Docstoc

MRI of the Shoulder (DOC)

Document Sample
MRI of the Shoulder (DOC) Powered By Docstoc
					MRI Clinical Applications II                                                             Joseph Castillo




3         MRI of the Shoulder

This section introduces a case study to address learning outcome 1 and both learning outcome 2 and 3.
Learning outcome 3 – evaluate the advantages and disadvantages of introducing protocol changes- is
addressed by discussing the introduction of MR arthrography in the investigation of rotator cuff tears.
Shoulder images related to this case study are presented at the end of the section.


Normal MRI Anatomy


The shoulder is composed of three joints: glenohumeral, acromioclavicular and
scapulothoracic articulations. A fourth joint, the sternoclavicular joint is associated
with shoulder and joint mechanics (John V Crues III et al., 1999). The glenohumeral
joint allows a greater range of movement than any other joint in the body. This is
facilitated by deepening of the shallow glenoid fossa by the fibrocartilaginous labrum.
The capsule is strengthened anteriorly by several thickenings, namely the anterior
and posterior bands of the inferior glenohumeral ligament, the variable middle
glenohumeral ligament and the smallest superior glenohumeral ligament.                             The
tendon for the long head of the biceps originates from the supraglenoid tubercle,
passes intra -articularly and extends into the bicipital groove between the
supraspinatus and subscapularis tendon. A synovial sheath surrounds the tendon,
which is an extension of the synovial lining of the glenohumeral joint.

The coracoacromial arch has a fundamental role in stabilizing the unstable
glenohumeral joint.        The arch is formed by the humeral head posteriorly, the
acromion superiorly and by the coracoid process and coracoacromial ligament
anteriorly (Kaplan et al, 2001). Located within the coracoacromial arch from superior
to inferior are the subacromial/subdeltoid bursa, the supraspinatus tendon and
muscle, and the long head of the biceps tendon. The coraco acromial ligament is an
important structure as it limits anterior and superior motion of the humeral head and
overlying tendons. Anything that decreases the space within the coracoacromial
arch could lead to impingement symptoms.                    Different planes of imaging must
evaluate these structures that form or are contained in the arch.

The Acromioclavicular joint is a fibrocartilaginous synovial joint reinforced superiorly
and inferiorly by the acromioclavicular ligament.

The rotator cuff is a thick tendinous sheet that surrounds the humeral head. The cuff
is a confluence of four tendons, which are the insertions of the subscapularis,
supraspinatus, infraspinatus and teres minor muscles. These are well delineated on



                                                  28
MRI Clinical Applications II                                             Joseph Castillo


MR images as black structures on all pulse sequences.          The rotator cuff and
associated muscles are important for synchronous rotation of the humerus. The
subscapularis, infraspinatus and teres minor also provide internal and external
rotation. The cuff is important for the stability and normal function of the shoulder
joint (Resnick & Kang, 1997)

Signal intensities of the various structures i.e. muscles, ligaments tendons and bones
have already been described in Section 1.




Clinical Indications
   Shoulder Impingement syndromes

   Rotator Cuff Tears

   Calcific Tendonitis

   Glenohumeral Joint Instability

   Adhesive capsulitis (Frozen shoulder)

   Arthritis


Case Study - Rotator Cuff Pathology


Tear of the rotator cuff at the critical zone - an area of the supraspinatus tendon
underlying the anterolateral corner of the acromion - is a common example of chronic
rotator cuff pathology. A normal rotator cuff rarely tears except in a severe trauma.
Many rotator cuff tears develop in response to chronic micro trauma in the form of
mechanical impingement. The commonest is mechanical irritation of the rotator cuff
by hard bony projections. Rotator cuff pathology based on impingement produces a
spectrum of disorder, beginning with a normal cuff and progressing to a tear. As the
actual process of degeneration is continuous episode it would be difficult to
discriminate between degenerative fraying and partial tears on physical examination
or arthrography. MR imaging with its multiplanar capabilities and soft tissue contrast
is sensitive in showing rotator cuff degeneration and tear, structural abnormalities
frequently associated with impingement syndrome and other disorders that may
clinically mimic pathology of the rotator cuff (Uri, 1997). The MR signal intensities
differentiating the continuum from tendonopathy to complete tear is shown in Table
3.1 below.



