Optimizing Musculoskeletal MR
Presented by Virtual Visiting Guest Lecturer:
Mini N. Pathria, M.D.
University of California San at Diego
Peer Review: Vincent B. Ho, MD and Maureen Hood, BS, RT(R)(MR)
HTML Conversion: James G. Smirniotopoulos, MD
PLEASE READ THE DISCLAIMER
Figure(click here)
Title slide
Figure(click here)
High quality images require adequate anatomic coverage of the area of interest and sufficient spatial and contrast resolution to visualize morphologic abnormalities and abnormalities in tissue composition. Adequate SNR is also essential. Improvements in image quality are often a trade-off against imaging time.
Figure(click here)
Images which contain adequate diagnostic information to allow accurate interpretation are the goal of the MR examination. In today's economic climate, it is important that the study also be completed in a reasonable amount of time.
Figure(click here)
The initial step in our protocol design is selecting patient positioning, FOV and coil because these decisions do not significantly impact on sequence selection.
Figure(click here)
Most examinations are obtained supine. We routinely position the patient prone if there is pain associated with the supine position (e.g. buttock mass, scapular pain). Lesions located posteriorly show less distortion when they are placed in a non-dependent position, thereby minimizing compression of the abnormal area.
Figure(click here)
We routinely use large FOV localizers and film them, frequently switching to smaller FOV and more local coils as the examination progresses.
Figure(click here)
Review of other imaging studies prior to writing protocol ensures coverage of the area of abnormality. This patient with a giant cell tumor of the tibia was imaged using the routine knee protocol. The lesion was not completely evaluated on this study. Review of the patient’s xrays prior to the MR study would have avoided this situation.
Figure(click here)
Routinely marking areas of palpable mass or maximal discomfort with a bath oil bead ensures covering the area of interest and helps the radiologist identify the lesion.
Figure(click here)
Markers are particularly helpful for lesions that are difficult to differentiate from surrounding tissues, such as this subcutaneous lipoma which was easily palpable but difficult to identify on the MR images.
Figure(click here)
A marker has been placed over a forearm mass which is difficult to see on this large FOV T1-w SE image.
Figure(click here)
The subsequent small FOV fat-suppressed T1-w SE image following intravenous Gd shows enhancement of the mass, though poor fat suppression still makes it difficult to see the lesion.
Figure(click here)
Large FOV localizers are helpful for evaluating long lengths of bone. This T1-w SE localizer in a patient with a proximal tibial osteosarcoma shows a skip lesion or metastasis in the distal femur.
Figure(click here)
Coronal localizers of the spine include the retroperitoneum and portions of the upper abdomen and pelvis, allowing a crude look at these soft tissue regions.
Figure(click here)
We routinely use a large FOV fast STIR sequence as our localizer sequence because of its high contrast resolution. We film the localizer, and then use small FOV SE or FSE sequences to look at the morphology in detail. This fast IR localizer of the knee shows an area of trabecular edema ("bone bruise") in the posterior tibia.
Figure(click here)
The area of bone bruise is not seen on the small FOV PD-w FSE sequence but the anatomic detail of the menisci is well shown.
Figure(click here)
High spatial resolution is necessary when imaging small structures but not every sequence in the protocol needs to have high spatial resolution. For example, large FOV fast IR sequences are used early in our examinations to localize the lesion because of their high contrast resolution. We do not emphasize high spatial resolution with this sequence.
Figure(click here)
It is more efficient to obtain high spatial resolution on sequences that are fast and that have high signal to noise (e.g. GE, T1, T2 FSE). Using small FOVs is typically the easiest way of improving the spatial resolution of the examination.
Figure(click here)
Spatial resolution is a function of the size of the voxel (3D pixel) being imaged. The factors that determine voxel size and therefore affect spatial resolution are slice thickness, FOV, number of FE steps (usually preset at 256), and the number of PE steps. The smaller the voxel, the higher the spatial resolution of the study.
