|
|
J Am Med Inform Assoc. 2008 Sep–Oct; 15(5): 671–678. doi: 10.1197/jamia.M2358. | PMCID: PMC2528035 |
Copyright © 2008, American Medical Informatics Association Modeling Functional Neuroanatomy for an Anatomy Information System Jörg M. Niggemann, MD, Dipl Inform, a 1 ![[low asterisk]](/corehtml/pmc/pmcents/lowast.gif) Andreas Gebert, MD, b and Stefan Schulz, MD c aKlinikum rechts der Isar, Technical University of Munich, Munich, Germany bInstitute of Anatomy, University of Lübeck, Lübeck, Germany cInstitute of Medical Biometry und Medical Informatics, Freiburg University Medical Center, Freiburg, Germany Received December 29, 2006; Accepted May 4, 2008. Objective Existing neuroanatomical ontologies, databases and information systems, such as the Foundational Model of Anatomy (FMA), represent outgoing connections from brain structures, but cannot represent the “internal wiring” of structures and as such, cannot distinguish between different independent connections from the same structure. Thus, a fundamental aspect of Neuroanatomy, the functional pathways and functional systems of the brain such as the pupillary light reflex system, is not adequately represented. This article identifies underlying anatomical objects which are the source of independent connections (collections of neurons) and uses these as basic building blocks to construct a model of functional neuroanatomy and its functional pathways. Design The basic representational elements of the model are unnamed groups of neurons or groups of neuron segments. These groups, their relations to each other, and the relations to the objects of macroscopic anatomy are defined. The resulting model can be incorporated into the FMA. Measurements The capabilities of the presented model are compared to the FMA and the Brain Architecture Management System (BAMS). Results Internal wiring as well as functional pathways can correctly be represented and tracked. Conclusion This model bridges the gap between representations of single neurons and their parts on the one hand and representations of spatial brain structures and areas on the other hand. It is capable of drawing correct inferences on pathways in a nervous system. The object and relation definitions are related to the Open Biomedical Ontology effort and its relation ontology, so that this model can be further developed into an ontology of neuronal functional systems. Functional Anatomy is concerned with anatomical entities grouped together to perform a physiological function. In Functional Neuroanatomy, a typical example is the visual system. Neuroanatomy textbooks are usually illustrated with drawings similar to the ones in which exhibit an important peculiarity of Functional Neuroanatomy: the main focus is not on single anatomical entities, such as in the musculoskeletal system, but on collections of uniform entities, neurons, which are considered its functional units. In textbook illustrations they are usually visualized as blobs. In the text, however, adequate expressions such as “a collection of neurons within the lateral geniculate body projects to a collection of neurons within area 17” are unusual. Such expressions are precise, but clumsy. Instead, the same scenario is expressed as “the lateral geniculate body projects into the area 17.” This compromise, however, obscures some detail—especially when the anatomical entities being described are constituted, in reality, by several qualitatively different collections of neurons. A symbolic representation of anatomy—especially for didactic purposes—should represent facts on a comprehensible, but detailed level, and yield output on the level of object names. To this end, we here diligently define both levels, and establish the transition rules between them. The model presented here is derived from an approach that had been implemented in the early anatomy e-learning system “Anatom-Tutor” 1 and is put into the context of current ontologies such as the Foundational Model of Anatomy (FMA). Representations of Anatomy and Neuroanatomy and Their Respective Focus The organization of the nervous system of various species is currently represented in several large systems. NeuroNames is a nomenclature of human and macaque brain structures, designed as a tool for indexing digital databases of neuroscientific information. It was started in the late 80's as a MacIntosh hypercard stack 2 and has since been developed into a web-accessible database. 3 Its primary objects are anatomical entities identified by names. 4 The Foundational Model of Anatomy (FMA) 5 is a frame-based ontology, originally designed to represent the macroscopic anatomy of the human thorax. It was assumed that this would cover all complexities of anatomy, and in this way the model could later be easily extended to other body areas and to other granularity levels such as cellular and subcellular structures. When incorporating neuroanatomy from NeuroNames, the authors themselves encountered diverse problems with this approach. 