Where is left and right position in CT axial image? And aboutDICOMDIRThis is a discussion on Where is left and right position in CT axial image? And aboutDICOMDIR within the DICOM forums, part of the Protocols category; Hi all, I'm new in Dicom, and I'd like to know if there is a rule which specifies the left and the right that axial image (MEDIA STORAGE SOP CLASS ... | ||||||||
| ||||||||
| Unix Content | Register | FAQ | Calendar | Search | Today's Posts | Mark Forums Read |
|
#1
08-11-2008, 09:45 AM
|
 Where is left and right position in CT axial image? And aboutDICOMDIR Hi all, I'm new in Dicom, and I'd like to know if there is a rule which specifies the left and the right that axial image (MEDIA STORAGE SOP CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in radiology the image must be always like that: ________________ |R L | | | | | | | | | | | |_______________| So i'd like to know if it is correct to assume for every DICOM image that I load and display, Left and Right like the picture below : ________________ |R L | | | | | | | | | | | |_______________| In this case when a dicom image is created means that who create the dicom image must create like below: ________________ |R L | | | | | | | | | | | |_______________| is it correct? Another question is about DicomDir: I'd like to know if a DicomDir must be referred only to one data-set or if is possible to have dicomdir that are related to more than a patient e/o study e/o serie. Thank in advance for your help vasmvr  |
|
#2
08-13-2008, 08:51 AM
|
| Re: Where is left and right position in CT axial image? And about DICOMDIR Bonjour Vasmvr, Your first idea of a fixed R-L view is incorrect  Instead, DICOM define 2 vectors for the horizontal and vertical axis of the image (0020,0037). These vectors give the 3D orientation of the image in relation to the patient. Add to this the slice origin (0020,0032) and you are able to get the 3D position of any pixels in a dataset. As for the DICOMDIR, you have one DICOMDIR per media, not per patient. So if you have multiple patient/studies on a media, they should all be referred by the DICOMDIR. Yves On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vasmvr@yahoo.com wrote: >Hi all, >I'm new in Dicom, and I'd like to know if there is a rule which >specifies the left and the right that axial image (MEDIA STORAGE SOP >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in >radiology the image must be always like that: >________________ >|R L | >| | >| | >| | >| | >| | >|_______________| > >So i'd like to know if it is correct to assume for every DICOM image >that I load and display, Left and Right like the picture below : >________________ >|R L | >| | >| | >| | >| | >| | >|_______________| > >In this case when a dicom image is created means that who create the >dicom image must create like below: >________________ >|R L | >| | >| | >| | >| | >| | >|_______________| >is it correct? > >Another question is about DicomDir: >I'd like to know if a DicomDir must be referred only to one data-set >or if is possible to have dicomdir that are related to more than a >patient e/o study e/o serie. > >Thank in advance for your help > >vasmvr |
|
#3
08-14-2008, 09:05 AM
|
| Re: Where is left and right position in CT axial image? And aboutDICOMDIR On 13 Ago, 14:51, Yves Martel > Bonjour Vasmvr, > > Your first idea of a fixed R-L view is incorrect > > Instead, DICOM define 2 vectors for the horizontal and vertical axis > of the image (0020,0037). These vectors give the 3D orientation of > the image in relation to the patient. Add to this the slice origin > (0020,0032) and you are able to get the 3D position of any pixels in a > dataset. > > As for the DICOMDIR, you have one DICOMDIR per media, not per patient. > So if you have multiple patient/studies on a media, they should all be > referred by the DICOMDIR. > > Yves > > On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vas...@yahoo.com wrote: > >Hi all, > >I'm new in Dicom, and I'd like to know if there is a rule which > >specifies the left and the right that axial image (MEDIA STORAGE SOP > >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in > >radiology the image must be always like that: > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > >So i'd like to know if it is correct to assume for every DICOM image > >that I load and display, Left and Right like the picture below : > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > >In this case when a dicom image is created means that who create the > >dicom image must create like below: > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > >is it correct? > > >Another question is about DicomDir: > >I'd like to know if a DicomDir must be referred only to one data-set > >or if is possible to have dicomdir that are related to more than a > >patient e/o study e/o serie. > > >Thank in advance for your help > > >vasmvr > >| | > >| | > >| | > >| | > >|_______________| > > >In this case when a dicom image is created means that who create the > >dicom image must create like below: > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > >is it correct? > > >Another question is about DicomDir: > >I'd like to know if a DicomDir must be referred only to one data-set > >or if is possible to have dicomdir that are related to more