                                            29
MRI Clinical Applications II                                                   Joseph Castillo




Cuff Disease               Short TE       Long TE        Liquid     in   Tendon
                                                         Bursae          Retraction

Normal                     Low            Low            No              No

Mild Degeneration          Intermediate   Low            No              No

Moderate Deg.              Intermediate   Intermediate   No              No

Partial Tear               Intermediate   High           Possible        No

Small Complete Tear        Intermediate   High           Possible        No

Full Thickness Tear        Intermediate   High           Yes             Yes



Table3.1: Differentiation of Rotator Cuff tendon Pathology (Modified from Crues III et al.,
1999)


Image optimization in MRI of the Shoulder
The shoulder joint due to its location relative to the isocenter of the bore and the
thorax places extraordinary technical demands on MRI. Since the shoulder joint is
located on the far lateral aspect of the upper torso, achieving high-resolution images
requires off-center positioning of the FOV.

The close association of the shoulder to the thorax causes motion artifacts due to
breathing. Instructing the patient to take shallow breathing and position the forearm
away from the chest eliminates these.

The axes of the shoulder are angled with respect to the primary axes of the scanner
and this requires manipulation of the gradient fields, unless one prefers to align the
patient with these fields (the patient however may not be amused). Accurate graphic
prescription is mandatory to achieve standardisation (not to confuse radiologists) and
reduce partial volume effects.

Manufacturers have successfully met most of these technical demands. Off-center
FOV and 2D oblique image acquisitions are now standard. 1.5T Scanners generally
provide improved SNR and decrease acquisition times when compared with low field
scanners. Nonetheless, attention must be given to technical details especially choice
of surface coils, artifact suppression and contrast resolution.




                                                 30
MRI Clinical Applications II                                               Joseph Castillo


Scan Plane


Optimal evaluation of the shoulder joint includes axial, coronal obliques and sagittal
obliques. Axial sections are obtained from the acromioclavicular joint through the
inferior glenoid margin using a coronal oblique localizer.   This view allows depiction
of labral and capsular structures and the long head of the biceps tendon as it course
down in the bicipital groove. This plane allows evaluation of the coracoid and its
relationship with the humerus and scapular neck (Uri, 1997). The axial images serves
as a localizer for prescribing the coronal oblique and sagittal oblique sections (See
images end of section).

Coronal oblique images are graphically prescribed parallel to the course of the
supraspinatus tendon. This view allows assessment of the subscapularis anteriorly
and the infraspinatus and teres minor muscle posteriorly. This view is optimal for the
evaluation of      the supraspinatus tendon, subacromial/subdeltoid bursa and
acromioclavicular joint (Uri, 1997).

Sagittal oblique images are graphically prescribed perpendicular to the supraspinatus
muscle and parallel to the glenohumeral joint. The boundaries of the FOV should
include from the most lateral aspect of the humeral head to the glenoid fossa. These
images allow evaluation of the coracoacromial arch, the shape of the acromion and
the rotator cuff (Resnick & Kang, 1997; Tsao & Mirowitz, 1997).       Masciocchi et al.
(1997) modified the sagittal oblique to assess better the coracoacromial arch. In this
technique, using the coronal oblique image as localizer, the sagittal planes are
inclined 10 degrees laterally. This technique demonstrates the whole coracoacromial
ligament without partial volume effect.




Coil Choice


Choice of coil is critical to shoulder MRI because it must provide good signal to noise
ratio. Because noise is inherent in the tissue being imaged, it is important that a coil
adequately covers the area of interest but cover as little unwanted tissue as possible.
Image quality of shoulder MRI depends significantly on improvements in radio
frequency coils.      Four-channel shoulder array has become available that permit
imaging with high resolution, small fields of view and thinner sections. Phased array
coils consist of two or more resonating loops. The output signal of each loop is fed



                                          31
MRI Clinical Applications II                                               Joseph Castillo


into an independent channel of the MRI system. Thus coils do not share noise as
long as they remain electrically isolated from each other. Combining quadrature
configuration with phased array will achieve good SNR at the required depth and
image quality with image homogeneity with the benefits of phased array coils
(Zlatkin, 1996)