Figure(click here)
High spatial resolution is necessary for imaging small structures such as the triangular fibrocartilage and intercarpal ligaments of the wrist.
Figure(click here)
Use of high SNR sequences and a well-designed surface coil ensure adequate signal to noise.
Figure(click here)
This image shows a tear of the ulnar collateral ligament at the 1st MCP joint, with interposition of the adductor aponeurosis. This injury, known as a Stener lesion, represents one form of a gamekeeper’s thumb and requires surgical repair. Seeing small structures like these requires high spatial resolution.
Figure(click here)
This image of the wrist is obtained at low spatial resolution. The tendons are difficult to separate and the bone margins are poorly defined.
Figure(click here)
An image of the same wrist at a higher spatial resolution allows visualization of the individual tendons.
Figure(click here)
Improvements in spatial resolution can be made by decreasing slice thickness (limits anatomic coverage obtained by a fixed number of slices), increasing PE steps (increases imaging time) or by decreasing FOV (limits anatomic coverage).
Figure(click here)
Small voxels have less SNR so the high spatial resolution image may be too noisy to be diagnostically useful if the voxels are too small for adequate signal. Small voxels are particularly a problem on low field imaging systems.
Figure(click here)
Fortunately, adequate SNR is easily obtained when imaging the musculoskeletal system due to the high concentration of fat in the mesenchymal tissues. There are a variety of strategies available if the images do show a large amount of noise.
Figure(click here)
Poor SNR can be recognized by a generalized “graininess” to the image. Noise results in an irregular, grainy appearance of the air surrounding the body and of other structures that are usually homogeneous.
Figure(click here)
Using higher NEX increases SNR but is very time-consuming. Increases in NEX increase SNR but only by the square root of NEX.
Figure(click here)
Doubling the NEX doubles the time of the examination but only increases the SNR by a factor of 1.4 (square root of 2).
Figure(click here)
Signal to noise is improved without any disadvantages by using the proper coil for the body part being imaged. There is no major disadvantage to using a local coil other than limited anatomic coverage and limited depth of penetration. This is an image of the pelvis obtained with the body coil.
Figure(click here)
This image of the left hip of the same patient was obtained using an anterior surface coil. There are improvements in both the spatial resolution and SNR but the field of coverage is much smaller than with the body coil.
Figure(click here)
A T1-w SE image of the arm in a patient with a cystic synovial cell sarcoma is degraded by noise. This image was obtained using the body coil.
Figure(click here)
The same arm imaged with a surface coil shows significant improvement of image quality. Switching to a more sensitive coil is an excellent way of increasing signal to noise.
Figure(click here)
Certain sequences have more SNR, typically due to greater signal from fat. Both T1-w and PD-w SE sequences generally have excellent SNR for musculoskeletal imaging.
Figure(click here)
T2-w SE images frequently have low SNR, particularly when small voxels are imaged. Images with inadequate SNR will not show contrast differences between different tissues.
Figure(click here)
Tissue parameters imaged with MR imaging include proton density, T1 relaxation, T2 relaxation and flow.
Figure(click here)
Image contrast in musculoskeletal tissues is very high due to the presence of a variety of tissues and the high contrast between fat and muscle and between fat and collagenous tissues.
Figure(click here)
Differences in proton density between various body tissues are minimal. It is differences in T1 and T2 relaxation times between the various tissues that affords high contrast resolution with MR imaging.
Figure(click here)
The proton density difference between muscle and fat is approximately 1% but their T1 relaxation times are very different. This patient had polio and shows severe muscle atrophy of the left lower extremity. There is marked asymmetry in the signal from the two legs.
Figure(click here)
T1 relaxation refers to the realignment of protons with the main magnetic field following removal of the RF pulse. T1 is the exponent of this recovery curve.
Figure(click here)
T2 relaxation refers to dephasing between spins within the XY plane and is always more rapid than T1 relaxation. T2-w sequences typically have high contrast between lesion and muscle due to the increased water content of most pathologic tissue.
