Precise UML modeling framework of the DICOM information model

[0001] The present invention relates to the Digital Imaging and Communications in Medicine (DICOM) standard, and more particularly relates to a standard representation of DICOM in the unified modeling language (UML), i.e., a UML profile for DICOM.

[0002] The Digital Imaging and Communications in Medicine (DICOM) standard is a detailed specification that describes a means for formatting and exchanging images, and associated information. relies on standard communication protocols and addresses the communication and viewing of images from such modalities such as CT, MR, Ultrasound, Nuclear medicine, Digital Cardiology, Angiography, RF equipment and Radiation Therapy devices and systems. Extensions are being developed to include modalities with visible light sources such as Ophthalmology, Microscopy for Pathology applications, as well as Endoscopy. It also allows the exchange of patient demographic data, exam status and scheduling information. The rapid adoption of DICOM by the medical imaging industry is opening new opportunities for healthcare organizations to increase the quality and cost effectiveness of patient care.

[0003] DICOM has an information model defining a set of Information Object Definitions (IODs) which provide an abstract definition of real-world objects applicable to communication of digital medical information. For each IOD, DICOM specifies any necessary information for the semantic description of the IOD, relationships to associated real-world objects relevant to the IOD and attributes which describe the characteristics of the IOD. Each IOD, moreover, is defined as a set of tables based on the entity-relationship (E-R) scheme. The problem is that IODs are not machine-readable and not easy to capture and follow for non-DICOM literates. This acts as a barrier to communication between information architects and software developers from different domains.

[0004] Hence, it is an object of this invention to provide a standard model or template for the Digital Imaging Communications in Medicine (DICOM) using UML.

[0005] It is another object of the invention to provide a standard DICOM model or template which may be used to guide UML modeling for DICOM to improve communication among the information architects and developers from medical imaging domains and software industry domains.

[0006] It is still another object of the invention to generate a UML document from DICOM Information Object Definitions (IODs).

[0007] It is still another object of the invention to generate a UML document based on a UML model or template derived from DICOM IODsc It is still another object of the invention to provide means and method to generate Document Type Definitions (DTDs) or XML schemas from UML documents.

[0008] It is yet another object of this invention to provide a UML profile for DICOM which is machine readable and therefore readily communicable within various medical domains.

[0009] It is yet another object of this invention to provide a means and method to build UML models for the future DICOM IODs, Modules, Macros, or Attributes.

[0010] These objects of the invention (and others) are achieved by providing a modeling technique to convert DICOM information from the DICOM relational model into an object-oriented representation. A methodology is presented for converting the DICOM specification into a UML-based object-oriented representation of the specification, and for converting DICOM reports into UML-compatible representations. This methodology for representing the DICOM specification provides a clear and comprehensive view of's semantically rich framework, its structures and constructs. By providing a mapping between DICOM and UML, developers, analysts and system architects will be better able to understand the DICOM specification, and better able to define and develop DICOM-aware applications.

[0011] The inventions disclosed herein set forth a unique UML profile for DICOM based on the standard stereotypes of UML. Using the inventive a UML profile, the DICOM constraints can be precisely represented in object models, realizing a precise representation of the DICOM information model, and UML documents translated from DICOM-defined documents may be generated.

[0012] The benefit of a common and precise representation of the DICOM information model is the ability to communicate between DICOM and non-DICOM literates for various reasons, including that such a model is machine-readable and can help to automate the development of DICOM applications. The DICOM information model or template is utilized to guide UML modeling for DICOM information object definitions, and to translate between DICOM documents and UML documents.

[0049] The Unified Modeling Language (UML), standardized by the Object Management Group (OMG), provides an industry standard modeling language for modeling object-oriented programs. Representing the DICOM information model in UML, as set forth and claimed herein, will make the DICOM information model in UML more clear and improve DICOM communication among information architects and software developers. Using UML tools, source code may be generated for common programming languages such as Java and C++. The prior art includes commonly owned and copending application Ser. No. 09/686,401, which discloses an invention which uses the Extensible Markup Language (XML) from OMG such that DICOM XML Document Type Definitions (DTDs) may be generated from UML models, as well as UML modeling of XML representation of DICOM, incorporated herein by reference. XML is a meta language which allows a user to define a proprietary markup to describe the structure of data, for example, a database for exchange of medical information between two entities linked by the world wide web.