6–8 The Foundational Model was further enhanced to accommodate so-called input/output relationships in order to represent the flow of information through the nervous system. The corresponding slots “Gets Input From” and “Sends Output To” are, however, not restricted in terms of domain and range constraints. For example, the class “organ region” has this slot. Most neuroanatomical objects are listed under this class, but also others such as “auricle of heart” and “segment of tooth.” Furthermore, it is not distinguishable whether two connections which an object exhibits are collaterals or whether they originate in different populations of neurons within the object. Therefore, an important subfield of neuroanatomy, viz. functional neuroanatomy, which is concerned with pathways and systems, can not readily be represented in the FMA in its current form. Brain Architecture Management System (BAMS) 9 and Brainmaps.org 10 are web accessible repositories and digital atlases on neuroanatomy, containing each a compilation from different atlases and for different species. Both contain data on connections, and in both, like in the FMA, collaterals can not be distinguished from axons originating in different neuron populations. Other web accessible anatomy information resources: Kim et al. have recently conducted a review of 40 online resources on anatomy, 11 most of them aimed at education. Six of them are dedicated to Neuroanatomy, four of which still in service (see Table A1 in the online appendix, available at http://www.jamia.org). None of them contains explicit information on functional systems. Other recent neuroanatomy teaching systems are Brain Project/3D-Brain 2.0 12 and BrainTrain. 13 Like the other resources cited above, these do not represent neural connections in functional systems. In summary, the aforementioned neuroanatomical systems address nomenclature, hierarchical taxonomy, part-of hierarchy, connections, cytoarchitecture, and different mappings of the cortex. In all of them the primary objects of representation are macroscopic, morphologically-defined anatomical structures. Their main disadvantage is their inability to represent internal subgroups of neurons and their connections, the “internal wiring” of neuroanatomical objects. Formal Ontologies Both terms and relations in an ontology have to be well-defined so that automated reasoning becomes feasible and yields meaningful results (see Smith et al., 2005). 14 Recently the Open Biomedical Ontologies (OBO) consortium has compiled the OBO ontology library, a repository of controlled vocabularies developed for shared use across different biological and medical domains. Smith et al. 14 reviewed the definition and use of relations in these ontologies and found, that even the most basic relations part-of and is-a are not always used in consistent fashion both within and between ontologies. The authors then proceeded to define an “ontology of relations,” 15 which we will use here as a starting point. Defining Anatomical Objects as Counts, Collections, Mass, etc The idea proposed in this paper is to define the basic concepts of functional neuroanatomy in terms of groups of uniform objects. This idea is related to the works of Bittner, Rector et al., 16–18 who suggested the description of the relation between collectives of uniform objects and each of their single constituents as has-granular-part, a subrelation of has-part. Functional Anatomy Johansson et al. 19 have introduced “pure structural anatomy” and “pure functional anatomy” as perspectives on anatomy. They argue that there are two ways to draw boundaries between spatial parts of the body: structural and functional. About the functional parts the authors state: “A part-of relation between such spatial-functional entities goes parallel to a sub-function relation among the functions associated to the spatial-functional entities”. Niggemann 20 outlined a similar perspective with respect to neuroanatomy, from the point of view of knowledge representation. He argued that the spatial dimension is not even necessary in order to define a “functional object” (however, this topic is beyond the scope of this article). Example: The Visual System Our model is introduced using the visual system as an example. We concentrate on the reflex systems of the pupillary light reflex and the accommodation reflex (). Both reflex arcs involve a group of neurons in the parvocellular oculomotor nucleus which send their signals (efferences) to a group of neurons in the ciliary ganglion (e, f, g in A and B). Important to note is that the two reflex arcs involve two different groups of neurons in the pretectal nucleus, the parvocellular oculomotor nucleus and the ciliary ganglion. The sentence “The parvocellular oculomotor nucleus sends output to the ciliary ganglion” is correct in both reflexes, but this information alone is not sufficient to distinguish between the two reflex arcs. We need a way to represent different connections within the structures, such as they are depicted in C. In general, we want to represent the most detailed information available. For this purpose we introduce a model which represents groups of neurons, as well as groups of neuronal cell bodies and groups of neuronal axons with their myelin sheaths. The model we are proposing addresses specific aspects of Neuroanatomy. This model is meant to be useful under many perspectives: both on an instance (token, individual, particular) and a class (type, concept, universal) level, and—from a philosophical perspective—both under a “conceptualist” and a “realist” point of view. The model of anatomical groups is introduced in an atemporal way, as we intend to model one idealized, prototypical, “normal” instance of a human nervous system at one specific point in time. This is the perspective taken e.g., in anatomical atlases, which introduce general “classes” such as “flat bone” but then describe every bone in the body as one instance of such a