Pulse Sequences



Conventional Spin-echo (CSE) sequences are still used because most experience
has been gained with them. Short TE spin echo such as Proton Density or T1 are
very sensitive for the detection of cuff degeneration, to demonstrate anatomical
detail, and for the evaluation of the marrow space and peribursal fat planes (Tsao &
Mirowitz, 1997; Uri, 1997). The T2W sequences are useful in evaluating the degree
of the disease. Because of the significant improvement in the speed of imaging using
Fast Spin echoes, are often used to evaluate the shoulder joint.      Studies cited by
Mirowitz (1993) have shown that tissue contrast is similar in fast spin echo imaging to
what is seen with conventional spin-echo imaging, but fat is more intense on T2-
weighted FSE. Therefore, as it becomes difficult to differentiate fat from fluid signal,
fat suppression technique is used.    The development of auto shimming techniques
allowed improved homogeneity of fat suppression away from the magnet center.
Potential disadvantage of the FSE technique include blurring of anatomic structures
on short TE images.

Gradient echo techniques also produce T1-weighted and T2-weighted sequences
and in less time than conventional spin echo sequences (Sahin-Akyar et al., 1998).
Ligaments and articular cartilage are well demonstrated with gradient echo
sequences, as is the fibrocartilaginous structure such as the glenoid labrum.
Gradient echo imaging can be performed using 2D or 3D technique. 3D imaging is
particularly suited in the assessment of small structures. Susceptibility artifacts, a
feature of gradient echo, could be advantageous to highlight subtle areas of
haemorrhage, but could also overestimate the size of osteophytes and miss marrow
pathology. They are useless in areas where metallic hardware in situ.

High spatial resolution is mandatory to depict the rather small structures found in the
shoulder joint such as the ligaments, tendons and labrum. In addition the even
smaller pathology affecting the shoulder joint such as partial tears, labral tears and
bony spurs require high spatial resolution.     This is achieved by using optimized
matrix. The imaging matrix determines voxel size in combination with the field of


                                          32
MRI Clinical Applications II                                             Joseph Castillo


view. Smaller voxels provide higher degrees of spatial resolution, allowing small
abnormalities and structures to be easily depicted.     At the same time, because
smaller voxels contain fewer protons, the signal-to-noise ratio is low. Thus optimum
imaging matrix must be balanced with adequate signal. In addition the higher the
number of phase-encoding steps the longer the scan time and the possibility of
patient movement is increased. In the shoulder joint (and other small joints) a small
FOV in the range of 12 to 14cm, high matrix in the range of 384 x 256 and slice
thickness in the range of 3 to 4mm is used. These factors are made possible by
using high gradients, and quadrature coils (Masciocchi et al., 1997, Lee & Lang,
2000).

Number of signal averages (NEX) are increased to 3 or even 4 to reduce artifacts
from respiration and to increase the signal-to-noise ratio. Again this increase the
scanning time and thus it is important to position the patient comfortably to minimize
the risk of movement during the long scans.


Use of Fat suppression in the Shoulder


Fat suppression is useful because it increase the conspicuity of an abnormality and
also provides assessment of bone marrow. This effect is prominent on all pulse
sequences especially T2w and T1w during MR arthroscopy. Fat suppression also
reduces chemical shift artifacts (Singson et al., 1996, Reinus et al., 1995). Our
protocol provides two types of fat suppression: 1) STIR and 2) Fat saturation. These
are explained in section 6 (Knee joint)


Additional Sequences


In the absence of native effusion and when conventional MRI findings are equivocal,
some centers prefer to perform either direct or indirect MR arthrography.        Good
results were reported in the determination of rotator cuffs, showing improved
sensitivity and increased conspicuity over conventional MRI. Gd-DTPA is especially
useful for evaluating the cuff undersurface and, when needed for distinguishing small
and partial tears from degeneration (Lee & Lang, 2000).          MR arthrography is
performed by first injecting small amount of iodinated contrast medium under
fluoroscopy to confirm intra-articular localization. 12 to 15 mL of saline/Gd-DTPA
mixture (1.0ml with 250 mL of saline) is then injected. The patient is then taken into
the MRI scanner and three planes T1W fat suppressed sequences are obtained. Fat