[0050] Information Model

[0051] The DICOM information model relies on explicit and detailed models of how the “things” (patients, images, reports, etc) involved in medical imaging operations are described, and how they are related. These models are called E-R models and are a way to be sure that manufacturers and users understand the basis for developing the data structures used in DICOM (FIG. 1). That is, FIG. 1 shows rectangular boxes 10, which may represent entities which singly or in combination, form the information objects. The diamond-shaped boxes 20 represent relationships, the arrows represent the connections between entities and relationships. Each IOD consists of a set of information entities (IEs), containing one or more modules, as seen in Table 1.

[0052]

TABLE 1
Comprehensive DICOM IOD definition
IE	Module	Usage
Patient	Patient	M
	Specimen	C - Required if the Observation Sub-
		ject is a Specimen
Study	General Study	M
	Patient Study	U
Series	SR Document Series	M
Equipmen	General Equipment	M
Document	SR Document General	M
	SR Document Content	M
	SOP Common	M

[0053] Usage indicates if such a Module is M—mandatory, C—conditional, or U—user optional. Each Module has a set of Attributes which may be atomic Attributes, Macros, or Sequence Attributes, as seen in Table 2.

[0054]

TABLE 2
General Study Module
Attribute Name	Tag	Type	Attribute Description
Study Instance	(0020,000D)	1	Unique identifier for the Study.
UID
Study Date	(0008,0020)	2	Date the Study started.
.
.
.
Study	(0008,1030)	3	Institution-generated description or
Description			classification of the Study
			(component) performed.
.
.
.
Referenced	(0008,1110)	3	A sequence which provides ref-
Study			erence to a Study SOP Class/
Sequence			Instance pair. The sequence may
			have zero or more Items.
>Referenced	(0008,1150)	1C	UID Uniquely identifies the ref-
SOP Class			erenced SOP Class. Required if
Sequence			Referenced Study (0008,1110) is
			sent.
>Referenced	(0008,1155)	1C	Uniquely identifies the referenced
SOP Instance			SOP Instance. Required if
UID			Referenced Study Sequence
			(0008,1110) is sent.
Procedure	(0008,1032)	3	A Sequence that conveys the type
Code Sequence			of procedure performed. One or
			more Items may be included in
			in this Sequence.

[0055] Table 2 should make clear that an atomic Attribute such as Study Instance UID does not contain any other Attributes. A Sequence Attribute, e.g., Referenced Study Sequence contains two atomic Attributes. Sometimes, a Sequence Attribute may contain other Sequence Attributes or Macros, e.g., Procedure Code Sequence contains Code Sequence Macro, not shown in this document. A Macro has the same structure as a Module. The Tag gives the DICOM Tag for a specific DICOM Attribute. The Type specifies if an Attribute is mandatory or optional with constraints. An Attribute is required and the length of its value field shall not be zero if it is Type 1. Type 1C has the same requirements as Type 1 under some conditions. An Attribute is required and the length of its value field may or may not be zero if it is Type 2. Type 2C has the same requirements as Type 2 under some conditions. Type 3 is optional and can be zero length or no value.

[0056] Within or among Modules or Macros, there may exist some constraints, e.g., Document Content Macro (see Table 3).

[0057]

TABLE 3
Part of Document Content Macro
Attribute Name	Tag	Type	Attribute Description
Value Type	(0040,A040)	1	The type of the value encoded in
			this Content Item.
			Defined Terms:
			TEXT, NUM, CODE, DATE-
			TIME, DATE, TIME, UIDREF,
			PNAME, COMPOSITE, IMAGE,
			WAVEFORM, SCOORD,
			TCOORD CONTAINER
Continuity of	(0040,A050)	1C	This flag specifies for a CON-
Content			TAINER whether or not its con-
			tained Content Items are logically
			linked in a continuous textual flow,
			or are separate items.
			Required if Value Type
			(0040,A040) is CONTAINER.
			Enumerated Values:
			SEPARATE
			CONTINUOUS
			See C.17.3.2 for further
			explanation.
DateTime	(0040,A120)	1C	This is the value of the Content
			Item. Required if Value Type
			(0040,A040) is DATETIME.
Date	(0040,A121)	1C	This is the value of the Content
			Item. Required if Value Type
			(0040,A040) is DATE.

