> Date:         Mon, 1 Jul 2002 16:59:48 +0200
> From:         Pieter Vandemaele <[log in to unmask]>

> Not really a topic on SPM, I know, but it's quit urgent. Does anyone
> know if you can convert Analyze images back to dicom images which can be
> sent back to the scanner (Siemens Symphony running Syngo). I tried
> medcon, but it seems that the primary key of the database cannot be
> fulfilled, so every slice is a different "patient".

Dear Pieter,

As one example, one software with which you can do this is
MEDx (http://medx.sensor.com), mentioned at this SPM page
http://www.fil.ion.ucl.ac.uk/spm/#MEDx

> Is there also a way
> to merge an anatomical and functional dataset to one set (in dicom)?

Yes, here are some:

1. A new, second kind of MR image (called Enhanced MR Image)
has been recently defined, with which one can represent fused
anatomical-functional data.
ftp://medical.nema.org/medical/dicom/final/sup49_ft.pdf
(Especially see section C.8.12.3.1.2.1 of this document.)
DICOM Enhanced MR images are 'multi-frame' and exhibit many
useful characteristics not found in the previous dicom MR image.
Unfortunately, as Enhanced MR Image is a new spec, I suspect
support for this image in various commercial (or public)
software will likely not appear widely this year. If your need is
this year, see 2 or 3 below (which are methods you may already be
familiar with). If you would have an interest in information on current
vendor activities on Enhanced MR Image, you might wish to contact
Stephen Vastagh (mailto:[log in to unmask]) at NEMA.

2. By using the old (non-enhanced) dicom MR object, assign
the slices of the anatomical volume to one dicom series
(by assigning the same Series Instance UID to their headers)
and the slices of the functional volume to a second dicom series
(represented by a second Series Instance UID), both in the same
study (i.e. assigh the same Study Instance UID to each slice
of -both- volumes). This method would enable you to subsequently
store the data in dicom archives/scanners, but you would not
be creating dicom images to represent the result of the merge.
Instead, your software would fuse the two volumes upon getting
them back from your dicom archive.

3. You could shrink your anatomical (which is probably 16 bits)
to for example 12 bits (potentially in a lossless manner,
depending on the dynamic range of the 16-bit data), and
store a color map in the last 4 bits to represent the
functional information. With this method, you can both
archive/restore the data from dicom archives (as in 2)
and have fused dicom images (unlike in 2). Unfortunately,
this method provides no color, and the functional information
would instead appear bright (different levels of white).

It is not hard to see that only method 1 offers an interoperable
fusion (merge) method; in 2 and 3, multiple dicom software
would not understand each other's fusion method (unless
you modify the software) and would not be able to display
the fusion correctly.

Yaman


--
Yaman Aksu
Sensor Systems Medical Products
103A Carpenter Drive
Sterling, VA 20164
Phone: (703) 437-7344 x407
Fax: (703) 437-0039
mailto:[log in to unmask]
http://medx.sensor.com