Hi all,

I am new to DICOM and have a question when it comes to modality within the study. Within a single study, is it possible for multiple modalities to exist? Initially, I was under the assumption that a single, study produced by a scanner would always have the same modality for all it's series. However, after reading an article "Philips DICOM Cookbook for Implementation on Modalities" it appears that multiple series under the same study can have different modalities. If this is a true statement, could you please provide a real world example so I can further solidify my understanding? Thank you very much for your time!

On Friday, July 7, 2023 at 9:48:21 AM UTC-4, Michael Ventresca wrote:
> Hi all,
>
> I am new to DICOM and have a question when it comes to modality within the study. Within a single study, is it possible for multiple modalities to exist? Initially, I was under the assumption that a single, study produced by a scanner would always have the same modality for all it's series. However, after reading an article "Philips DICOM Cookbook for Implementation on Modalities" it appears that multiple series under the same study can have different modalities. If this is a true statement, could you please provide a real world example so I can further solidify my understanding? Thank you very much for your time!

Hi, one example I can think of right now, is a study containing both CT and PET scans.
https://www.radiologyinfo.org/en/info/pet
....from a PET/CT scanner:
https://en.wikipedia.org/wiki/PET-CT
....where you have multiple series for the PT and CT scans, in this same study.

Highly doubt that a single series could have multiple modalities though, doesn't seem right, since the series has special logical associations by modality.

-thanks

On Friday, July 7, 2023 at 11:03:59 AM UTC-4, Chris O'Donnell wrote:
> On Friday, July 7, 2023 at 9:48:21 AM UTC-4, Michael Ventresca wrote:
> > Hi all,
> >
> > I am new to DICOM and have a question when it comes to modality within the study. Within a single study, is it possible for multiple modalities to exist? Initially, I was under the assumption that a single, study produced by a scanner would always have the same modality for all it's series. However, after reading an article "Philips DICOM Cookbook for Implementation on Modalities" it appears that multiple series under the same study can have different modalities. If this is a true statement, could you please provide a real world example so I can further solidify my understanding? Thank you very much for your time!
> Hi, one example I can think of right now, is a study containing both CT and PET scans.
> https://www.radiologyinfo.org/en/info/pet
> ...from a PET/CT scanner:
> https://en.wikipedia.org/wiki/PET-CT
> ...where you have multiple series for the PT and CT scans, in this same study.
>
> Highly doubt that a single series could have multiple modalities though, doesn't seem right, since the series has special logical associations by modality.
>
> -thanks

Hi Chris,

Thank you for your response. I will look into the information in which you linked. In regards to your last statement, I was asking if each series within the same study could have different modalities and it looks like based upon your answer and these articles, the answer is yes. Thanks again!

Yes, Modality is a Series level attribute [1], so by definition, a Study can contain Series with different Modality values, and the Instances within a single Series must all be of the same Modality.

"Modality" covers a multitude of sins [2], and does not just represent the acquisition modality, but can include Structured Reports, Key Object Selection documents, Presentation States, RT Structure Sets, Doses, Plans, etc., each of which has a distinct string value for Modality, ergo has to be in a separate Series, but which can be included in the same Study.

In the query, consider also Modalities in Study [3], which is multi-valued.

David

1. https://dicom.nema.org/medical/dicom/current/output/chtml/part03/chapter_A.html#sect_A.1.2.3
2. https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.3.html#sect_C.7.3.1.1.1
3. https://dicom.nema.org/medical/dicom/current/output/chtml/part04/sect_C.6.2.html#sect_C.6.2.1.2

On Friday, July 7, 2023 at 11:39:54 AM UTC-4, Michael Ventresca wrote:
> Thank you for your response. I will look into the information in which you linked. In regards to your last statement, I was asking if each series within the same study could have different modalities and it looks like based upon your answer and these articles, the answer is yes. Thanks again!

Thank you very much, David! I'd like to re-iterate my understanding with the below use-cases:

As mentioned in your above post and defined in the standards: "A Study may include Composite Instances that are created by a single modality, multiple modalities or by multiple devices of the same modality."