[0058] As shown above, Value Type, a Type 1 Attribute, has defined terms such as DATE, DATATIME, and NUM. And Date Attributes, a Type 1C, is required only if Value Type is DATE. specifies the data structures and Value Representations (VRs). Table 4 lists some of the VRs.

[0059]

TABLE 4
Definition of VRs
VR
Name	Definition Character	Repertoire	Length
AE	A string of characters with leading	Default Character	16 bytes
Appli-	and trailing spaces (20H) being	Repertoire ex-	maxi-
cation	non-significant. The value made of	cluding control	mum
Entity	16 spaces, meaning “no application	characters LF,
	name specified”, shall not be used.	FF, CR and ESC.
AS	A string of characters with one of	“0”–“9”, “D”,	4 bytes
Age	the following formats -- nnnD,	“W”, “M”, “Y”	fixed
String	nnnW, nnnM, nnnY; where nnn	of Default Char-
	shall contain the number of days	acter Repertoire
	for D, weeks for W, months for M,
	or years for Y. Example: “018M”
	would represent an age of 18
	months.
CS	A string of characters with leading	Uppercase char-	16 bytes
Code	or trailing spaces (20H) being	acters, “0”–“9”,	maxi-
String	non-significant.	the SPACE char-	mum
		acter, and under-
		score “_”, of the
		Default Character
		Repertoire

[0060] Each VR specifies its value pattern or restricts the length of the value field, e.g., AS—Age String, is a value of three digits followed by “D”, “W”, “M”, or “Y”. There are 26 VRs defined in the version of DICOM2001 specification. As shown above, the DICOM IODs are not represented in a standard modeling language and are not machine-readable, which forms a barrier to the communication between DICOM and non-DICOM information architects and developers. Therefore, a standard modeling representation of The DICOM information model is needed for common communication. UML, an industry modeling standard, is such a language. In order to represent such finer information over UML class diagrams clearly, a UML profile for DICOM is needed. The UML modeling of representing DICOM constraints, DICOM Attribute information including DICOM Tags, Types, and Enumerated values is called precise UML modeling.

UML Profile for Dicom Information Model

[0061] A UML profile is a stereotyped package that contains model elements that have been customized for a specific domain or purpose by extending the meta-model using UML extension mechanisms including stereotypes, tagged definitions, and constraints. The inventive UML profile for DICOM provides a set of model elements customized specifically for DICOM using extension. It serves as a set of class model templates to help build UML models for DICOM IODs. As discussed above, DICOM specifies a set of IOs such as IODs, IEs, Modules, Macros, and Attributes. The UML profile for DICOM provides an ability to build UML models for DICOM IODs where an associated UML stereotype with each DICOM IO and DICOM data type. Such a UML stereotype may have tagged values if it is defined for DICOM Attributes, where the tagged values specify the DICOM Attribute name, DICOM Tag, and DICOM Type. The inventive UML profile for DICOM may also have constraints, e.g., DICOM IOD which is always a top-level class. Each stereotype is named based on DICOM terms, not UML terms. These new and unique stereotypes, tagged values, and constraints together constitute the inventive UML profile for DICOM. The inventive UML profile for DICOM provides for guidance for the generation of UML models for The DICOM information model.

[0062] FIG. 2A illustrates an example flow diagram for converting the DICOM specification 110 into UML 130 DICOM DTD170 representations that facilitate program development in accordance with this invention. As noted above, although applications can be developed that utilize DICOM's relational structured reporting scheme directly, it can be expected that the number of programmers and other computer professions who are familiar with UML and object-oriented technologies and techniques will be substantially greater than those who are familiar with DICOM and relational technologies and techniques.

