This summary should not be
construed as a definitive statement of RANZCR or Royal Adelaide Hospital
When preparing image for the Clinical
Image Review you should consult the latest RANZCR documents.
Under the draft RANZCR Accreditation
Standards For Diagnostic And Interventional Radiology (Version
4 Draft 1 available from the web ) , each MRI unit must participate
in the RANZCRs MRI Clinical Image Review Program (CIRP). In addition,
each MRI unit must submit images of an approved phantom for review by RANZCR
approved consultants, at intervals determined by the RANZCR. (clause T1.2.2)
The Clinical Image Review Program is conducted
on a small set of submitted images that are reviewed by an unspecified
number of unnamed radiologists. Reasonable steps are taken to ensure
that the reviewing radiologists have no obvious clue which site they are
reviewing. Reviewers are to assess images against an
unspecified set of objective criteria as well giving an overall impression
of the images. The site under review will receive "detailed
feedback on the objective criteria fro each examination".
Clearly the concepts of the CIRP and much of
the original RANZCR MRI quality accreditation guidelines are drawn from
similar American College of Radiologists documents.
One sheet of film per required sequence, containing
15 20 images.
Images for each anatomical region must come from
the same patient examination & should be acquired with normal imaging
parameters. The use of "extended acquisition periods will not be
acceptable". The documents don't' explain if this means the images
should not contain pathology, or if the panel will judge if the sequence
has been extended from normal practice in an attempt to misrepresent the
site's standard clinical technique.
Technical parameters must be shown on the images,
but patient identification and practice identification must be removed.
It isn't clear if full sequence details should be supplied as well, or
if the reviewer will try to determine all technical details from the available
The required sequences are loosely defined (Axial
Long TR Long TE).
As usual, spatial resolution is poorly handled.
(FOV standard size to include all specified anatomy, minimum matrix specified,
and in one instance a maximum slice thickness) The requirements are
not in any sense technically demanding, and do not represent more than
a minimum acceptable standard. I would expect most sites will
submit images far exceeding the implied minimum requirements.
The minimum matrix required is specified a little
unusually with the phase steps first. I have used the more conventional
method in this page. There is no mention of rectangular matrices so I assume
they will accept a rectangular voxel.
Required Clinical Examinations & Sequences
Knee: Routine examination for internal derangement
or meniscal injury
2 sequences. One coronal and one sagittal.
One sequence with bright fluid. (Presumably the other has dark fluid)
Minimum Matrix 256 x 192.
Brain A: Routine Examination for suspected Multiple
Film 1: Axial . Long TR Long TE
Minimum matrix 256 x 192
Film 2: Sagittal or Coronal FLAIR. Minimum
matrix 256 x 192
Film 2: Sagittal or Coronal Long TR Short
TE Minimum matrix 256 x 192
Brain B: Evaluation of Circle of Willis for Stroke
MRA of Circle of Willis (The posterior fossa and
neck is not required.) Minimum matrix 256 x 128
Cervical Spine: Routine Examination for suspected
Thin section axial images from C4 to T1. Dark
or Bright fluid. Slice thickness 4 mm or less. Minimum matrix 256 x 128.
Lumbar Spine: Routine examination for sciatica
Film 1: Sagittal Long TR Long TE Minimum Matrix
256 x 192
Film 2: Axial Long TR Lng TE at least
from L3 to S1 Minimum matrix 256 x 192
The most obvious point is that the required images
are technically very undemanding. Its hard to imagine a scanner
manufactured in the past 10 years that couldn't produce the required images
with sequences under 5 minutes.
The other obvious point is that the objective
criteria for assessment should be public. I would hope they are a
lot better defined and more appropriate to modern MRI than the loose
requirements recorded above.
The document should cite minimum spatial resolution
in terms of slice thickness and in-plane resolution. The use of inter
slice interpolation (common in 3D MRA sequences ) should be mentioned as
acceptable or not. the current references to spatial resolution are
way too open to interpretation., for example what FOV is sufficient to
include the required anatomy of the axial lumbar spine? 120 mm to
cover paravertebral structures, or 360 mm to include the flanks and avoid
aliasing ? With the specified minimum matrix there is
a 300% variation in inplane resolution which could yield a 900% variation
in SNR (based on voxel volume) implying a theoretical 8400% variation in
ere is no consideration of the T2 induced blurring,
presumably the assessors will eyeball inplane resolution and not rely on
the indicated figures.
Filtering is not specifically banned, so some
sites will submit images with significant digital modification, reflecting
their normal clinical practice.
Region specific comments
Knee: Many sites use a group of sequences
designed to detect mensical injury , bone bruise, cartilaginous damage.
The only bright fluid sequence apart from the fat suppressed T2 would
be in the axial plane to assess patellar cartilage. Would the panel accept
STIR as a bright fluid sequence ? If so what about a photonegative
fat suppressed SE T1 sequence?
Brain A Long TE Long TE. Does the panel
have guidelines here. The appearance of a 2500/90 T2 TSE is quite
different to a 4500/120.
The weighting available within the requirement
of a long TR and short TE will have very variable appearances from real
PD (2500/17) with CSF isointense or darker than white matter, through to
moderate T2 weight (4500/30) where the CSF will be bright. The echo
spacing and ETL will also affect quality here.
Brain MRA: The specifications here are
so loose as to be ridiculous. IMHO anyone producing MRA at 250 FOV
256 x 128 matrix should be ashamed. The slice thickness and TE are
also significant issues here, and there is no indication of what projections
should be provided, or if base images are to be supplied. MRA images
on a 20 format is quite inappropriate.
Cervical Spine: "dark or bright CSF" What
a cop out. A dark CSF image for radiculopathy could be a T1 or a T2 TSE
with poor CSF signal control. How about asking for one of each ?
The minimum matrix of 256 x 128 is really not appropriate any more
unless the FOV is unusually small.
Lumbar Spine: OK, the CIRP asks for T2
axial and sagittal even though the fast T1 images are the most taxing given
contrast, resolution and SAR issues.
What's Next ?
In the near future I hope to publish our submitted
images and sequences, along with a set of images on the same subjects that
would meet the minimum requirements of the CIR documents.
I'll try to get more information on the objective
Please contact me with your experience, details
of feedback etc., and we should be able to compile a site that will assist
those undergoing review and highlight points that the RANZCR
can consider to enhance the role of this review method.