]. Thin sections allow detection of small lesions, e.g. small metastases or granulomas. Furthermore, preoperative evaluation of masses is much easier because of its multiplanar reconstruction capabilities. Another common use of 3D GRE sequences is in providing the morphological base for fiber tractography and functional MRI (fMRI). The tractography or fMRI images are usually superimposed on a baseline 3D sequence for anatomic localization. However, these sequences are usually susceptible to patient motion. Also owing to very thin sections, they have low contrast to noise ratio. Therefore, contrast enhancement is better appreciated on conventional SE T1W sequences which are used to supplement 3D sequences. Currently, T2W volume sequences are also available (e.g. CUBE from GE healthcare, SPACE from SIEMENS).
Three-dimensional CISS is a type of fast GRE sequence. [17],[18] The advantage of CISS is in providing an excellent contrast between CSF and other structures. It is relatively insensitive to motion, susceptibility and flow-related artifacts. Thin contiguous sections and high in-plane resolution enable clear visualization of minute structures, and it can be acquired and/or reconstructed in any desired plane. The limitations include long image acquisition time and poor gray-white matter differentiation. CISS is useful for evaluating structures close to or surrounded by CSF and for lesions which are relatively isointense to CSF on T1W and T2W images. It is good for evaluation of cranial nerves and is routinely used for the assessment of cerebellopontine angle lesions, inner ear structures and the internal auditory canal [Figure 5]. In cases of pulsatile tinnitus, facial spasms and trigeminal neuralgia, this sequence plays a major role in identifying any vascular loop which may be compressing/displacing the adjacent nerve. It has a unique role in evaluation of cisternal spaces to identify non-enhancing nodules or infiltrates, which would be missed on other sequences. Similarly, it is very useful for evaluation of the ventricular system and to study the CSF pathways. It can reveal causes of obstructive hydrocephalus such as congenital aqueduct stenosis, membranes, and can delineate margins of intra-ventricular lesions. CISS is the sequence of choice for the evaluation of cysts and cystic lesions. CISS readily identifies the scolex in neurocysticercosis, which is essential for making a definitive diagnosis. In postoperative cases, CISS may help to differentiate between a recurrent arachnoid cyst and a postoperative cavity by demonstrating adhesions.
Magnetization transfer (MT) sequences exploit differences in relaxation between water transiently bound to macromolecules and water protons not bound to macromolecules. [19] Its use is mainly based on the background suppression of the normal brain. This sequence is also used for characterization of tuberculomas [Figure 6] and demyelinating lesions. MT ratios are reduced in demyelinating lesions. [20],[21] The disadvantage of MT sequence is the need for a prolonged RF exposure to the patient.
Magnetic resonance angiography may be done with noncontrast or contrast-enhanced techniques. [22] Noncontrast MRA is the flow-based time-of-flight angiography (TOF) or phase contrast (PC-MRA) sequence which uses the inherent motion of blood to generate signals. 3D TOF MRA is more sensitive for fast arterial flow and produces high spatial resolution images of intracranial arteries [Figure 2]. Tissues with short T1 relaxation time (fat, blood and tissues that take up contrast) may simulate vessels and a comparison with T1W images may be required. Presence of coils, clips and susceptibility artifacts cause signal loss. Signal loss due to turbulence or very slow flow may cause overestimation of stenosis and artifacts in the depiction of aneurysms. PC-MRA provides data on flow velocity and direction also. Contrast-enhanced MRA (CE-MRA) relies on the T1 shortening effects of gadolinium (Gd) chelate in the blood. The basic sequence used for CE-MRA is the ultra-short TR/TE 3D spoiled GRE sequence. The CE-MRA reduces or eliminates most of the artifacts associated with the TOF and PC angiography. It can be acquired in any plane, and large areas can be covered. It is possible to perform a time-resolved examination similar to catheter angiography (Time-Resolved Imaging of Contrast KineticSCombining, GE healthcare). This is good for the dynamic depiction of vessels which is required for the imaging work-up of vascular malformations. [22] MRV may also be done by optimizing the TOF, PC or CE-MRA for venous imaging [Figure 3]. Magnetic resonance spectroscopy differs from other sequences in that it yields graphs of the spectrum instead of images. [23] MR spectrum consists of resonances or peaks that represent signal intensities as a function of frequency (commonly expressed as parts per million). Resonance frequency of some important metabolites are as follows: (N-acetyl-asparate at 2.0 ppm, creatine/phosphocreatine [Cr] at 3.0 ppm, choline compounds [Cho] at 3.2 ppm, Myo-inositol [mI] at 3.5 ppm, Lactate [Lac] at 1.35 ppm and free lipids [Lip] at 1.3 and 0.9 ppm). Single voxel or multi voxel MRS can be done. Single voxel spectroscopy receives signal of a volume limited to a single voxel and the acquisition is fairly fast (1-3 min). Two types of sequences are commonly used (Point-RESolved Spectroscopy and STimulated Echo Acquisition Mode). Multi-voxel spectroscopy or chemical-shift imaging, obtains spectroscopic information from multiple adjacent voxels over a large volume of interest in a single measurement. The main clinical applications of MRS are in the evaluation of tumors [Figure 7], demyelinating diseases, infections, and metabolic disease. [23]