class. We introduce the terms group and individual with the relations has-member and member-of. From the OBO relations ontology 15 we choose the following ones as primitive relations in our model: part-of, located-in, adjacent, instance-of, is-a. (New terms which we introduce as names of classes/universals are printed in small caps, such as F unctional-G roup. We consider relations netween instances only, they are printed in italics.) Analyzing the Underlying Structures In order to express “groups of neurons” and “groups of neuron parts,” we start with a brief inspection of the underlying particulars, the neurons themselves. We do this on the “cell” and “subcellular” levels of granularity. Here we look at a neuron and its relevant segments, and then define the basic relations of adjacency and efference. Afterwards, we can transpose these definitions into the higher granularity level of groups, which lies between the “cell” and the “tissue” granularity levels. Nerve cells (neurons) are the individual information processors of the neural system. depicts a typical neuron. It has a long process, the axon, which conducts electric signals to target cells (targets can be other neurons or effectors such as muscle cells). Every axon is part of some neuron. Most axons are covered by a myelin sheath which consists of cells or parts thereof. An axon together with its myelin sheath is called a fiber. Fibers can be partitioned into segments according to the Ranvier's nodes in the myelin sheath. The segment of a fiber between two nodes is called the internodal segment. The terminals of the fiber form part of the conducting connection to the target cell. The connection or synapse overlaps the source neuron as well as the target cell. The central part of the neuron which includes the cytoplasm, cell nucleus and the surrounding plasma membrane but excludes the axon, dendrites (see below) and any other neurites is called the cell body. Many neurons also have dendrites: protuberances to which synapses connect and which conduct the excitation to the cell body. In this functional model, it is irrelevant whether a terminal connects to a dendrite or directly to the cell body, we therefore abstract from this detail. Definition (R-Segment, Fiber-Segment) The smallest functional segment of a fiber is an internodal segment together with its following node; we will give it the name R-Segment. We also need to define fiber segments of arbitrary length: A Fiber-Segment is one R-Segment or a sequence of connected R-Segments, possibly ended with a terminal.
This definition is needed when we want to express the meaning of sentences like: “The fibers originating in the retina run through the optic nerve”. This can be translated as “Each of the fibers has some Fiber-Segment as part which is spatially located-in the optic nerve”. Definition (Neuron-Structure) We define as Neuron-Structure all sections of a neuron introduced in the previous description and definition.
Relations Between Neuron-Structure The relations between the cell-structure-level objects serve as an orientation for defining relations between group objects. Definition (Efferent-to, Afferent-to, Direct-efferent-to, Direct-afferent-to) N euron-S tructures pass electrical excitation: A Neuron-Structure is in relation efferent-to another one, if and only if it passes signals to it. Efferent-to is a transitive relation. Direct-efferent-to is a subrelation of efferent-to, it additionally requires spatial adjacency and therefore is not transitive. The same holds for the inverse, afferent-to and direct-afferent-to.
A neuron is in relation direct-efferent-to another one, if and only if it has as a part a terminal, which is direct-efferent-to, the second neuron.
From Neuron-Structures to Groups Now the transition is made from Neuron-Structure to groups. To achieve this, we will define an equivalent in the group level for each object and each relation on the Neuron-Structure level (). We uniformly name all group objects with the postfix “-group”. Functional-Group is the common subsumer of all groups that can be derived from the Neuron-Structure defined above, together with groups of neurons themselves. It can be defined more generally by use of the criteria of a common connection pattern or participation in a common function. shows a taxonomy of the classes defined in this section. Definition (Functional-Group) A F unctional-G roup is a collection of neurons or neuron segments of the same type, such that: A F unctional-G roup is therefore a kind of “ObjectAggregate” as defined in the Basic Formal Ontology (BFO) 21,22 as “An independent continuant that is a mereological sum of separate objects and possesses non-connected boundaries”. Similarly, the definition of F unctional-G roup is consistent with Simons' notation of “group.” 23 Relations between Functional-Groups We define two sets of relations between Functional-Groups. One describes relations with respect to signal conduction, the other with respect to their members. Definition (Relations with Respect to Connections) The same relations that are defined for Neurons and N euron-S tructure can be used between F unctional-G roups: efferent-to/afferent-to: We define efferent-to as the transitive relation which represents signal passing. A Functional-Group g 1 is in relation efferent-to to another Functional-Group g 2 if and only if for every member m1 of g 1 there is a member m2 of g 2 such that m1 efferent-to m2. afferent-to is the inverse relation.