than a > >patient e/o study e/o serie. > > >Thank in advance for your help > > >vasmvr Thank you for your answer Yves, so you mean that who generate a Dicom Image must not generate it like this: ________________ |R L | | | | | | | | | | | |_______________| but can generate it like this ________________ |L R | | | | | | | | | | | |_______________| or _______________ |R | | | | | | | | | | L | |_______________| or |L | | | | | | | | | | R | |_______________| Right? I explain you my problem: i have to import dicom data set of axial CT (MEDIA STORAGE SOP CLASS UID 1.2.840.10008.5.1.4.1.1.2) so I'd like to know if i can only open the image and SET R_____L to the image. With the (0020,0037) and (0020,0032) how can i understand the position R and L of the image, sorry but i'm new in Dicom. Which combination are there? Can you give me some example? About DicomDir you mean that one data-set = one dicomdir (one patient, one study , one serie) ? Thanks in advance for your help |
|
#4
08-15-2008, 11:10 AM
|
| Re: Where is left and right position in CT axial image? And about DICOMDIR Bonjour Vasmvr, The tag (0020,0037) has 6 float values. The first 3 values define the horizontal axis (going from the top left pixel to the top right pixel), and the last 3 define the vertical axis (going from top left to bottom left). These axis are expressed in the patient coord. system, with the X axis going from the patient right to the patient left, the Y axis going from the patient front to the patient back and the Z axis going from the patient feet to the patient head. So, for a standard axial image you may have (0020,0037) = 1/0/0 / 0/1/0. In this case, you can place R on the left portion of the image and L on the right portion of the image. But if you have a pilot, a sagittal image or if the patient is on his side you may have to place different labels on the left and right of the image. If the image is at an angle, you may want to place more than one label on each side... Here's one way to do it: - Look for the biggest component of the vector horizontal vector (compare absolute values of the first 3 values of (0020,0037)) - if it is in X - if it is > 0 Label on the left is "Right", on the right is "Left" - if it is < 0 Label on the left is "Left", on the right is "Right" - if it is in Y - if it is > 0 Label on the left is "Front", on the right is "Back" - if it is < 0 Label on the left is "Back", on the right is "Front" - if it is in Z - if it is > 0 Label on the left is "Feet", on the right is "Head" - if it is < 0 Label on the left is "Head", on the right is "Feet" - Look for the second biggest component of the vector - if it is not 0, then you may want to add a second label using the same algo... You can also use the same technique for the top and bottom labels using the last 3 values of (0020,0037). And for the DICOMDIR, the standard specify that there can only be one DICOMDIR per media. So if for example you have a CD with 20 studies from 10 patients, then the DIMDIR will have information on all 20 studies. It will contain a tree with 10 patients at the root, each patient will have links to its studies, each study links to its series and each series link to its individual images, and finally, in each image description you will have the path of the filename for the image. All of this inside one DICOMDIR file. Hope this help... Yves On Thu, 14 Aug 2008 06:05:42 -0700 (PDT), vasmvr@yahoo.com wrote: >On 13 Ago, 14:51, Yves Martel >> Bonjour Vasmvr, >> >> Your first idea of a fixed R-L view is incorrect >> >> Instead, DICOM define 2 vectors for the horizontal and vertical axis >> of the image (0020,0037). These vectors give the 3D orientation of >> the image in relation to the patient. Add to this the slice origin >> (0020,0032) and you are able to get the 3D position of any pixels in a >> dataset. >> >> As for the DICOMDIR, you have one DICOMDIR per media, not per patient. >> So if you have multiple patient/studies on a media, they should all be >> referred by the DICOMDIR. >> >> Yves >> >> On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vas...@yahoo.com wrote: >> >Hi all, >> >I'm new in Dicom, and I'd like to know if there is a rule which >> >specifies the left and the right that axial image (MEDIA STORAGE SOP >> >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in >> >radiology the image must be always like that: >> >________________ >> >|R L | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >> >So i'd like to know if it is correct to assume for every DICOM image >> >that I load and display, Left and Right like the picture below : >> >________________ >> >|R L | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >> >In this case when a dicom image is created means that who create the >> >dicom image must create like below: >> >________________ >> >|R L | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >is it correct? >> >> >Another question is about DicomDir: >> >I'd like to know if a DicomDir must be referred only to one data-set >> >or if is possible to have dicomdir that are related to more than a >> >patient e/o study e/o serie. >> >> >Thank in advance for your help >> >> >vasmvr > >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >> >In this case when a dicom image is created means that who create the >> >dicom