[0063] This invention is also based on the premise that DICOM-related application programs can be more efficiently developed as UML enabled applications. This efficiency is gained by presenting the DICOM structure in the context of a UML structure, i.e., the model or template, thereby eliminating the learning curve and context-switch difficulties typically encountered when dealing with new languages. As is known in the art, the assimilation of the rules, conventions, idiosyncrasies, etc. of a language generally comes with time and experience. In like manner, the appreciation of the interrelationships among data items and entities is highly dependent upon an appreciation of the interactions and dependencies that are implied by the modeling language used to express these interrelationships. By presenting the DICOM specification as a UML representation, the experiences of the UML programmer, systems analyst, technical writer, engineer, manager, etc. are advantageously used, thereby potentially reducing the costs associated with an application program development.

[0064] As illustrated in FIG. 2A, a converter 120 converts the elements of the DICOM specification 110 into elements of a UML structure 130. This UML structure 130 can be used to facilitate the development 140 of an application program 150 that uses the UML structure 130, or it can be further converted, via converter 160, into elements of an DICOM Document Type Definition (DTD), illustrated at 170. The DICOM DTD 170 facilitates the development 180 of an DICOM-enabled application program 190.

[0065] Note that the UML application program development 140 and the DICOM application program development 180 can be expected to require fewer resources than an application program development that uses the DICOM specification directly. A number of UML utility routines can be expected to be available for use in this development, based on the increasing use of UML in the computer industry. Also, UML is an object-oriented language, and an objective of the object-oriented paradigm is to facilitate the transport and re-use of object-oriented software.

[0066] FIG. 2B illustrates an example flow diagram for converting the DICOM reports 210 into UML 230 and DICOM 270 documents for processing by the aforementioned UML 150 and DICOM 190 enabled applications, in accordance with this invention. That is, the programs 150, 190 whose development was facilitated by the DICOM converters 120, 160 of this invention, or other UML programs, can be used to process individual DICOM reports 210, merely by converting the reports 210 into corresponding UML 230 and DICOM 270 documents, respectively. Note that the term “document” is used herein for ease of reference, and includes any of a variety of embodiments, including data that is temporarily or permanently stored in a file, data that is stored in memory, or on a disk, data that is communicated among processing systems, and so on. The UML 150 or DICOM 190 applications may include, for example, applications that render 280 the content of the individual DICOM report 210 for viewing by a diagnostician, applications that transfer the content of the individual DICOM report 210 to processing facilities that do not have DICOM enabled applications, applications that collect the information from individual DICOM reports 210 and perform analyses for clinical studies, and so on.

[0067] FIG. 3 illustrates an example DICOM to UML converter 300. Example DICOM entities 311-317 are illustrated on the left of the converter 300, and corresponding UML entities 321-327 are illustrated on the right of the converter 300. In accordance with this invention, particular rules are associated with the various DICOM entity types to effect the conversion. FIG. 3 is provided for exemplary purposes, and is by no means exhaustive, as will be readily seen by those skilled in the art.

[0068] FIG. 4A illustrates an example flow diagram for converting DICOM Attribute names to UML Class Attribute names. The DICOM attribute name is initially copied to form the UML attribute name, at 410, and converted to lower case, at 420. Via the loop 430-480, each blank and hyphen character is identified, at 440, and replaced with an underscore character, at 450, and each apostrophe or bracket character is identified, at 460, and deleted from the UML attribute name, at 470. The resultant UML attribute name is stored, at 490.

[0069] The DICOM enables Content Items unlimited recursion. FIG. 4B illustrates this recursion in the DICOM Document Content module 500, via the Document Relationship Macro 520. As illustrated, the Document Relationship Macro 520 may include a Content module 530, and this Content module 530 may contain a document relationship 532 that includes another Content module 540, and so on. In accordance with this invention, as illustrated in FIG. 4C, the DICOM Document Content UML representation 560 includes elements DocumentRelationship 570 and DocumentContent 590, and the Document Relationship representation 570 and the Content representation 580 each reference each other, thereby effecting a recursive relationship.

[0070] Using the above rules for mapping the elements of DICOM into corresponding elements of UML, DICOM specification documents and DICOM content documents are converted into a modeling language that is more often used by computer professionals, thereby providing the opportunity to ease the task and cost of producing application programs that can be used for processing DICOM related material.