                                          33
MRI Clinical Applications II                                                Joseph Castillo


suppression in this case increase the conspicuity of the contrast medium and
decrease the signal intensity of the peribursal fat plane around the subacromial
subdeltoid bursa. Without fat suppression, the fat plane can mimic contrast and
leads to false interpretation of a cuff tear. The patient is positioned as before contrast
administration. Further images with the arm in abduction and various degrees of
humeral rotation may be taken to visualize subtle cuff defects. On the other hand,
indirect MR arthrography requires only an intravenous injection and then the patient
is scanned after a delay of 8 to 15 minutes (Allman et al., 1999). One study reported
that two readers improved their accuracy for cuff tears from 67% and 62% with
conventional MRI to 92% and 96% with indirect MR arthrography. Others found that
exercising the joint does not improve accuracy (Jaovisidha et al., 1999, Brenner et
al., 2000). This makes the use of indirect MRA more appropriate to our situation.
Still there is a substantial increase in examination time. Despite these studies, MR
arthrography has not been widely accepted for evaluating the rotator cuff as it has
been for imaging the glenoid labrum. Most authors cited by Tuite (2002) have found
that Fat suppressed Fast spin echo T2-weighted images obtained with a quality
shoulder coil is adequate for routine imaging of the cuff.


Patient


63-year-old male patient presenting with chronic pain in the lateral aspect of the
shoulder and on abduction. For the assessment of rotator cuff syndrome.


Patient consideration and equipment


MRI was performed on a GE Signa MR/i (GE, Milwaukee) at 1.5T, using a
quadrature shoulder coil.      The patient and I completed the metal screening
questionnaire. The procedure was explained to the patient and was informed that
the image quality will be adversely affected if the patient does heavy breathing and
moves his arm. The patient mentioned that when he lies down for a considerable
length of time his shoulders start to ache. He also mentioned particular shoulder
pain in certain positions of humeral rotation. He was then asked to change into a
hospital gown and taken to the scanner. The patient was asked to lie down supine,
with the shoulder placed in the coil. He was then asked to place his affected arm by
his side in order to minimize respiratory motion to the shoulder. The arm is placed in
the neutral to slight external rotation to avoid overlap of the supraspinatus and


                                           34
MRI Clinical Applications II                                                 Joseph Castillo


infraspinatus tendons on the coronal oblique images (Uri, 1997; Zlatkin, 1996). For
this reason internal rotation is avoided. He mentioned that he might get tired in this
position. I then asked him to put his four fingers under his thigh and the thumb next
to the thigh. Foam pads were placed under the whole arm and he was provided with
a pillow. A large triangular foam pad was placed under the knees. When the patient
indicated that he was comfortable a large Velcro strap was fastened across the chest
to maintain immobilisation of the arm. A panic buzzer was given in the other hand
and earphones were provided to prevent hearing impairment.

The centering laser light was placed over the middle of the coil, landmarked and
patient was moved into the bore of the scanner.


Sequence Parameters


Our site protocol (Table 3.1) for shoulder includes axial proton density (high
resolution), Proton Density Coronal Oblique with fat suppression, T2- weighted FSE
Coronal Oblique, T2W Sagittal Oblique and STIR Sagittal oblique. A Saturation slab
was placed over the Thorax for the Coronal Oblique sequences. Due to the position
of the shoulder at the periphery of the magnetic field Auto shimming always on.




        TR       TE       TI    F     Matrix         NEX   ETL   FOV   THK   SAT     Time
                                A
        msec     msec
Ax      4150     30                   512x384        2     10    14    3/1   I       3:31
PD
Cor     2250     34                   256x224        3     6     14    4/1   L       2:42
PD fs                                                                        FAT
Cor     3000     100                  256x224        4     16    14    4/1   L       3:18
T2
Sag     2250     34                   256x224        3     6     14    4/1   Fat     4:21
PD fs
STIR    3716     50       150         256x192        2     6     14    4/1           4:52
Ax      500      Min            30    512x256        3           14    4/1           4:35
T2*G
RE


Table 3.1 Protocol as used in Malta




                                                35
MRI Clinical Applications II                                                Joseph Castillo


Findings


Degeneration and partial tears are indistinguishable on T1W (short TE) sequences
where they appear as focal or diffuse increased signal intensity. On T2W if these
areas appear the same signal intensity as muscle then it is consistent with signal
degeneration. If these areas exhibit high signal intensity, similar to fluid, then they
represent partial tears.

Full thickness tear by MRI is evidenced by discontinuity of the tendon with high signal
intensity fluid traversing the gap between the tendon fragments from the articular to
the bursal surfaces of the tendon on T2W sequences. On T1w the tear appears as
intermediate signal and thickened rather than the normal low signal intensity.