Figure(click here)
Contrast resolution is also determined by the imaging parameters we select. Contrast resolution is heavily dependent on the type of pulse sequence employed and TR, TE, TI and flip angle.
Figure(click here)
With proper selection of imaging parameters, pathology such as this quadriceps rupture can be seen in exquisite detail.
Figure(click here)
Multiple pulse sequences are routinely employed for musculoskeletal imaging.
Figure(click here)
The most commonly used are the SE sequences, which have been shown to be highly accurate for imaging internal derangement of joints and other pathology.
Figure(click here)
Multiple optional techniques can be applied in conjunction with the previously mentioned pulse sequences.
Figure(click here)
Spin echo sequences have been the most frequently used sequences for musculoskeletal imaging. Standard SE sequences using T1, PD and T2-weighting are typically employed in most routine protocols.
Figure(click here)
This image shows an example of the T1-w (top), PD-w (middle), T2-w (bottom) sequences in a patient with a tear of the supraspinatus tendon and a large degenerative cyst of the greater tuberosity.
Figure(click here)
The T1-w sequence is good for looking at marrow abnormalities but affords less soft tissue contrast than the T2-w SE sequence. This T1-w SE image shows a large hemorrhagic sarcoma within the right buttock.
Figure(click here)
The lesion can be easily differentiated from muscle on the T2-w SE image.
Figure(click here)
The T1-w SE image shows a large cyst in the greater tuberosity. There is excellent contrast between the cyst and the bright fatty marrow.
Figure(click here)
The signal of the subacromial fluid and the tuberosity cyst increase with T2 weighting. Overall SNR is less than on the T1-w image.
Figure(click here)
In standard SE imaging, imaging time is a function of TR, number of PE steps and NEX. (Imaging time = TR * PE * NEX) Increasing any of these 3 parameters increases imaging time. (The number of slices is not a factor that changes imaging time because all the slices are obtained during the TR interval).
Figure(click here)
In SE imaging, the most efficient way to decrease imaging time is to reduce NEX in sequences with high SNR or to decrease the number on PE steps on sequences with low SNR.
Figure(click here)
Decreasing the phase encoding steps in half reduces the imaging time by 50% but also diminishes the spatial resolution of the examination.
Figure(click here)
GE sequences are faster than SE due to their shorter TR but do change the type of image contrast and have magnetic susceptibility artifacts. Using sequences that obtain more information per TR, such as FSE and fast IR techniques, are an increasingly popular alternative approach for faster imaging.
Figure(click here)
Echo-planar MR promises to provide even more rapid imaging but is not yet widely clinically utilized for musculoskeletal imaging.
Figure(click here)
Gradient echo sequences are a family of pulse sequences (FLASH, SPGR, FISP, GRASS, etc.) that use small flip angles (less than 90 ), short repetition times (TR) and gradient reversal (rather than the 180 refocusing pulse used in SE imaging) to refocus spins.
Figure(click here)
All gradient echo sequences, however, share some common advantages: high signal-to-noise per unit time, short imaging times, the ability to obtain thin (less than 1mm) slices and 3D imaging capability. These features can be extremely advantageous in musculoskeletal imaging, particularly for obtaining kinematic data.
Figure(click here)
T2 decay is faster with GE sequences than with SE sequences because there is no correction for local field inhomogeneities since the 180o pulse (which will correct for approximately 50 ppm inhomogeneity) is not applied. There are other factors which lead to the shortened T2, therefore it is designated as T2* with GE imaging.
Figure(click here)
Susceptibility effects which may cause signal loss at interfaces between tissues with differing magnetic susceptibilities. The interface between trabecular bone and marrow results in lower signal within the bone than with SE sequences.
Figure(click here)
GE images are helpful for identifying blood products. This T2-w SE image of the knee in a patient with PVNS shows low signal areas of hemosiderin, particularly in the infrapatellar fat pad.