[0071] FIG. 5 is a description of prior art UML profile for XML schema. XML applications in UML are discussed by Carlson, David, MODELING XML APPLICATIONS WITH UML-PRACTICAL E-BUSINESS APPLICATIONS, Addison-Wesley, 2001.

[0072] As discussed above, the DICOM information model contains a set of IODs, IEs, Modules, Macros, Sequences, and atomic Attributes. In order to guide the generation of UML models for each IOD, a stereotype is defined for each IO. The inventive UML profile for DICOM is shown in FIG. 6. Those skilled in the art will understand that the FIG. 6 description is not exhaustive, and may be extended with new stereotypes if necessary. Each of the FIG. 6 stereotypes is also described in Table 5.

[0073]

TABLE 5
Summary of stereotype description of DICOM profile
Stereotype	Base Class	Description	Tagged values	Constraints
IOD	XSDtopLevelElement	A IOD stereotype	None.	This class must
		represents a DICOM		be assigned to
		IOD definition		exactly one class.
		containing a set of
		DICOM IEs. It can be
		represented as a top level
		element in XML Schema
		representation.
IE	XSDcomplexType	A IE stereotype	None.	None.
		represents a DICOM IE
		which contains one or
		more Modules. It can be
		mapped to a
		complexType in XML
		Schema specification.
Module	XSDcomplexType	A Module is a stereotype	None.	None.
		describing the DICOM
		attributes within a
		DICOM Module. It is
		represented as a
		complexType in XML
		Schema specification.
Macro	XSDcomplexType	A Macro is a stereotype	None.	None.
		that describing a DICOM
		Macro. It is represented
		as a complexType in
		XML Schema
		specification.
Sequence	XSDcomplexType	A UML class with this	codeId - a value	The class with
		stereotype represents a	indicates the DICOM tag	this stereotype
		DICOM sequence	of this sequence.	must be the
		attribute that contains a	codeMeaning - a string	destination of
		set of DICOM attributes.	gives its DICOM name	unidirectional
		It is represented as a	of this sequence.	association in the
		complexType in XML	vr - a value indicates the	UML model.
		Schema specification.	value representation of
			DICOM sequence. Its
			default value is ‘SQ’.
SequenceItem	XSDcomplexType	A UML class with this	None.
		stereotype represents a
		Sequence item within a
		Sequence. It is
		represented as a
		complexType in XML
		Schema specification.
Attribute	XSDcomplexType	A UML class with this	codeId - same as above.	None.
		stereotype represents a	codeMeaning - same as
		DICOM attribute. It is	above.
		represented as a	vr - same as above
		complexType in XML	without default value.
		Schema specification.	nillable - a Boolean
			indicates if the value of
			this attribute can be
			NULL or not.
Datatype	XSDsimpleType	A Datatype is a	None.	These
		stereotype describing a		simpleTypes are
		DICOM value		derived from
		representation. It is		built-in types in
		represented as a		XML Schema
		simpleType in XML		specification.
		Schema specification.
Enumeration (from	See Carlson, pp. 307.	This stereotype	See Carlson, pp. 307.	None.
UML profile for XML)		represents the defined
		terms or enumerated
		values of a DICOM
		Attribute.
XSDelement (from	See Carlson, pp. 309.	This stereotype	See Carlson, pp. 309.	None.
UML profile for XML)		represents the DICOM
		Attribute within a
		Module, Macro, or
		Sequence.
XSDattribute (from	See Carlson, pp. 309.	This stereotype	See Carlson, pp. 309.	The attribute
UML profile for XML)		represents the attributes		datatype may not
		(tag, DICOM name,		refer to a class
		DICOM type) of a		specification.
		DICOM Attribute.

[0074] For general purpose of UML modeling, the stereotypes defining the inventive profile for DICOM refer to the general stereotype base constructs such as Package, Component, and Class. The XML stereotype bases are referenced if the main objective is to generate XML schemas from UML construct models. To that end, OCL is a new notational language, a subset of the industry standard UML. OCL allows software architects, designers, and software developers to write constraints over the object models. In OCL, a constraint is a restriction on one or more values of an object-oriented model or system. There are three types of constraints defined in OCL: preconditions, postconditions, and invariants. Pre- and postconditions are defined for operations in objects. Invariant is a constraint that states a condition expressed in a mathematical expression that must always be met by all instances of a class, type, or interface. Constraints convey a number of benefits such as better documentation, improved precision, and communication without misunderstanding.