direct-efferent-to/direct-afferent-to: Similarly, the non-transitive relations direct-efferent-to/direct-afferent-to can be derived from the underlying Neuron-Structures: A Functional-Group g 1 is in relation direct-efferent-to to another Functional-Group g 2 if and only if for every member m1 of g 1 there is a member m2 of g 2 such that m1 direct-efferent-to m2. direct-afferent-to is the inverse relation.
Definition (Mereological Relations) subgroup-of: A Functional-Group g 1 is subgroup-of another Functional-Group g 2 if and only if every member of g 1 is also member of g 2. Thus the subgroup-of relation corresponds to a part-of relation between groups.
overlap: A Functional-Group g 1 overlaps another Functional-Group g 2 if and only if there is at least one member of g 1 which is also member of g 2.
equal: Two Functional-Groups are equal if and only if they have exactly the same members.
sum: A Functional-Group g 1 is sum of Functional-Groups g 2 and g 3 if and only if every member of g 1 is also member of g 2 or g 3 and both g 2 and g 3 are subgroups of g 1.
Subclasses of Functional-Group Definition (Group of Complete Neurons: Neuron-Group) A group of neurons will be called Neuron-Group.
Definition (Group of Cell-Bodies: Cell-Body-Group) A Cell-Body-Group is a group of cell bodies, all of which share a common function or common connections. Therefore, looking at functions and subfunctions or at connections in different levels of granularity leads to different formations of Cell-Body-Groups.
Definition (Group of Fibers: Fiber-Group) Fiber-Group refers to a group of complete fibers (axons with myelin sheath if applicable) from cell bodies to terminals. Note that a complete fiber is a (maximal) Fiber-Segment, such that Fiber-Group must be considered a subtype of Fiber-Segment-Group.
On the group level like on the Neuron-Structure level we need to define a smallest (functional) segment of a Fiber-Group. We derive this from the smallest (functional) segment of a fiber, the R-Segment. So we can define: Definition (Group of R-Segments: Atomic-Fiber-Segment-Group) An Atomic-Fiber-Segment-Group is a group of parallel R-Segments within the same Fiber-Group.
Definition (Group of Fiber-Segments: Fiber-Segment-Group) A Fiber-Segment-Group is a collection of parallel Fiber-Segments
The example used in the definition of Fiber-Segment can be reformulated using Fiber-Group: “The Fiber-Group originating in the retina runs through the optic nerve”, with the interpretation that there is a Fiber-Segment-Group which is part of the Fiber-Group and which is spatially located-in the optic nerve. Definition (Group of terminals: Terminal-Group) Finally, Terminal-Group can naturally be defined as a group of terminals which belong to the same Fiber-Group and connect to the same target.
The Level of Macroscopic Objects The goal for a readable output of an anatomy information system is to transform the fact “The neural excitations which contribute to the right pupillary light reflex are conducted via cbg1, cbg2, cbg3 …” (following efferent-to relations between Cell-Body-Groups) into the sentence: “… conducted via: right temporal retina, right lateral geniculate corpus, …”. That means we have to replace the unnamed Cell-Body-Groups with named anatomical objects. In order to make the transition from the level of groups to the level of macroscopic, named anatomical objects, we need to define the kind of objects that are candidates for “named anatomical objects” in this context. Definition (Macroscopic-Neuroanatomical-Entity) A M acroscopic-N euroanatomical-E ntity is an object of neuroanatomy which - 1 is visible to the human eye without the help of a microscope
- 2 occupies a connected space
(An object occupies a connected space if for every two spatial points A and B within the object there exists a path from A to B which is entirely in the object). Relations of Macroscopic-Neuroanatomical-Entities In the scope of this article, the only interesting relation between Macroscopic-Neuroanatomical-Entities is part-of/has-part. As mentioned above, we consider this to be a primitive relation. We now define located-in as a relation between Functional-Groups and Macroscopic-Neuroanatomical-Entities. We define the relation through the group's members, whose location in the respective object can be microscopically verified. Definition (Located-in as Relation between Functional-Group and Macroscopic-Neuroanatomical-Entity) A Functional-Group g is located-in a Macroscopic-Neuroanatomical-Entity m if and only if each member of g is located-in m.