Figure(click here)
The loss of signal around the hemosiderin deposition is more pronounced than on the SE image. GE imaging is very sensitive to magnetic susceptibility effects due to heterogenous distribution of ferromagnetic or paramagnetic substances.
Figure(click here)
GE sequences are less “reliable” than SE sequences for assessment of marrow pathology. This enchondroma is well seen on this GE sequence.
Figure(click here)
The same lesion is almost impossible to identify on this GE sequence. In my experience, it is more difficult to reliably predict signal characteristics of tissues with GE imaging and I use GE imaging primarily to assess articular pathology.
Figure(click here)
Image contrast on GE images depends on TR and TE (same effects as with SE imaging) and on the flip angle (small flip angles are more T2-weighted, flip angle approaching 90 produce T1 weighting).
Figure(click here)
GE sequences for useful for evaluating specific internal derangement of joints, particularly for assessment of hyaline cartilage and osteochondritis dissecans. This fat-suppressed SPGR image of the knee shows a bony bar across the posterior femur due to a prior physeal injury.
Figure(click here)
In fast spin-echo imaging (FSE), multiple echoes (usually 8) are collected during each 90 degree pulse.
Figure(click here)
The effective TE is the TE for the pulse that does not have a phase-encoding gradient.
Figure(click here)
Because multiple lines of K space are encoded for each TR FSE offers considerable time savings compared to SE. FSE images can be obtained approximately 4 times more quickly than conventional SE. The time saved is usually partially "traded in" for better spatial resolution (more PE steps) or increased SNR (higher NEX).
Figure(click here)
Selecting how to encode the echoes in K-space allows one to select image contrast. The echoes in the middle of K-space largely determine the type of image contrast, whereas the periphery of K-space contains high spatial frequency information and determines spatial resolution.
Figure(click here)
On PD-w FSE images, the early echoes are encoded in the center of K-space.
Figure(click here)
On T2-w FSE images, the late echoes are encoded in the center of K-space.
Figure(click here)
FSE sequences have become very popular and are now routinely used in many anatomic areas, including the spine.
Figure(click here)
The myelogram effect of the T2-w FSE image makes it ideal for assessment of the CSF space and fluid-containing articulations.
Figure(click here)
However, the FSE sequence has some disadvantages. For musculoskeletal imaging, the major disadvantages of FSE are blurring of small structures and persistent high signal from marrow fat, even on the T2-w images.
Figure(click here)
Bright fat is less of a problem with very long TE FSE imaging as seen in this example of a patient with a chondrosarcoma of the pelvis. The high signal of fat can be eliminated with the addition of selective fat suppression.
Figure(click here)
This T1-w SE image shows a subcutaneous lipoma.
Figure(click here)
The T2-w FSE image is almost identical and there is minimal signal loss in the fat, different than is seen with SE imaging.
Figure(click here)
Marrow fat also tends to stay bright on the T2-w FSE sequences, making distinction of fat from abnormal tissue more difficult. This T1-w image shows an undisplaced fracture of the left anterior acetabulum.
Figure(click here)
The fracture can not be seen on the T2-w FSE image and the high signal of the marrow masks any edema around the fracture site.
Figure(click here)
This T1-w SE images also shows abnormal signal within the acetabulum due to a insufficiency fracture.
Figure(click here)
The T2-w FSE image fails to show the lesion because the high signal of marrow edema can not be seen against the background of bright marrow fat. Note that there is less artifact around the left sided total hip arthroplasty than on the conventional SE image.
Figure(click here)
A T2-w FSE image of the spine in a patient with a proven hemangioma shows minimal increased signal within the lesion.
Figure(click here)
The addition of fat suppression to the T2-w FSE sequences enhances its sensitivity to small changes in free water concentration, particularly within the marrow space. The hemangioma is now clearly visible.
Figure(click here)
Another disadvantage of FSE imaging is blurring. This conventional SE image shows a tear of the posterior horn of the medial meniscus.