[0075] Following a set of syntaxes defined in OCL, various constraints can be put on attributes, operations, classes, and other types. FIG. 7 shows an example of putting an invariant constraint on an association. FIG. 7 is meant to communicate the constraint that a House can have a Cellar or a Basement, but it can't have both at the same time.

A Framework of Precise UML Modeling

[0076] A modeling framework for a DICOM IOD is shown in FIG. 8. First, naming conventions must be defined in order to keep a variety of construct names consistent. Second, how to model different 10 is discussed. Then the modeling rules for different DICOM constraints and finally, a way of expressing DICOM VRs is presented.

[0077] Naming Conventions

[0078] The following steps are required for maintaining consistency for construct names:

• [0079] 1. Change all the upper case letters to lower case.
• [0080] 2. Replace any blank space between two words with an underscore (_).
• [0081] 3. Remove apostrophes and brackets.
• [0082] 4. Replace hyphen (−) and slash (/) with underscore (_).
• [0083] 5. A string ‘IE’ is appended at the end of names for IE elements; ‘Module’ at the end of names for Module elements.
• [0084] 6. For class names or class type names, capitalize the first letter of each word.
• [0085] 7. ‘Seq’ is short for ‘Sequence’ in the class name; ‘seq’ in the referring class variable name; and ‘seq_item’ in the referred class variable name.

[0086] Packages and Namespace

[0087] To avoid repeat definitions of IEs, Modules, Sequences, Sequence Items, Macros, and atomic Attributes and also to reuse their definitions, a UML package associated with each IO type is created. For the current version of DICOM specification, IODs package contains 18 IODs (FIG. 8). FIG. 9 shows the UML package diagram of the DICOM information model that illustrates all the packages and their relationships.

[0088] Each package contains different types of IOs. DICOM IODs contains IOD classes, DICOM IEs contain IE classes, DICOM Modules contain Module classes, DICOM Sequences contain Sequence type classes, DICOM SequenceItems contain Sequence Item type classes, DICOM Macros contain Macro type classes, DICOM Attributes contain Attribute Type classes, DICOM Datatypes contain VR type classes, and DICOM Enumerations contain enumerated or defined terms.

[0089] Information Object Definitions

[0090] An example UML profile for DICOM that specifies the stereotypes for various DICOM IOs including IODs, IEs, Modules, Macros, Sequences, and Attributes has been introduced above. These stereotypes will be used for modeling different IOs. Therefore, an IOD is modeled as a IOD type, an IE as IE type, a Module as Module type, a Sequence as a Sequence type, a Sequence item as a SequenceItem type, a Macro as Macro type, and an Attribute as a Attribute type. Comprehensive DICOM SR IOD is one of three DICOM Document IODs which serve as an example of a UML model shown in FIG. 10.

[0091] It is clear to see that the IOD class is at the root level followed by IE classes and then Module classes. It also shows what is the base stereotype of each class and which package a class is derived from. Down to the Module level, for example, PatientModule, whose DICOM definition shown in Table 6, is represented as in FIG. 11. Table 6. Definition of Patient Module

[0092]

Attribute Name	Tag	Type	Attribute Description
Patient's Name	(0010,0010)	2	Patient's full name.
Patient ID	(0010,0020)	2	Primary hospital identification
			number or code for the patient.
Patient's Birth	(0010,0030)	2	Birth date of the patient.
Date
Patient's Sex	(0010,0040)	2	Sex of the named patient.
			Enumerated Values:
			M = male
			F = female
			O = other
Referenced	(0008,1120)	3	A sequence which provides refer-
Patient			ence to a Patient SOP Class/In-
Sequence			stance pair. Only a single Item
			shall be permitted in this Sequence.
>Referenced	(0008,1150)	1C	Uniquely identifies the referenced
SOP Class UID			SOP Class. Required if Referenced
			Patient Sequence (0008,1120) is
			sent.
>Referenced	(0008,1155)	1C	Uniquely identifies the referenced
SOP Instance			SOP Instance. Required if Ref-
UID			erenced Patient Sequence
			(0008,1120) is sent.
Patient's Birth	(0010,0032)	3	Birth time of the Patient.
Time
Other Patient	(0010,1000)	3	Other identification numbers or
IDs			codes used to identify the patient.
Other Patient	(0010,1001)	3	Other names used to identify the
Names			patient.
Ethnic Group	(0010,2160)	3	Ethnic group or race of the patient.
Patient	(0010,4000)	3	User-defined additional
Comments			information about the patient.