With the above statement, does this imply that the below statements are all true?:
1.) Scenario: A patient, on a single day gets scanned on 2 separate devices that both produce an MRI. In this use-case, these scans can be grouped within the same study.
2.) Scenario: A patient, on a single day gets scanned on 2 separate devices where one device produces a CT and the other device an MRI. In this use-case, there should be a study for the CT and a separate study for the MRI
3.) Scenario: A patient, on a single day gets scanned on a single device which is capable of producing multiple modalities (e.g., PET/CT). In this use-case, since a single device is capable of producing multiple modalities, this would be a single, study

In addition to the above, a single study, could include other series which does not represent the acquisition modality (e.g., MRI) but represents supplemental information such as reports (e.g., physician/reading reports), presentation, etc.. These could technically have a different modality value than the acquisition modality (if exists). For example, "SR".

Thank you very much!

On Saturday, July 8, 2023 at 11:28:51 AM UTC-4, Michael Ventresca wrote:

> With the above statement, does this imply that the below statements are all true?:

> 1.) Scenario: A patient, on a single day gets scanned on 2 separate devices that both produce an MRI. In this use-case, these scans can be grouped within the same study.

Yes, theoretically, but this would require either the operator of both devices selecting the same worklist entry (when the Study Instance UID is obtained from the RIS via MWL), or a "study merge" operation performed on the PACS by an administrator. In environments where there is a 1 request/order : 1 study : 1 report relationship required, this is sometimes necessary (the standard does not require this).

A more practical example would be a CT that was partially completed and then the tube blew and it was finished off on another CT device; probably not so common these days. Theoretically any acquisition modality could have some sort of failure during a procedure and be completed on another machine with a desire not to discard partial results (which may already have been auto-sent anyway). There may be regulatory/policy requirements not to discard partial results that were the result of ionizing radiation delivery, even if the procedure is completely restarted.

Also, there are cases where a procedure has a delayed follow up, e.g., on the next day, and might be requested/ordered once; the same Study Instance UID might be included in multiple successive scheduled procedure steps for such procedures, or the same one retained and re-used - so single studies can span several days, whether they are on the same or different devices. These kinds of delayed studies happen in NM for example.

There are also midnight crossing studies .... so don't get hung up on a "single day".

> 2.) Scenario: A patient, on a single day gets scanned on 2 separate devices where one device produces a CT and the other device an MRI. In this use-case, there should be a study for the CT and a separate study for the MRI

It is theoretically possible that a CT and MR could be combined into one study on the worklist or retrospectively, but it would be very unlikely (unless for some peculiar reason there was a request/order for both together), so yes, it is more likely they would be in separate studies (and separately requested/ordered and reported), as is the normal routine.

> 3.) Scenario: A patient, on a single day gets scanned on a single device which is capable of producing multiple modalities (e.g., PET/CT). In this use-case, since a single device is capable of producing multiple modalities, this would be a single, study

This is routine for hybrid modalities like PET/CT or PET/MR and again there is usually a single request/order and report.

> In addition to the above, a single study, could include other series which does not represent the acquisition modality (e.g., MRI) but represents supplemental information such as reports (e.g., physician/reading reports), presentation, etc.. These could technically have a different modality value than the acquisition modality (if exists). For example, "SR".

Yes, again routine.

David

Thank you very much for the informative response, David. It is really helpful to hear this in context of real world applications. Thanks again!

David,

Could you please help me with a few additional clarifications if you don't mind?:

1.) Since the SOP Class UID defines the IOD + services, the SOP Class UID would also be considered a series-level attribute not study-level, correct?
2.) I cannot imply that there is a direct correlation between the value for the modality attribute and the associated, SOP Class UID. In the standards, I see that there is a 1:1 mapping between SOP Class UID AND IOD. However, it would not be proper logic to infer a mapping from the value for the modality attribute to one or more SOP Class UIDs.

Thank you

On Sunday, July 9, 2023 at 1:05:52 PM UTC-4, Michael Ventresca wrote:

> 1.) Since the SOP Class UID defines the IOD + services, the SOP Class UID would also be considered a series-level attribute not study-level, correct?