Figure(click here)
On the FSE image, the tear is difficult to see due to blurring. (Courtesy of Dr. Mark Anderson, Charlottesville VA)
Figure(click here)
Blurring is most noticeable in the phase-encoding direction and is due to exponential T2 decay. This blurring can be minimized by selecting a short echo train and decreasing the echo interval.
Figure(click here)
A PD-w FSE image with fat suppression shows a complex tear of the medial meniscus. While large tears are usually seen with FSE imaging, the sensitivity for meniscal pathology is reported to be significantly less than with conventional SE techniques.
Figure(click here)
Magnetization transfer contrast is another technique that can be applied to musculoskeletal imaging, particularly for assessment of cartilage.
Figure(click here)
Differentiation between cartilage and joint fluid is enhanced by using MTC.
Figure(click here)
The principle of MTC is imaging the difference in T1 between water coupled with macromolecules and uncoupled water.
Figure(click here)
The principles of MTC imaging are well described in the article referenced on this diagram.
Figure(click here)
MTC has not been widely used for musculoskeletal imaging. At this time, its major utility appears to be in the field of cartilage imaging.
Figure(click here)
This axial MTC image of the knee shows excellent depiction of the patellar cartilage and high contrast between cartilage and joint fluid. (Courtesy of Dr. S. Eilenberg, San Diego CA)
Figure(click here)
There are a variety of methods available for obtained images where the signal from fat is suppressed. The most popular of these are short tau inversion recovery (STIR) and selective fat suppression techniques.
Figure(click here)
Adding fat-suppression to SE sequences results in lowered contrast resolution for most structures. However, we use these routinely for visualization of "black stuff", such as menisci, ligaments and tendons.
Figure(click here)
STIR is less dependent on field inhomogeneities and excellent fat suppression can be obtained throughout the body. This STIR image of the ankle shows thickening and abnormal signal within the Achilles tendon.
Figure(click here)
The selective fat suppressed sequences are much more sensitive to field inhomogeneities and the fat suppression is often less uniform.
Figure(click here)
STIR, like all inversion recovery sequences, uses a 180 RF pulse to invert longitudinal magnetization, followed by the conventional SE pulse train (90 -180 ). STIR sequences provide images with selective suppression of fat (and all other tissues with T1 = to fat).
Figure(click here)
By choosing an appropriate inversion time (TI) (TI=time between the initial 180 and 90 pulses) it is possible to null out signal from fat. STIR is not used following contrast enhancement because the enhanced lesion’s T1 may be shortened , making it equal to the T1 of fat. If this occurs, the pathologic tissue's signal is nulled, just like fat.
Figure(click here)
The STIR sequence allows for an additive effect of prolonged T1 and T2. The feature is in contrast to SE images where prolonged T1 and T2 times have opposing effects. Since most pathology has prolonged T1 and T2 compared to normal tissue, this additive effect increases the contrast between lesions and normal tissue on STIR images compared to SE images.
Figure(click here)
STIR is very sensitive for detecting marrow abnormalities, as normal fatty marrow is nulled, while pathology remains high in signal intensity. In this patient with multifocal osteomyelitis, the abnormal marrow in the distal right femur and proximal left tibia is bright, as is the abscess surrounding the femur.
Figure(click here)
This T2-w SE image of the distal femur in a patient with an osteosarcoma shows the tumor and its extension into the soft tissues through the posterior femur.
Figure(click here)
The tumor is more conspicuous on the STIR image and its soft tissue component is easier to identify. Musculoskeletal tumors are usually more conspicuous on STIR than on SE images and STIR usually detects a greater volume of abnormality. It is often more difficult to differentiate tumor from peritumoral edema on STIR images.
Figure(click here)
However, STIR imaging does suffer from some significant disadvantages including decreased signal-to-noise compared with SE imaging, decreased spatial coverage, and loss of anatomic detail due to the loss of fat planes.
Figure(click here)
The high signal around the medial epicondyle of the elbow is well seen on the STIR image but it is difficult to identify the ligaments and tendons. The anatomic planes are difficult to see due to fat suppression.