[0093] Each DICOM Attribute is modeled as a variable member of the PatientModule class with a Class type associated with it. The stereotype of PatientModule class is Module. And the stereotype of each class member is XSDelement. The DICOM Macros are represented in the same way as the Modules. The only difference is that the stereotype base of the Macros is Macro (see Table 7 and FIG. 12).

[0094]

TABLE 7
DICOM definition of Numeric Measurement Macro
Attribute Name	Tag	Type	Attribute Description
Measured	(0040,A300)	2	This is the value of the Content
Value			Item.
Sequence			Zero or one Items shall be permit-
			ted in this sequence.
>Numeric	(0040,A30A)	1	Numeric measurement value. Only
Value			a single value shall be present.
>Measurement	(0040,08EA)	1	Units of measurement. Only a sin-
Units Code			gle Item shall be permitted in this
Sequence			sequence.
>>Include ‘Code Sequence Macro’ Table 8.8-1
Defined Context ID is 82.

[0095] The DICOM Sequences are represented as Sequence typed classes with an XSDelement typed sequence item attribute and three XSDattribute typed auxiliary attributes: codeId, codeMeaning, and vr. Here vr takes the default value of ‘SQ’. If a Sequence only contains a Macro, the sequence item is a type of Macro (see FIG. 13).

[0096] Actually, there is no Referenced SOP UID Macro defined in the current version of the DICOM specification. The author modeled ‘referenced_sop_class_uid’ (Referenced SOP Class UID) and ‘referenced_sop_instance_uid’ (Referenced SOP Instance UID) together as a Macro typed class named ReferencedSOPUID for convenience because these two DICOM Attributes are referred to many times as shown in Table 2.

[0097] If a Sequence contains more Attributes it will be a type of SequenceItem which represents all the Attributes within the Sequence, see Table 7 and FIG. 14. Because of the specific characteristics of DICOM data elements and structures, e.g., each DICOM Attribute with its own attributes such as name, DICOM Tag, Type, and description, each atomic Attribute is represented as a class that contains not only its data element but also its attributes. This approach is taken mainly to help automate the generation of XML schemas (see FIG. 15).

Constraints

[0098] The constraints in the DICOM information model are represented as a type of invariant constraint. They can be put on attributes or associations. FIG. 16 shows a sample UML diagram of Document Content Macro with OCL constraints.

[0099] As shown above, the multiplicity of date time is [0 . . . 1]. Based on its DICOM description in Table 1, it shows that it is required if Value Type (0040,A040) is DATETIME. An Invariant constraint is put to express this constraint as shown in FIG. 16 and highlighted in FIG. 17. However, some DICOM constraints cannot be represented using Invariant constraint such as the condition for Specimen Identification Module.

[0100] Types of Data Elements

[0101] Five Types of data elements are defined in . They are Type 1, Type 1C, Type 2, Type 2C, and Type 3, where two parameters combined together express this kind of constraint, shown in Table 8.

[0102]

TABLE 8
Representation of Types of data elements
	Multiplicity	Value of nillable attribute
	Type 1	1	False
	Type 1C	0 . . . 1	False
	Type 2	1	True
	Type 2C	0 . . . 1	True
	Type 3	0 . . . 1	True

[0103] The SOP Common Module is an example (Table 9) which shows the first two Attributes are Type 1, the third Type 1C, and the rest type 3.