image must create like below: >> >________________ >> >|R L | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >is it correct? >> >> >Another question is about DicomDir: >> >I'd like to know if a DicomDir must be referred only to one data-set >> >or if is possible to have dicomdir that are related to more than a >> >patient e/o study e/o serie. >> >> >Thank in advance for your help >> >> >vasmvr > >Thank you for your answer Yves, > >so you mean that who generate a Dicom Image must not generate it like >this: > >________________ >|R L | >| | >| | >| | >| | >| | >|_______________| > >but can generate it like this > >________________ >|L R | >| | >| | >| | >| | >| | >|_______________| > or > _______________ >|R | >| | >| | >| | >| | >| L | >|_______________| > >or > >|L | >| | >| | >| | >| | >| R | >|_______________| > > > >Right? > >I explain you my problem: i have to import dicom data set of axial CT >(MEDIA STORAGE SOP >CLASS UID 1.2.840.10008.5.1.4.1.1.2) so I'd like to know if i can only >open the image and SET R_____L to the image. > >With the (0020,0037) and (0020,0032) how can i understand the position >R and L of the image, sorry but i'm new in Dicom. Which combination >are there? Can you give me some example? > >About DicomDir > >you mean that one data-set = one dicomdir (one patient, one study , >one serie) ? > >Thanks in advance for your help |
|
#5
08-18-2008, 04:21 AM
|
| Re: Where is left and right position in CT axial image? And aboutDICOMDIR On 15 Ago, 17:10, Yves Martel > Bonjour Vasmvr, > > The tag (0020,0037) has 6 float values. The first 3 values define the > horizontal axis (going from the top left pixel to the top right > pixel), and the last 3 define the vertical axis (going from top left > to bottom left). > > These axis are expressed in the patient coord. system, with the X axis > going from the patient right to the patient left, the Y axis going > from the patient front to the patient back and the Z axis going from > the patient feet to the patient head. > > So, for a standard axial image you may have (0020,0037) = 1/0/0 / > 0/1/0. In this case, you can place R on the left portion of the > image and L on the right portion of the image. But if you have a > pilot, a sagittal image or if the patient is on his side you may have > to place different labels on the left and right of the image. If the > image is at an angle, you may want to place more than one label on > each side... > > Here's one way to do it: > > - Look for the biggest component of the vector horizontal vector > (compare absolute values of the first 3 values of (0020,0037)) > - if it is in X > - if it is > 0 > Label on the left is "Right", on the right is "Left" > - if it is < 0 > Label on the left is "Left", on the right is "Right" > - if it is in Y > - if it is > 0 > Label on the left is "Front", on the right is "Back" > - if it is < 0 > Label on the left is "Back", on the right is "Front" > - if it is in Z > - if it is > 0 > Label on the left is "Feet", on the right is "Head" > - if it is < 0 > Label on the left is "Head", on the right is "Feet" > > - Look for the second biggest component of the vector > - if it is not 0, then you may want to add a second label using the > same algo... > > You can also use the same technique for the top and bottom labels > using the last 3 values of (0020,0037). > > And for the DICOMDIR, the standard specify that there can only be one > DICOMDIR per media. So if for example you have a CD with 20 studies > from 10 patients, then the DIMDIR will have information on all 20 > studies. It will contain a tree with 10 patients at the root, each > patient will have links to its studies, each study links to its series > and each series link to its individual images, and finally, in each > image description you will have the path of the filename for the > image. All of this inside one DICOMDIR file. > > Hope this help... > > Yves > > On Thu, 14 Aug 2008 06:05:42 -0700 (PDT), vas...@yahoo.com wrote: > >On 13 Ago, 14:51, Yves Martel > >> Bonjour Vasmvr, > > >> Your first idea of a fixed R-L view is incorrect > > >> Instead, DICOM define 2 vectors for the horizontal and vertical axis > >> of the image (0020,0037). These vectors give the 3D orientation of > >> the image in relation to the patient. Add to this the slice origin > >> (0020,0032) and you are able to get the 3D position of any pixels in a > >> dataset. > > >> As for the DICOMDIR, you have one DICOMDIR per media, not per patient. > >> So if you have multiple patient/studies on a media, they should all be > >> referred by the DICOMDIR. > > >> Yves > > >> On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vas...