Now we can proceed to reconstruct the FMA relation sends-output-to between Macroscopic-Neuroanatomical-Entities. Definition (Receives-input-from, Sends-output-to) A Macroscopic-Neuroanatomical-Entity m 1 sends-output-to another Macroscopic-Neuroanatomical-Entity m2 if and only if there are some Cell-Body-Groups f1, f2 such that f1 located-in m and f2 located-in m2 and f1 efferent-to f2.
receives-input-from is the inverse.
Validation through Example Proof of Completeness, Correctness, Novelty and Advantage In this section we will prove that the Functional-Group model is a valid and useful extension to models such as the FMA or BAMS. We use the example of the visual system to illustrate the steps of the proof. Claim: Augmenting a current model such as the FMA with the Functional-Group model - 1 does not lead to any loss in information or capabilities
- 2 adds functionality which these models currently do not have
In order to prove that the F unctional-G roup model can indeed add functionality to current models, we choose the FMA as an example and take the following steps: - 1 We show that the current FMA can be transformed into the Functional-Group enhanced format. Especially, we show that
- a) this process can be reverted, which proves that no information is lost in the conversion (completeness)
- b) the classes which are automatically added correctly reflect anatomical reality (correctness)
- 2 We prove that in the Functional-Group enhanced FMA, additional information (namely internal wiring) can be stored
- a) which could not be stored in the original FMA (novelty)
- b) which allows inferences that could not be drawn in the original FMA (advantage)
The complete proof is given in the online appendix. Here we outline the main steps. 1. Transformation of the FMA into a Functional-Group enhanced format The transformation mainly involves - 1 Augmentation of the class hierarchy by the classes defined in section IV
- 2 Processing of all “Sends Output To” links
The result is illustrated in A and B: The information contained in the former “Sends Output To” links (A) is now represented by Functional-Groups contained in the objects and their efferent-to connections (B). The detailed proof also shows that the original information can be recovered from the extended form. The process corresponds to redefining the relation “Sends Output To” as in Definition (receives-input-from, sends-output-to) in section IV. Completeness: As we have only added classes, nothing is lost in this step. The original information in the “Sends Output To” links can be recovered, proving that no information is lost. Correctness: As outlined in section IV, the new classes by design correctly reflect anatomical entities. Since the procedure introduces new objects (the individual F unctional-G roups and their relations), we must prove that these reflect anatomical reality. It is a fundamental truth in neuroanatomy, that every neuronal signal - – can only originate in a neuron cell body (resp Cell-Body-Group)
- – must be conducted by an axon (resp Fiber-Group), and
- – can be delivered/transmitted only through a terminal (resp Terminal-Group)
Therefore it is a valid conclusion that each “Sends Output To” relation implies the existence of such structures, and it is correct to represent them explicitly. In summary, we have shown that the Functional-Group enhanced FMA (with no additional information added) and the original can be mutually transformed into each other. This proves claim 1, that no information is lost in the transformation process. Also we have shown that the enhanced model correctly reflects anatomical reality. 2. Adding new information The Functional-Group enhanced FMA now allows collators to add new information which wasn't possible in the original model. In this section we have to prove that the same information cannot be stored in the current FMA (novelty) and that it is useful (advantage). 1. “Internal wiring” In this step, decisive new information can be added which makes the Functional-Group model so effective. It is the step from B to C: New Functional-Groups are added and new efferent-to links are established between Terminal-Groups and Cell-Body-Groups. Novelty: By design of the FMA, internal connections within an object cannot be represented. This is because connections are not objects themselves, they are just attributes of the macroscopic anatomical objects. In the extension however, the connections are represented by objects within the macroscopic anatomical objects (Functional-Groups) which can again be linked to each other (direct-efferent-to). Therefore, internal connections within a macroscopic anatomical object can only be represented in the extension. Advantage: Representing internal connections is a basic step for distinguishing known pathways from candidate pathways (see next paragraph). 