On the PD Coronal oblique fat suppressed image there is an intermediate signal in
the Rotator cuff tendon proximal to its insertion. This focal area of high signal is
again seen on the T2W sequence and STIR images. Imaging demonstrates "an
extensive tear of the rotator cuff tendon proximal to its insertion to the major tubercle
(Radiologist report). Images also demonstrate an acromial slope, which narrows the
space between the humeral head and acromion, a predisposing condition to tears.
The images also demonstrate acromioclavicular joint degenerative changes. (see
images end of section)


Critical Evaluation


The site protocol used is standard and it has been developed so that it not only
evaluates rotator cuff disease but also evaluate the disease continuum and other
pathologies. A high-resolution axial proton density (long TR/Short TE) is obtained
first in order to evaluate all the muscles surrounding the joint and particularly the
glenouhumeral cartilage and labrum. It is also used to plan accurately the oblique
coronals.    Two sequences are obtained in this plane: 1) Fast Spin Echo Proton
Density (Long TR/Short TE) with Fat Suppression and 2) a Fast Spin Echo (Long
TR/TE).     Various studies have shown that most cuff tears can be seen with this
sequence. Patten et al also suggested that oblique sagittal images provide 10%
improvement in the accuracy of tears. While this may seem insignificant I feel more
confident to identify tear/pathology from another perspective.       Thus two sagittal
oblique sequences are also obtained: 1) Fast Spin Echo Proton Density (long
TR/Short TE) with fat saturation and 2) STIR. This last sequence is sometimes



                                           36
MRI Clinical Applications II                                                Joseph Castillo


prescribed in the coronal plane if assessing denervation of the supraspinatus muscle.
As we do not have experience with T2* gradient echo sequence in the shoulder joint,
T2* gradient echo is just used when more information is required about the state of
the labrum.



Although our site protocol seems too address very well rotator cuff pathology, I still
feel that there is more scope for improvement. MR arthrography is one area, which I
would like to introduce in the protocol. Several authors have reported that direct MR
arthrography is close to 100% sensitive and specific for full thickness tear and
articular surface partial tears of the rotator cuff (Guckel&Nidecker, 1997). Kaplan et
al. (2001) stressed that shoulder MR arthrogram is now routine investigation in their
institution, unless the physician asks for conventional MRI of the shoulder. This has
resulted in a considerable increase in their referrals for MRI of the shoulder,
indicating satisfaction from referring physicians.

The disadvantage of direct MR arthrography is that it requires an injection into the
joint making it a semi-invasive examination. As Fluoroscopy is required the total
examination time is increased. In addition, it may be logistically difficult to organise
and perform if the scanner and the fluoroscopic unit are in different sections of the
department. This is exactly our situation as with only one Fluoroscopy unit and fully
loaded, adding MR Arthrography patients will surely disrupt the throughput in this unit
and possibly the MRI unit. Alternatively, the shoulder examinations will be done in
the afternoon (extended time) when the department is closed and Fluoroscopy unit is
not used. This would be ideal, but with increased costs as staff would have to be
paid on overtime basis. In addition FDA has not approved the use of gadolinium
intra-articular, and so it requires permission from hospital authorities and informed
consent from the patient. Acceptance by the radiology and orthopaedic department
to include MR arthrography requires marketing, which is not within my power to
implement. The best way to speed up the change process is to introduce indirect MR
arthrography. This only requires intravenous injection but would still create pressure
on time. In addition it does not distend the joint (Tuite, 2002) and thus might not be
favourable with radiologists. Still it's worth a try although this involves a lot of work
through quality circles.




                                           37
MRI Clinical Applications II                                                  Joseph Castillo


Treatment and Prognosis


Appropriate management of rotator cuff tears relies on the proper classification of the
injury as chronic, acute extension of a chronic injury, or acute tear. Chronic injuries
are initially managed conservative with anti-inflammatory followed by physical
rehabilitation.     Acute extension of chronic tears require early oerthopaedic
consultation for probable surgical repair (Soohoo & Rosen, 1996)



Alternative imaging modalities


Standard radiographs pf the shoulder joint remain the first study in any patient
suffering from a shoulder trauma, pain or instability. There are many projections to
assess the shoulder joint, and x-rays will show any gross osseous lesions. However
for soft tissue lesions their value is limited and highly dependable on the radiologist
experience (Guckel & Nidecker, 1997)

Ultrasound seems to be an attractive alternative because it is quiet simple, non-
invasive, available and relatively inexpensive technique. However, the sensitivity and
specificity reported in literature varies widely as it depends in the operator (ibid). In
addition the difficulty in documenting the images in a standard manner and therefore
gain reproducible diagnosis is a major drawback. Contrary to MR imaging, US has
not yet demonstrated that it influences the clinicians diagnostic certainty and
treatment protocol (Zanetti & Hodler, 2000).