[0104]

TABLE 9
Part of SOP Common Module
Attribute Name	Tag	Type	Attribute Description
SOP Class UID	(0008,0016)	See	Uniquely identifies the SOP
		C.12.1.1.1	Class. See PS 3.4.
SOP Instance	(0008,0018)	See	Uniquely identifies the SOP
UID		C.12.1.1.1	Instance. See PS 34.
Specific	(0008.0005)	1C	Character Set that expands or
Character Set			replaces the Basic Graphic
			Set.
			Required if an expanded or
			replacement character set is
			used.
			See C.12.1.1.2 for Defined
			Terms.
Instance	(0008,0012)	3	Date the SOP Instance was
Creation Date			created.
Instance	(0008,0013)	3	Time the SOP Instance was
Creation Time			created.
Instance	(0008,0014)	3	Uniquely identifies device
Creator UID			which created the SOP
			Instance.

[0105] FIG. 18 shows that SOP Class UID and SOP Instance UID are Type 1 so that the multiplicity of sop_class_uid and sop_instance_uid is 1 and their nillable attributes take the default value, false, based on the stereotype of Attribute. Specific Character Set is a Type 1C in that the multiplicity of specific_character_set is [0 . . . 1] in SOPCommonModule class and the nillable attribute takes a value of true. This is also true for the Type 3 data elements such as Instance Creator UID Attribute.

[0106] Value Representations

[0107] In the current version of the DICOM specification there are 26 VRs defined. Each of them has different constraints or patterns which can be expressed in the OCL notation language. FIG. 19 gives the UML representation (Table 4) of a subset of VRs. The AS—Age string as described in Table 4, is represented to be a Class inherited from string Type with a pattern of ‘[0-9] {3} [DWMY]’. MaxLength restricts the length of string AE and CS to 16 bytes.

[0108] Enumerated Values and Defined Terms

[0109] There are two ways to represent enumerated values or defined terms. One is to use the ‘enum’ OCL type to list all the enumerated values as shown in FIG. 20. The other is to use stereotype enumeration as shown in FIG. 21 The latter is good for more than two enumerated items.

Sample Modeling: A UML Model for Comprehensive DICOM IOD

[0110] Based on the above framework, a UML model of comprehensive DICOM IOD was created. The UML model of comprehensive DICOM IOD captures all the IEs, Modules, Macros, Sequences, and Attributes including various constraints, data structures, DICOM tags, and DICOM VRs. FIGS. 22-31 highlight DICOM the UML class diagrams of the Modules of DICOM Comprehensive DICOM IOD and DICOM VRs. Such a model assists system architects and developers in understanding the DICOM information model. Also, XML schemas can be easily generated using XMI from this model, which will not be addressed in this document. The UML models for other DICOM IODs can also be built based on this modeling framework. Some common Modules such as Patient, General Study, and SOP Common can be reused.

[0111] There are three data types related to date or time in the DICOM specification: DA, DT, and TM. They are of string type with different constraints. There is no problem in representing these constraints using regular expressions. But the issue is the date representation from DICOM specification is different from that of the XML Schema specification which conforms to the ISO definition. Take date as an example; the format from ISO is ‘CCYY-MM-DD’ (where CC—century, YY—year, MM—month, and DD—day) and the format from DICOM is ‘CCYYMMDD’. This makes it inconvenient to represent DICOM date in XML format and other programming languages. It would be better to modify the DICOM DA, DT, and TM representation so that they are consistent with ISO date and time format.

[0112] Unexpressed Constraints

[0113] Most of the DICOM constraints can be expressed in OCL. However, a few of them cannot be expressed. Take a Type 1C data element, Verifying Observer Sequence as an example, it is required only if the Verification Flag has the value of VERIFIED. In this case, both Verifying Observer Sequence and Verification Flag are DICOM Attributes, and also Verification Flag has known values, which can be expressed in OCL, as shown in FIG. 32. On the other hand, some conditions cannot be expressed in OCL expressions, for example the condition of requiring Specimen Identification in Table 1 is if the Observation Subject is a Specimen. It cannot be represented in a UML diagram because there is no definition of Observation Subject in the DICOM information model. We proposed a temporary constraint—Other, to express these constraints which cannot be expressed in Invariant, Post- and Precondition, see FIG. 33. This is an open issue to the DICOM information model.

[0114] The foregoing merely illustrates the principles of the invention. It will thus be appreciated that those skilled in the art will be able to devise various arrangements which, although not explicitly described or shown herein, embody the principles of the invention and are thus within the spirit and scope of the following claims.