2. Known Pathways and Functional Systems The Functional-Group extension allows one to distinguish between complex known pathways and candidate pathways, and to represent complete functional systems such as the accommodation reflex system as in C. Novelty: In current models such as FMA or BAMS, candidate pathways can be generated by recursively following “Sends Output To” links. It is not guaranteed that these candidates represent any anatomical or physiological reality. In contrast, the addition of the “internal wiring” of direct-efferent-to links between Terminal-Groups and Cell-Body-Groups where these are known, allows the enhanced model to discern different functional systems. Advantage: Known pathways can be represented and can be searched and followed, and they can be discerned from candidate pathways. One of the most important kind of entities in functional neuroanatomy, functional systems, can now be represented. A detailed example of how the system follows a known pathway is given in section “Following pathways in the Functional-Group model versus FMA and BAMS” in the online appendix. 3. “White matter” pathway information The resulting pathway structures can also represent the path that a connection takes through “white matter” structures (those that primarily contain axons or parts thereof). shows the structures through which fibers from the retina pass on their way to the lateral geniculate body (numbers 3–5 in C). With the Functional-Group extension, this information is now explicitly represented so that an information system can output information like “fibers from the retina run through the optic nerve, the optic chiasm, and the optic tract”. This is equivalent to the statement “each of the Fibers has a Fiber-Segment as part which is spatially located-in the optic nerve, the optic chiasm, and the optic tract.” Details are given in the online appendix. Novelty: The “Continuous With” relation of the original FMA is not constrained to objects of neuroanatomy; it is also used to link blood vessels, tendons etc. Therefore, the “white matter” pathway information can only be correctly represented using the Functional-Group extension. Advantage: An anatomical information system containing this information can be used to reason about and to teach structure-function relationships such as consequences of lesions to white matter structures. In summary, we have shown that the Functional-Group extended FMA can store information and draw conclusions, which the design of the original FMA does not allow. This proves claim 2 and completes the proof that the Functional-Group model is a valid and useful extension of the FMA. The Functional-Group model has been designed to address the topology of neural connections as a specific aspect of Neuroanatomy. So far we have presented a basic form of the model. Three aspects need further attention: The usability of the model with incomplete information, the specific phenomenon of the cross-connection of pathways, and the addition of more aspects of neuronal connections such as the transmitters involved. Use with Incomplete Information, and Aggregation of Scientific Results The steps described in the section “Validation Through Example” illustrate how incomplete information is handled: when embedded in a “host system” such as the FMA, it is first set up to reconstruct that system's “Sends Output To” information. That information is necessarily incomplete because information about internal wiring is lacking. In this stage, candidate pathways are found by assuming links from every incoming to every outgoing connection of an object. When scientific results about internal connections become available, these can be added to the model as described in the step “Adding new information,” thus enabling the model to represent known pathways and distinguish those from candidate ones. The Cross Connection of the Pathways Bright light causes a constriction of the pupil but not simultaneously an accommodation reaction. However, looking at nearby objects also leads to a constriction of the pupil. This observation is a subject matter of physiology and, as such, not directly represented in a neuroanatomical model. Together with the general axiom that in the nervous system there is no function without connection, we can conclude that “there is some connection from at least one Cell-Body-Group involved in the accommodation pathway to at least one other that is involved in the pupillary reflex pathway”. This first-order logic statement can not be represented in the Functional-Group model (nor can it be represented in one of the other models mentioned). As soon as it is known which