References


       Brenner M.L., Morrison W.B., Carrino J.A., Nusser C.A., Sanders T.G., Howard R.F.,
        Meier P. (2000) Direct MR Arthrography: is exercise prior to imaging beneficial or
        detrimental, Radiology, 215(2):491-496
       Creus III J.V., Stoller D.W., RYU R.K.N. (1999) IN:Magnetic Resonance Imaging
                                        rd
        (eds:Stark DD Bradley WG Jr), 3 ed., St.Louis, USA, Mosby,
       Lee S.E. & Lang P. (2000) MR and MR arthrography to identify degenerative and
        posttraumatic diseases in the shoulder joint, European Journal of Radiology, 35:126-
        135.
       Guckel C. & Nidecker A. 1997, Diagnosis of tears in rotator cuff injuries, European
        Journal of Radiology 25, 168-176.




                                            38
MRI Clinical Applications II                                                      Joseph Castillo


       Jaovisidha S., Jacobson J.A., Lenchi K.L., Resnick D. (1999) MR imaging of rotator
        cuff tears: is there a diagnostic benefit of shoulder exercise prior to imaging, Clinical
        Imaging, 23(4):249-253 (abstract).
       Masciocchi C., Michel C., Barile A., Maurizi Enrici R., (1997) The Shoulder: MR
        Technique, La Radiologia Medica, 93:143-144.
       Mirowitz S.A. (1993) Imaging Techniques, Normal Variatios and Diagnostic Pitfalls in
        Shoulder Magnetic resonance Imaging, MRI Clinics of North America, 1(1): 19-36.
       Reinus W.R., Shady K.L., Mirowitz S.A., Totty W.G., (1995) MR diagnosis of rotator
        cuff tears of the shoulder: value of using T2-weighted fat saturated images. American
        journal of Roentgenology, 164(6):1451-1455.
       Resnick D. & Kang H.S., (1997) Internal Derangements of Joints, Philadelphia, USA,
        WB Saunders Company
       Sahin-Akyar G., Miller T.T., Staron R.B., McCarthy D.M., Feldman F., (1998) Gradient
        Echo versus Fat suppressed Fast spin echo MR Imaging of rotator cuff tears,
        American Journal of Roentgenology, 171(1): 223-227.
       Singson R.D., Hoang T., Dan S., Friedman M. (1996) MR Evaluation of Rotator Cuff
        Pathology using T2-weighted fast spin echo with and without fat suppression,
        American Journal of Roentgeonology, 166(5): 1061-1065
       Soohoo F.N & Rosen (1996) Diagnosis & Treatment of Rotator Cuff tears in the
        Emergency Department, The Journal of Emergency Medicine, 14(3): 309-317.
       Tsao L.Y. & Mirowitz S.A. (1997) MR Imaging of the Shoulder, Imaging techniques,
        Diagnostic pitfalls and normal variants, MRI Clinics of North America, 5(4):683-704.
       Tuite M, “Shoulder, Rotator cuff injury (MRI)”.online. emedicine.com. July, 2002.
        Available:http://www.emedicine.com/radio/topic894.htm (25/09/2002).
       Uri D.S. (1997) MR Imaging Shoulder Impingement and Rotator Cuff Disease,
        Radiologic Clinics of North America, 35(1): 77-94.
       Zanetti M. & Hodler J. (2000) Imaging of degenerative and post traumatic disease in
        the shoulder joint with US, European Journal of Radiology, 35: 119-125.


       Zlatkin M.B. (1996) Shoulder IN: Clinical Magnetic Resonance Imaging (Eds:
                                                     nd
        Edelman R.R., Hesselink J.R., Zlatkin M.B.) 2 ed, Philadelphia, USA, WB Saunders,




                                               39

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:101
posted:5/8/2011
language:English
pages:12