No. SOP Class UID is an Instance level attribute, and a Series may contain instances with different SOP Classes. Accordingly, it is
specified in the SOP Common Module, which is at the Instance (Image) level.

> 2.) I cannot imply that there is a direct correlation between the value for the modality attribute and the associated, SOP Class UID.

Correct. Some Modality values may be encoded in a multitude of different SOP Classes. E.g., legacy, legacy converted and enhanced MR,.
not to mention secondary capture, parametric map SOP Classes, etc ., could all be MR Modality (and theoretically in the same Series).

> In the standards, I see that there is a 1:1 mapping between SOP Class UID AND IOD.

No, not always ... see for example the For Processing and For Presentation Digital X-Ray SOP Classes, which both map to the same IOD.

> However, it would not be proper logic to infer a mapping from the value for the modality attribute to one or more SOP Class UIDs.

No, but it is often useful to have a subset of "plausible" SOP Classes per Modality value, e.g., when negotiating a C-GET. See for example
the source code for "https://www.dclunie.com/pixelmed/software/javadoc/com/pixelmed/dicom/SOPClass.html#getPlausibleStandardSOPClassUIDsForModality-java.lang.String-"

Thank you very much for the clarification and link, David! I will have my development team look into that method to see how modality value is mapped to plausible, SOP Class UIDs. Enjoy the rest of your weekend

Michael Ventresca schrieb am Freitag, 7. Juli 2023 um 15:48:21 UTC+2:
> Hi all,
>
> I am new to DICOM and have a question when it comes to modality within the study. Within a single study, is it possible for multiple modalities to exist? Initially, I was under the assumption that a single, study produced by a scanner would always have the same modality for all it's series. However, after reading an article "Philips DICOM Cookbook for Implementation on Modalities" it appears that multiple series under the same study can have different modalities. If this is a true statement, could you please provide a real world example so I can further solidify my understanding? Thank you very much for your time!

Hi Michael,

some more very common examples can be found in eye care. There you'll often see multi-modality devices acquiring data for multiple modalities from a single MWL order.

Imagine an order for cataract pre-op diagnosis. Lots of devices are able to aquire eye biometry (axial lengths), keratometry (corneal curvature) and various quality control images in a single "shot". Each of these data sets will be stored in different series (OAM, KER, OP) but all of them in the same study associated with the same MWL order and accession number.

Or corneal topographic analysis which leads to corneal topographic maps (OPM) as well as source images (OP) all in the same study.

Regards,
Patrick

Hi Patrick,

Thank you very much for the real-use case. It helps greatly to help all of this knowledge sink in!

Dr. Clunie,

I noticed you mentioned the C_GET operation in your response about plausible SOP Classes. Would this mapping also hold true for C_STORE operations?

My use-case:
I own an application that stores clinical trial data for sponsors. In this application, site users (e.g., CRCs) will retrospectively upload copies of their acquisition modalities so these images can be reviewed/seen by the sponsor. In terms of configuration processes in our system, we want to keep it simple so the configuration specialist can configure which modalities (e.g., CT, MRI, XRAY, etc..) are expected at each protocol visit. When the configurator selects the "modality", the modality will be automatically mapped to a possible/plausible set of SOP Class UIDs. In this manner, we can "anticipate" how we can/will receive that modality per DICOM standards.

For example:
IF a visit has been configured to receive a "CT" then it would anticipate ANY of the follow set of SOP Class UIDs below (Note, this mapping was taking off the source code you linked above):

Modality - SOP Class Name
CR - Secondary Capture Image Storage
CR - Multiframe Grayscale Byte Secondary Capture Image Storage
CR - Multiframe Grayscale Word Secondary Capture Image Storage
CR - Computed Radiography Image Storage
CR - Digital Xray Image Storage For Processing
CR - Digital Xray Image Storage For Presentation
CR - Digital Intra oral Xray Image Storage For Processing
CR - Digital Intra oral Xray Image Storage For Presentation
CR - Digital Mammography Xray Image Storage For Processing
CR - Digital Mammography Xray Image Storage For Presentation
CR - Xray 3D Craniofacial Image Storage

NOTE, I understand that the above list doesn't include things such as structured reports of presentation states. That would also be something I'd need to address.