@yahoo.com wrote: > >> >Hi all, > >> >I'm new in Dicom, and I'd like to know if there is a rule which > >> >specifies the left and the right that axial image (MEDIA STORAGE SOP > >> >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in > >> >radiology the image must be always like that: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >So i'd like to know if it is correct to assume for every DICOM image > >> >that I load and display, Left and Right like the picture below : > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >In this case when a dicom image is created means that who create the > >> >dicom image must create like below: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > >> >is it correct? > > >> >Another question is about DicomDir: > >> >I'd like to know if a DicomDir must be referred only to one data-set > >> >or if is possible to have dicomdir that are related to more than a > >> >patient e/o study e/o serie. > > >> >Thank in advance for your help > > >> >vasmvr > > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >In this case when a dicom image is created means that who create the > >> >dicom image must create like below: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > >> >is it correct? > > >> >Another question is about DicomDir: > >> >I'd like to know if a DicomDir must be referred only to one data-set > >> >or if is possible to have dicomdir that are related to more than a > >> >patient e/o study e/o serie. > > >> >Thank in advance for your help > > >> >vasmvr > > >Thank you for your answer Yves, > > >so you mean that who generate a Dicom Image must not generate it like > >this: > > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > >but can generate it like this > > >________________ > >|L R | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > or > > _______________ > >|R | > >| | > >| | > >| | > >| | > >| L | > >|_______________| > > >or > > >|L | > >| | > >| | > >| | > >| | > >| R | > >|_______________| > > >Right? > > >I explain you my problem: i have to import dicom data set of axial CT > >(MEDIA STORAGE SOP > >CLASS UID 1.2.840.10008.5.1.4.1.1.2) so I'd like to know if i can only > >open the image and SET R_____L to the image. > > >With the (0020,0037) and (0020,0032) how can i understand the position > >R and L of the image, sorry but i'm new in Dicom. Which combination > >are there? Can you give me some example? > > >About DicomDir > > >you mean that one data-set = one dicomdir (one patient, one study , > >one serie) ? > > >Thanks in advance for your help Hi Yves, thank you very much for your help, i think your answer will help me. So to know the left and right side i only have to use (0020,0037) the first three value, right? And what are the other three values do i need to import dicom data set? What about (0020,0032) i think i need this value only the third value to know the position (which is the first image and so on) can i assume the small value that is the first axial? For the dicomdir i don't understand what you mean "for media" i understand like this DICOMDIR Patient1 Patient2 Patient3 Studies Studies1 Studies2 Studies series serie1 serie2 serie images images1 images2 images Is it correct? Now i have to understand how to recognize the link from patient and correct study, serie and images Thanks in advantage for your help vasmvr |
|
#6
08-20-2008, 03:23 AM
|
| Re: Where is left and right position in CT axial image? And aboutDICOMDIR On 15 Ago, 17:10, Yves Martel > Bonjour Vasmvr, > > The tag (0020,0037) has 6 float values. The first 3 values define the > horizontal axis (going from the top left pixel to the top right > pixel), and the last 3 define the vertical axis (going from top left > to bottom left). > > These axis are expressed in the patient coord. system, with the X axis > going from the patient right to the patient left, the Y axis going > from the patient front to the patient back and the Z axis going from > the patient feet to the patient head. > > So, for a standard axial image you may have (0020,0037) = 1/0/0 / > 0/1/0. In this case, you can place R on the left portion of the > image and L on the right portion of the image. But if you have a > pilot, a sagittal image or if the patient is on his side you may have > to place different labels on the left and right of the image. If the > image is at an angle, you may want to place more than one label on > each side... > > Here's one way to do it: > > - Look for the biggest component of the vector horizontal vector > (compare absolute values of the first 3 values of (0020,0037)) > - if it is in X > - if it is > 0 > Label on the left is "Right", on the right is "Left" > - if it is < 0 > Label on the left is "Left", on the right is "Right" > - if it is in Y > - if it is > 0 > Label on the left is "Front", on the right is "Back" > - if it is < 0 > Label on the left is "Back", on the right is "Front" > - if it is in Z > - if it is > 0 > Label on the left is "Feet", on the right is "Head" > - if it is < 0 > Label on the left is "Head", on the right is "Feet" > > - Look for the second biggest component of the vector > - if it is not 0, then you may want to add a second label using the > same algo... > > You can also use the same technique for the top and bottom labels > using the last 3 values of (0020,0037). > > And for the DICOMDIR, the standard specify that there can only be one > DICOMDIR per media. So if for example you have a CD with 20 studies > from 10 patients, then the DIMDIR will have information on all 20 > studies. It will contain a tree with 10 patients at the root, each > patient will have links to its studies, each study links to its series > and each series link to its individual images, and finally, in each > image description you will have the path of the filename for the > image. All of this inside one DICOMDIR file. > > Hope this help... > > Yves > > On Thu, 14 Aug 2008 06:05:42 -0700 (PDT), vas...