Functional magnetic resonance imaging is used to map the eloquent cortex associated with language, memory, and motor networks. [24] This information is very useful in the presurgical planning of patients with brain tumors, AVMs, seizures and cerebro-vascular diseases. DTI may be coupled with fMRI to delineate the associated white matter tracts. The goal is to minimize damage to these networks by mapping the location of areas of cortex and the white matter tracts relative to the surgical lesions. The basic sequence used for fMRI is the blood oxygen level dependent (BOLD) sequence. BOLD imaging involves using the paramagnetic properties of deoxyhemoglobin to image the local tissue oxygen concentration. The activated brain cortex during a particular activity changes the local deoxyhemoglobin concentration producing a signal on BOLD images. The four common paradigms used for mapping are language comprehension, memory, eye-movement paradigm and hand motor paradigm. Diffusion tensor imaging detects the diffusivity of water molecules in various directions. [25] In white matter tracts, the direction of maximum diffusivity is along the white matter tract orientation and is called anisotropic diffusion. White matter tracts can be delineated by postprocessing the DTI images (tractography) [Figure 8]. DTI sequences are being used to probe white matter structures in a number of disorders including epilepsy. [25] DTI sequences are used for surgical planning of cerebral neoplasms to assess the integrity and location of white matter tracts in relation to the tumor. [9] Another application of DTI is in investigating the white matter damage in traumatic brain injury, infection related white matter diseases like HIV encephalitis, and progressive multifocal leukoencephalopathy. [26],[27]
Perfusion MRI is used to measure the cerebral perfusion. [28] It may be done using exogenous contrast agents (dynamic susceptibility contrast MR [DSC-MR] or dynamic contrast- enhanced MR [DCE-MR] perfusion scan) or endogenous contrast agents (arterial spin labeling [ASL] perfusion technique). Dynamic susceptibility contrast MR Perfusion is a technique in which the first pass of a bolus of the gadolinium-based contrast agent through brain tissue is monitored by a series of T2- or T2*-weighted MRI. The method is based on a fast echo-planer imaging acquisition. The susceptibility effect of the paramagnetic contrast agent leads to a signal loss in the signal intensity - time curve. From this data, parametric maps of cerebral blood volume (CBV) and flow (CBF) can be derived. DCE-MR Perfusion is based on the acquisition of serial T1W images before, during, and after administration of gadolinium based MR contrast media. The resulting signal intensity-time curve reflects a composite of tissue perfusion, vessel permeability, and extravascular-extracellular space. The most frequently used metric is k trans. It can have different interpretations depending on blood flow and permeability. ASL technique uses magnetically labeled blood as an endogenous tracer [Figure 9].
Dynamic susceptibility contrast-derived relative CBV is the most widely used and robust method to evaluate brain tumors. Disadvantages of this technique include the difficulty in determining absolute quantification, the susceptibility artifacts and user dependence. With DSC MR perfusion visualization and quantification of the whole brain can be done in less than a minute of acquisition time. This is much less when compared with ASL, which has much longer scanning times. The dynamic contrast-enhanced-technique allows the quantitative assessment of the blood-brain barrier and microvascular permeability. It yields a complete assessment of tumor angiogenesis. The drawbacks of DCE-MRI include the complexity of image acquisition and postprocessing. The main advantage of ASL technique is that there is no need for a gadolinium-based contrast agent. This enables easier repeated measurements. Permeability, a major confounder in measuring relative CBV with the DSC MR technique, is of less concern in ASL. [28] Currently, ASL can provide values only of CBF. ASL sequences have a lower signal-to-noise ratio compared to DSC and DCE-MR perfusion and provide images of lower temporal and spatial resolution.