Figure(click here)
This T1-w image shows a lipoma of the calcaneus, containing a central focus of low signal calcification.
Figure(click here)
The fatty lipoma can no longer be seen on the STIR image.
Figure(click here)
Selective fat suppression uses the difference in the precessional frequency between fat and water to suppress signal from fat. Chemical shift selective saturation techniques (e.g.. Chemsat) take advantage of this difference by applying a frequency selective presaturation pulse immediately prior to the imaging sequence.
Figure(click here)
The presaturation pulse effectively eliminates the longitudinal magnetization of fat protons, eliminating signal from fat. In practice, the suppression is often imperfect. In this example, there is inadequate suppression of the tibia and soft tissues inferior to the knee joint.
Figure(click here)
A T1-w SE image of the soft tissues posterior to the scapula is shown.
Figure(click here)
The fat suppressed image shows loss of signal within the subcutaneous fat. Note the uniformity of the fat suppression, achieved with an water bag adjacent to the skin.
Figure(click here)
Use of water bags, silicone baffles or commercially available saturation pads is helpful for obtaining uniform suppression. This image of both feet using a saline bag under the right foot shows better fat suppression adjacent to the saline bag than on the contralateral side.
Figure(click here)
IV bags or travel pillows filled with Kaopectate Plus are inexpensive and function as excellent saturation pads due to the low signal of Kaopectate Plus. This image of the foot was obtained with a Kaopectate Plus filled bag on the dorsum of the foot. There is inadequate suppression of the plantar surface.
Figure(click here)
Fat saturation techniques can be employed following contrast enhancement because Gd is not suppressed, as it is with STIR. Enhancement of multiple neurofibromas can be seen on this T1-w SE fat suppressed image.
Figure(click here)
Another image in the same patient shows neurofibromas. There is poor fat suppression of the right thigh so the normal marrow appears brighter than on the left. Inhomogeneous fat suppression creates difficulty distinguishing gadolinium enhancement from fat, particularly within the marrow.
Figure(click here)
The major indications for fat suppression are listed.
Figure(click here)
Fat suppression can confirm the fatty nature of a mass. This T1-w SE fat suppressed image of a lipoma anterior to the right femoral neck shows low signal within the mass. The small low-density structures at the edge of the mass represent areas of dystrophic ossification within the lesion.
Figure(click here)
A T1-W SE image shows a lipohemarthrosis and a fracture of the posterior tibial plateau.
Figure(click here)
The PD-w FSE image with fat suppression shows circular areas of low signal within the effusion, representing fat globules suspended within hemorrhagic effusion.
Figure(click here)
T1-w SE sequences with fat-sat are routinely used following Gd-DTPA enhancement because they exhibit high contrast between lesion and fat. Enhancement of a pretibial varix is easily differentiated from the subcutaneous tissues.
Figure(click here)
This T1-w SE fat-suppressed image shows a postoperative desmoplastic proliferation involving the neck.
Figure(click here)
This T1-w SE image was obtained following intraarticular injection of a dilute solution of Gd-DTPA. It is difficult to distinguish gadolinium from fat.
Figure(click here)
The fat suppressed image affords differentiation between the darker fat and the bright gadolinium. The bright area overlying the deltoid tendon insertion is caused by inadequate fat suppression at the curved interfaces with air.
Figure(click here)
Using fat suppression alters the contrast scale of the image. This T1-w SE image shows enlargement of both Achilles tendons due to xanthomatosis.
Figure(click here)
The T1-w SE image obtained following fat suppression shows more of the internal architecture of the enlarged tendons. The contrast range is expanded, similar to manipulating window width with CT.
Figure(click here)
This T1-w SE image of a chondrosarcoma involving the right ilium shows marrow and soft tissue involvement.
Figure(click here)
Following fat suppression, the internal architecture of this cartilage lesion can be identified. (Courtesy of Dr. Steven Eilenberg, San Diego CA)
Figure(click here)
Title slide