@yahoo.com wrote: > >On 13 Ago, 14:51, Yves Martel > >> Bonjour Vasmvr, > > >> Your first idea of a fixed R-L view is incorrect > > >> Instead, DICOM define 2 vectors for the horizontal and vertical axis > >> of the image (0020,0037). These vectors give the 3D orientation of > >> the image in relation to the patient. Add to this the slice origin > >> (0020,0032) and you are able to get the 3D position of any pixels in a > >> dataset. > > >> As for the DICOMDIR, you have one DICOMDIR per media, not per patient. > >> So if you have multiple patient/studies on a media, they should all be > >> referred by the DICOMDIR. > > >> Yves > > >> On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vas...@yahoo.com wrote: > >> >Hi all, > >> >I'm new in Dicom, and I'd like to know if there is a rule which > >> >specifies the left and the right that axial image (MEDIA STORAGE SOP > >> >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in > >> >radiology the image must be always like that: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >So i'd like to know if it is correct to assume for every DICOM image > >> >that I load and display, Left and Right like the picture below : > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >In this case when a dicom image is created means that who create the > >> >dicom image must create like below: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > >> >is it correct? > > >> >Another question is about DicomDir: > >> >I'd like to know if a DicomDir must be referred only to one data-set > >> >or if is possible to have dicomdir that are related to more than a > >> >patient e/o study e/o serie. > > >> >Thank in advance for your help > > >> >vasmvr > > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > > >> >In this case when a dicom image is created means that who create the > >> >dicom image must create like below: > >> >________________ > >> >|R L | > >> >| | > >> >| | > >> >| | > >> >| | > >> >| | > >> >|_______________| > >> >is it correct? > > >> >Another question is about DicomDir: > >> >I'd like to know if a DicomDir must be referred only to one data-set > >> >or if is possible to have dicomdir that are related to more than a > >> >patient e/o study e/o serie. > > >> >Thank in advance for your help > > >> >vasmvr > > >Thank you for your answer Yves, > > >so you mean that who generate a Dicom Image must not generate it like > >this: > > >________________ > >|R L | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > >but can generate it like this > > >________________ > >|L R | > >| | > >| | > >| | > >| | > >| | > >|_______________| > > or > > _______________ > >|R | > >| | > >| | > >| | > >| | > >| L | > >|_______________| > > >or > > >|L | > >| | > >| | > >| | > >| | > >| R | > >|_______________| > > >Right? > > >I explain you my problem: i have to import dicom data set of axial CT > >(MEDIA STORAGE SOP > >CLASS UID 1.2.840.10008.5.1.4.1.1.2) so I'd like to know if i can only > >open the image and SET R_____L to the image. > > >With the (0020,0037) and (0020,0032) how can i understand the position > >R and L of the image, sorry but i'm new in Dicom. Which combination > >are there? Can you give me some example? > > >About DicomDir > > >you mean that one data-set = one dicomdir (one patient, one study , > >one serie) ? > > >Thanks in advance for your help Hi Yves, thanks again for your help, now I understand the tag 0020,0037 but I have another question: the value of the float value must be -1 or 0 or 1 or the can have a floating point value (esample 0.6) and is it possible to have some value different from 0 esample 0.6/-0.5/0.2 0.5/0.8/1 or it works as a switch and a can have only one value different from 0 in each axis? Example 1/0/0/0/1/0? Thanks again for your help. Vasmvr |
|
#7
08-22-2008, 04:11 PM
|
| Re: Where is left and right position in CT axial image? And about DICOMDIR On Wed, 20 Aug 2008 00:23:03 -0700 (PDT), vasmvr@yahoo.com wrote: This is becoming confusing, me answering at the top and you at the bottom  So, this time my answer is at bottom. >On 15 Ago, 17:10, Yves Martel >> Bonjour Vasmvr, >> >> The tag (0020,0037) has 6 float values. The first 3 values define the >> horizontal axis (going from the top left pixel to the top right >> pixel), and the last 3 define the vertical axis (going from top left >> to bottom left). >> >> These axis are expressed in the patient coord. system, with the X axis >> going from the patient right to the patient left, the Y axis going >> from the patient front to the patient back and the Z axis going from >> the patient feet to the patient head. >> >> So, for a standard axial image you may have (0020,0037) = 1/0/0 / >> 0/1/0. In this case, you can place R on the left portion of the >> image and L on the right portion of the image. But if you have a >> pilot, a sagittal image or if the patient is on his side you may have >> to place different labels on the left and right of the image. If the >> image is at an angle, you may want to place more than one label on >> each side... >> >> Here's one way to do it: >> >> - Look for the biggest component of the vector horizontal vector >> (compare absolute values of the first 3 values of (0020,0037)) >> - if it is in X >> - if it is > 0 >> Label on the left is "Right", on the right is "Left" >> - if it is < 0 >> Label