Epilepsy For epilepsy, in addition to the basic sequences, a GRE or SWI sequence is added to see any hemorrhagic or calcific foci. Postcontrast T1W imaging is done to observe any enhancing lesion. T1W (preferably isotropic 3D T1W imaging) sequences are superior for evaluating cortical thickness and grey-white matter interface. T1W images should be carefully reviewed for cortical dysplasias and grey matter heterotopias. T2 FLAIR images should be carefully assessed for cortical and subcortical hyperintensities, which can be very subtle. For possible mesial temporal sclerosis, thin T2W and T2W FLAIR coronal sequences should be taken perpendicular to the long axis of the hippocampus. Multiple sclerosis Multiple sequences are used, including T2W, FLAIR, precontrast and postcontrast T1W sequences. These show different sensitivities depending on the anatomic location of lesions. T2 FLAIR images provide the highest sensitivity for detection of supratentorial lesions close to the CSF, especially in juxtacortical and periventricular white matter, but are less sensitive for posterior fossa lesions. T2W sequences are more sensitive for detecting infratentorial lesions. DIR pulse sequence has a high sensitivity for lesions in both these anatomical locations. [8] These sequences also show cortical lesions well. Additional sagittal sections should be acquired for visualization of the corpus callosal, periventricular and cervical cord lesions. Perfusion MR increases the confidence of diagnosis in tumefactive MS by demonstrating hypoperfusion. MRS shows increased glutamine and glutamate peaks in tumefactive MS which are not seen in aggressive tumors. [19] Intracranial infections Diffusion MR has utility in the identification of brain abscess and infected extra-axial collections; it can demonstrate cortical and basal ganglionic lesions in Creutzfeldt Jacob disease that are often otherwise invisible. [29] Postcontrast imaging should be obtained to observe a leptomeningeal enhancement or any focal brain lesions. For brainstem and midline lesions, a sagittal postcontrast T1W sequence is useful. MRS is of utility in making a differential diagnosis between toxoplasmosis and lymphoma. [30] Suspected venous sinus thrombosis A TOF, PC or CE-MRV is done. In addition to the basic sequences, a GRE or SWI sequence is obtained to look for any hemorrhage. Patients suspected of having cavernous sinus thrombosis require postcontrast T1W coronal scans. Stroke The most important sequence is DWI which confirms or excludes acute ischemic changes. An MRA sequence should be added for evaluation of the intracranial vessels, and if required for carotid vessels also. SWI is also done to observe any hemorrhagic transformation. Perfusion MRI may be done to visualize the salvageable "penumbra." Tumor Postcontrast isotropic 3D T1W imaging should be done for a multiplanar reconstruction. MRS is usually done to detect the presence of a choline peak. In the cases where the tumor is closely related to vascular structures, a MRA and/or MRV sequence is also added.
Magnetic resonance imaging of the spine may be done for abnormalities of the spinal cord or for extramedullary/bony abnormalities. Imaging is required in sagittal and axial planes [Figure 10]. Sagittal T1W and FSE T2W sequences are routinely used. Axial T2W images are preferred to evaluate the neural foramina for the presence of an osteophyte or lateral disc herniation. In the cervical spine, T2 GRE sequences are useful for differentiating osteophytes from disc bulges.
Steady-state sequences with balanced gradients (i.e. True FISP, FIESTA and CISS), can be used for visualization of the intradural cervical roots in combination with axial FSE T2W or the 3D GRE technique. [31] This technique is also used for imaging brachial plexus injuries [Figure 11].
Fat suppressed sequences are helpful for detection of vertebral metastatic diseases and marrow edema. The most commonly used MRI techniques for fat suppression includes frequency-selective fat saturation, inversion recovery, water excitation, or a combination of these techniques. Limitations of these techniques include failed or inhomogeneous signal suppression. To overcome these problems, fat-water separation techniques are now being utilized based on the Dixon principle. [32] Further modifications have resulted in an IDEAL technique (iterative decomposition of water and fat with echo asymmetry and least-squares estimation, GE Healthcare), which is being used to obtain a homogenous fat suppression in thoracic and cervical spine. The indications for the gadolinium-enhanced imaging are infective, inflammatory and neoplastic lesions. To appreciate postcontrast enhancement in vertebral lesions, fat-suppressed T1W sequences are essential; however, for spinal cord lesions, postcontrast T1W images are sufficient. For degenerative disc disease, postcontrast imaging is generally not needed except for postoperative cases. Whole spine screening MRI may be done with sagittal T1W or T2W sequence using combined head and spine phased-array coils. The entire spine is imaged by moving the patient table stepwise without repositioning the patient or changing the coil. Indications for such studies include suspected multilevel pathology in patients with tuberculosis of the spine, multiple myeloma, metastatic disease, degenerative disease, drop metastasis from brain tumors, preoperative evaluation of patients with neurofibromatosis type 1 and spinal dysraphism. Spinal DWI has been used for distinguishing between acute benign osteoporotic and malignant vertebral compression fractures. However, as discussed earlier, most DWI techniques are EPI based and prone to image distortion. Non-EPI based DWI and DTI methods are now being used to detect the presence of spinal cord infarction and to evaluate the integrity of white matter tracts in patients with spinal cord injury. [33],[34] The role of DTI is being explored to assess the microstructural changes of the white matter caused by cord compression. It can be a useful tool to determine the pathological spinal cord levels in multilevel cervical spondylotic myelopathy. [35] The information from orientation entropy-based DTI analysis would help in deciding the optimal surgical strategy. Magnetic resonance neurography is being used for the demonstration of peripheral nerve anatomy and pathology. [36] Three-dimensional anatomic nonselective MR neurography and anatomic nerve-selective MR neurography sequences are available. The sequences are termed differently by various vendors as SPACE (Siemens), Cube (General Electric Healthcare) and VISTA (Philips). [36] They can be obtained with and without fat suppression and can generate various contrasts depending on the clinical situation: T1, PD, T2, spectral adiabatic inversion recovery (SPAIR), and STIR. The 3D SPAIR SPACE is used in extremities due to its higher signal-to-noise ratio, 3D STIR SPACE is used for plexuses due to better and more homogeneous fat suppression and 3D T2 SPACE is used for spine evaluation (also referred to as MR myelography).
We would like to acknowledge Mr. Dharmendra, MR technician, Department of Radiodiagnosis, SGPGIMS, Lucknow for his help in the image collection.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 11]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||