on the left is "Left", on the right is "Right" >> - if it is in Y >> - if it is > 0 >> Label on the left is "Front", on the right is "Back" >> - if it is < 0 >> Label on the left is "Back", on the right is "Front" >> - if it is in Z >> - if it is > 0 >> Label on the left is "Feet", on the right is "Head" >> - if it is < 0 >> Label on the left is "Head", on the right is "Feet" >> >> - Look for the second biggest component of the vector >> - if it is not 0, then you may want to add a second label using the >> same algo... >> >> You can also use the same technique for the top and bottom labels >> using the last 3 values of (0020,0037). >> >> And for the DICOMDIR, the standard specify that there can only be one >> DICOMDIR per media. So if for example you have a CD with 20 studies >> from 10 patients, then the DIMDIR will have information on all 20 >> studies. It will contain a tree with 10 patients at the root, each >> patient will have links to its studies, each study links to its series >> and each series link to its individual images, and finally, in each >> image description you will have the path of the filename for the >> image. All of this inside one DICOMDIR file. >> >> Hope this help... >> >> Yves >> >> On Thu, 14 Aug 2008 06:05:42 -0700 (PDT), vas...@yahoo.com wrote: >> >On 13 Ago, 14:51, Yves Martel >> >> Bonjour Vasmvr, >> >> >> Your first idea of a fixed R-L view is incorrect >> >> >> Instead, DICOM define 2 vectors for the horizontal and vertical axis >> >> of the image (0020,0037). These vectors give the 3D orientation of >> >> the image in relation to the patient. Add to this the slice origin >> >> (0020,0032) and you are able to get the 3D position of any pixels in a >> >> dataset. >> >> >> As for the DICOMDIR, you have one DICOMDIR per media, not per patient. >> >> So if you have multiple patient/studies on a media, they should all be >> >> referred by the DICOMDIR. >> >> >> Yves >> >> >> On Mon, 11 Aug 2008 06:45:13 -0700 (PDT), vas...@yahoo.com wrote: >> >> >Hi all, >> >> >I'm new in Dicom, and I'd like to know if there is a rule which >> >> >specifies the left and the right that axial image (MEDIA STORAGE SOP >> >> >CLASS UID 1.2.840.10008.5.1.4.1.1.2) must have . I know that in >> >> >radiology the image must be always like that: >> >> >________________ >> >> >|R L | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >|_______________| >> >> >> >So i'd like to know if it is correct to assume for every DICOM image >> >> >that I load and display, Left and Right like the picture below : >> >> >________________ >> >> >|R L | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >|_______________| >> >> >> >In this case when a dicom image is created means that who create the >> >> >dicom image must create like below: >> >> >________________ >> >> >|R L | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >|_______________| >> >> >is it correct? >> >> >> >Another question is about DicomDir: >> >> >I'd like to know if a DicomDir must be referred only to one data-set >> >> >or if is possible to have dicomdir that are related to more than a >> >> >patient e/o study e/o serie. >> >> >> >Thank in advance for your help >> >> >> >vasmvr >> >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >|_______________| >> >> >> >In this case when a dicom image is created means that who create the >> >> >dicom image must create like below: >> >> >________________ >> >> >|R L | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >| | >> >> >|_______________| >> >> >is it correct? >> >> >> >Another question is about DicomDir: >> >> >I'd like to know if a DicomDir must be referred only to one data-set >> >> >or if is possible to have dicomdir that are related to more than a >> >> >patient e/o study e/o serie. >> >> >> >Thank in advance for your help >> >> >> >vasmvr >> >> >Thank you for your answer Yves, >> >> >so you mean that who generate a Dicom Image must not generate it like >> >this: >> >> >________________ >> >|R L | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> >> >but can generate it like this >> >> >________________ >> >|L R | >> >| | >> >| | >> >| | >> >| | >> >| | >> >|_______________| >> > or >> > _______________ >> >|R | >> >| | >> >| | >> >| | >> >| | >> >| L | >> >|_______________| >> >> >or >> >> >|L | >> >| | >> >| | >> >| | >> >| | >> >| R | >> >|_______________| >> >> >Right? >> >> >I explain you my problem: i have to import dicom data set of axial CT >> >(MEDIA STORAGE SOP >> >CLASS UID 1.2.840.10008.5.1.4.1.1.2) so I'd like to know if i can only >> >open the image and SET R_____L to the image. >> >> >With the (0020,0037) and (0020,0032) how can i understand the position >> >R and L of the image, sorry but i'm new in Dicom. Which combination >> >are there? Can you give me some example? >> >> >About DicomDir >> >> >you mean that one data-set = one dicomdir (one patient, one study , >> >one serie) ? >> >> >Thanks in advance for your help > >Hi Yves, > >thanks again for your help, now I understand the tag 0020,0037 >but I have another question: the value of the float value must be -1 >or 0 or 1 or the can have a floating point value (esample 0.6) >and is it possible to have some value different from 0 esample >0.6/-0.5/0.2 0.5/0.8/1 or it works as a switch and a can have only one >value different from 0 in each axis? Example 1/0/0/0/1/0? > >Thanks again for your help. > >Vasmvr The tag is composed of 2 3D vectors. They can have any values! For example if you have an axial image with a 30 deg. gantry tilt, you will have: 1/0/0 / 0/0.866/0.5 Any float values are permitted. Though since these should be normed vectors, the sqrt( x*x + y*y + z*z ) should be == 1. But even that I don't believe it is required by the specs (somebody correct me if I'm wrong...) Yves |
|
#8
08-22-2008, 06:01 PM
|
| Direction Cosines are required to be normed (was Re: Where is leftand right position in CT axial image? And about DICOMDIR) On Aug 22, 10:11*pm, Yves Martel > Any float values are permitted. *Though since these should be normed > vectors, the sqrt( x*x + y*y + z*z ) should be == 1. *But even thatI > don't believe it is required by the specs (somebody correct me if I'm > wrong...) I reported that earlier. AFAIK Image Orientation (Patient) (0020,0037) is simply VR=DS with VM=6 that's all what the standard says. But clearly if one reads C.7.6.2.1.1 Image Position And Image Orientation, the values have to be cos/sin values (no normalization appear in the matrix). The fact that x*x + y*y + z*z == 1 is not enough, you also need that dot product is 0 (x[0]*y[0] + x[1]*y[1] + x[2]*y[2] == 0) From the top of my head I think that : * Rescale Slope=0 is also a valid value. * Pixel Spacing (0,0) is also valid ... Again those are good example for a DICOM Conformance Tests -Mathieu |
|
#9
09-01-2008, 12:57 PM
|
| Re: Direction Cosines are required to be normed (was Re: Where isleft and right position in CT axial image? And about DICOMDIR) On Aug 23, 12:01*am, Mathieu Malaterre wrote: > [...] AFAIK Image Orientation (Patient) (0020,0037) > is simply VR=DS with VM=6 that's all what the standard says. > But clearly if one reads C.7.6.2.1.1 Image Position And Image > Orientation, the values have to be cos/sin values (no normalization > appear in the matrix). The fact that x*x + y*y + z*z == 1 is not > enough, you also need that dot product is 0 (x[0]*y[0] + x[1]*y[1] + > x[2]*y[2] == 0) [...] That's absolutely correct. I just want to remember a problem I found many years ago, regarding the Image Position (Patient). Image Position (Patient) is made by three DS components that must be normalized (x*x + y*y + z*z == 1), very similar to the two vectors that compose the Image Orientation (Patient). Unfortunately the single components of these two DICOM attributes are DS, so the number of figures (that is, the precision) you find into them depends only on the application that produces the images. In the images our MR scanner used to produce, I just put three or four digits, because I calculated that it was impossible to get better spatial resolution with our MR scanner, and being a physicist I was trained to never put non significant digits. I expected the test about x*x + y*y + z*z == 1 was made using a similar reasonable number of digits. Unfortunately one (big vendor's) PACS rejected our images, because it used to perform the test using more than 8 figures ! So, the lesson I learnt was: is you have to produce the images, put as many (reasonable) figures as you can, even if they are not significant; if you have to check the images about the contents of Image Position (Patient) and Image Orientation (Patient), use a reasonable precision (three or four figures are more then enough, that is, x*x + y*y + z*z <= 1.001 or >= 0.999 should be accepted). Regards. Luigi Pampana-Biancheri mailto:luigi.pampana@removethis.esaote.com |
|
#10
09-01-2008, 03:07 PM
|
| Re: Direction Cosines are required to be normed (was Re: Where isleft and right position in CT axial image? And about DICOMDIR) On Aug 23, 12:01 am, Mathieu Malaterre wrote: > [...] AFAIK Image Orientation (Patient) (0020,0037) > is simply VR=DS with VM=6 that's all what the standard says. > But clearly if one reads C.7.6.2.1.1 Image Position And Image > Orientation, the values have to be cos/sin values (no normalization > appear in the matrix). The fact that x*x + y*y + z*z == 1 is not > enough, you also need that dot product is 0 (x[0]*y[0] + x[1]*y[1] + > x[2]*y[2] == 0) [...] That's absolutely correct. I just want to remember a problem I found many years ago, also regarding the Image Orientation (Patient). Unfortunately the single components of this DICOM attribute are DS, so the number of figures (that is, the precision) you find into them depends only on the application that produces the images. In the images our MR scanner used to produce, I just put three or four digits, because I calculated that it was impossible to get better resolution with our MR scanner, and being a physicist I was trained to never put non significant digits. I expected the test about x*x + y*y + z*z == 1 was made using a similar reasonable number of digits. Unfortunately one (big vendor's) PACS rejected our images, because it used to perform the test using more than 8 digits ! So, the lesson I learnt was: is you have to produce the images, put as many (reasonable) digits as you can, even if they are not really significant; if you have to check the images about the contents of Image Orientation (Patient), use a reasonable precision (three or four digits are more then enough, that is, x*x + y*y + z*z <= 1.001 or >= 0.999 should be accepted). Regards. Luigi Pampana-Biancheri mailto:luigi.pampana@removethis.esaote.com |
|
#11
09-23-2008, 03:39 AM
|
| Re: Direction Cosines are required to be normed (was Re: Where isleft and right position in CT axial image? And about DICOMDIR) On 1 Set, 21:07, luigipb > On Aug 23, 12:01 am, Mathieu Malaterre > wrote: > > > [...] AFAIK Image Orientation (Patient) (0020,0037) > > is simply VR=DS with VM=6 that's all what the standard says. > > But clearly if one reads C.7.6.2.1.1 Image Position And Image > > Orientation, the values have to be cos/sin values (no normalization > > appear in the matrix). The fact that x*x + y*y + z*z == 1 is not > > enough, you also need that dot product is 0 (x[0]*y[0] + x[1]*y[1] + > > x[2]*y[2] == 0) [...] > > That's absolutely correct. I just want to remember a problem I found > many years ago, also regarding the Image Orientation (Patient). > > Unfortunately the single components of this DICOM attribute are > DS, so the number of figures (that is, the precision) you find into > them depends only on the application that produces the images. > > In the images our MR scanner used to produce, I just put three or four > digits, because I calculated that it was impossible to get better > resolution with our MR scanner, and being a physicist I was > trained to never put non significant digits. I expected the test about > x*x + y*y + z*z == 1 was made using a similar reasonable number of > digits. Unfortunately one (big vendor's) PACS rejected our images, > because it used to perform the test using more than 8 digits ! > > So, the lesson I learnt was: is you have to produce the images, put as > many (reasonable) digits as you can, even if they are not > really significant; if you have to check the images about the contents > of > Image Orientation (Patient), use a reasonable precision (three or four > digits are more then enough, that is, x*x + y*y + z*z <= 1.001 or >= > 0.999 should be accepted). > > Regards. > > Luigi Pampana-Biancheri > mailto:luigi.pamp...@removethis.esaote.com So if I understand good, every Dicom image produced must have this two condition: x*x + y*y + z*z == 1 for every 3D vector and (x[0]*y[0] + x[1]*y[1] + x[2]*y[2] == 0). Right? So I can consider, during import of a Dicom data set, that this two condition are rispected. Right? Now I'd like to understand better this part: The tag is composed of 2 3D vectors. They can have any values! For example if you have an axial image with a 30 deg. gantry tilt, you will have: 1/0/0 / 0/0.866/0.5 : it means that i will have and image like this: _____________ | AF | | | | | | R | | | | | |____________| Is it right? So I have to check if the values in the (0020,0037) are different from 0 and in case associate the correct label. Am I correct? What do you think to not check, before importing the images, this condition x*x + y*y + z*z == 1 for every 3D vector and (x[0]*y[0] + x[1]*y[1] + x[2]*y[2] == 0). I think that if the image is Dicom must be correct. I hope some one can answer me. Thanks in advance. Vasmvr |
|
#12
09-23-2008, 06:13 AM
|
| Re: Direction Cosines are required to be normed (was Re: Where isleft and right position in CT axial image? And about DICOMDIR) On Sep 23, 9:39*am, vas...@yahoo.com wrote: > What do you think to not check, before importing the images, this > condition x*x + y*y + z*z == 1 for every 3D vector and *(x[0]*y[0] + > x[1]*y[1] + x[2]*y[2] == 0). I think that if the image is Dicom must > be correct. I was not suggesting not to check these conditions, I was just suggesting to do the check using a reasonable number of digits, that's all. In your example, the lenght of the second vector is 0.999956, it you check it against the interval 0,999999-1.000001 you'll find it is not a unit vector, while in fact probably it is, at least according to the precision of the application that produced it. Regards. Luigi Pampana-Biancheri mailto:luigi.pampana@removethis.esaote.com |
|
#13
09-24-2008, 08:45 AM
|
| Re: Direction Cosines are required to be normed (was Re: Where isleft and right position in CT axial image? And about DICOMDIR) On Sep 23, 5:13*am, luigipb > On Sep 23, 9:39*am, vas...@yahoo.com wrote: > > > What do you think to not check, before importing the images, this > > condition x*x + y*y + z*z == 1 for every 3D vector and *(x[0]*y[0] + > > x[1]*y[1] + x[2]*y[2] == 0). I think that if the image is Dicom must > > be correct. > > I was not suggesting not to check these conditions, I was just > suggesting to do the check using a reasonable number of digits, that's > all. In your example, the lenght of the second vector is 0.999956, it > you check it against the interval 0,999999-1.000001 you'll find it is > not a unit vector, while in fact probably it is, at least according to > the precision of the application that produced it. > > Regards. > > Luigi Pampana-Biancheri > mailto:luigi.pamp...@removethis.esaote.com Just something additional to keep in mind - I have seen images in the past with direction of cosines = 5\1\0\0\1\0. The first vector is obviously not even close to being a unit vector. So, whether there is a rule or not, there are implementations that create bad data, and unfortunately, this happens more often that you'd think. Anyways, something else to consider in your implementation... |