Macroscopic-Neuroanatomical-Entity contains this cross connection, the situation can be modeled, for example by adding a collateral from an incoming connection belonging to the accommodation reflex to the Cell-Body-Group of the pupillary light reflex, or by establishing other internal wiring as the scientific results suggest. Scope and Possible Extension: More Detailed Description of Signal Transmission Apart from the visual system, this approach has been successfully tested with representations of the vestibular system, the auditive system, the accessory optical system, long corticofugal tracts of the motoric systems, the extrapyramidalmotoric system, the central limbic continuum, ascending reticular tracts, efferent connections of the cerebellum and thalamo-cortical connections. In its basic form, as presented here, the model only represents the “bare” connections from terminal to target cell. It does not represent dendrites, different kinds of synapses, or different neurotransmitters. In a more advanced form, our model can be expanded to cover such details. The existing objects can then be enriched with attributes such as “adrenergic,” and new intermediate objects can be introduced between synapse and target cell bodies in order to represent dendrites. Further extensions can express characteristics of signal conduction in the fibers and modulations thereof by axo-axonal synapses. We have proposed a model based on groups of cells or cell parts, which serves to represent neuronal connections between brain structures in more detail than it has been thus far possible. The model is capable of drawing valid inferences on pathways in a nervous system. In combination with an ontology of functions, the group-oriented model of functional neuroanatomy can be the starting point for a comprehensive ontology of functional pathways in nervous systems, and can thus augment the Foundational Model of Anatomy when used in an anatomy information system. The authors thank Dr. Reinhard Eggers (Department of Anatomy, University of Luebeck) for valuable discussions of neuroanatomical details. The groundwork for this article was partly funded by a fellowship of Deutsche Forschungsgemeinschaft (DFG) and by the German Ministry of Science and Technology (BMFT) under contract ITW 9106. The authors also thank the anonymous reviewers for valuable discussions and hints. 1. Niggemann J, Beaumont I, Brückner R, Paul V. Anatom-Tutor—das wissensbasierte Ausbildungssystem zum Einsatz im Studentenunterricht Annals of Anatomy 1994;176(Suppl):247. 2. Martin R, Dubach J, Bowden D. NeuroNames: human/macaque neuroanatomical nomenclature Fourteenth Annual Symposium on Computer Applications in Medical Care. Los Alamitos, CA: IEEE Computer Society Press; 1990. pp. 1018-1019. 3. Braininfo [homepage on the internet]Washington: Neuroscience Division, National Primate Research Center, University of Washington; c2007 [updated 2007; cited 2007 Mar 20]http://www.braininfo.org 1990. accessed Mar 20, 2007. 4. Bowden D, Martin R. Neuronames Brain Hierarchy Neuroimage 1995;2(1):63-83. [PubMed] 5. Rosse C, Shapiro L, Brinkley J. The Digital Anatomist foundational model: Principles for defining and structuring its concept domainIn: Chute CG, editor. AMIA'98—Proceedings of the 1998 AMIA Annual Fall Symposium. . A Paradigm Shift in Health Care Information Systems: Clinical Infrastructures for the 21st Century. Orlando, FL, November 7–11, 1998. Philadelphia, PA: Hanley & Belfus; 1998. pp. 820-824. 6. Martin R, Mejino J, Bowden D, Brinkley J, Rosse C. Foundational model of neuroanatomy: Implications for the human brain project Proc AMIA Symp 2001:438-442. 7. Martin R, Rickard K, Mejino Jr JL, Agoncillo A, Brinkley J, Rosse C. The evolving neuroanatomical component of the Foundational Model of Anatomy AMIA Annual Symposium Proceedings 2003:927. 8. Rickard K, Mejino J, Martin R, Agoncillo A, Rosse C. Problems and solutions with integrating terminologies into evolving knowledge bases MEDINFO 2004: Proceedings of the 11th World Congress on Medical Informatics. Amsterdam: IOS Press; 2004. pp. 420-424. 9. Bota M, Dong H, Swanson LW. Brain Architecture Management System Neuroinformatics 2005;3(5):15-48. [PubMed] 10. Brainmaps [homepage on the internet]. Davis, California: University of California Regents Davis campus; c2005–2006 [updated 2007; cited 2007 Mar 20]http://brainmaps.org 2005. accessed Mar 20, 2007. 11. Kim S, Brinkley J, Rosse C. Profile of on-line anatomy information resources: Design and instructional implications Clinical Anatomy 2003;16(1):55-71. [PubMed] 12. Rydmark M, Kling-Petersen T, Pascher R, Philip F. 3D visualization and stereographic techniques for medical research and education Studies in health technology and informatics 2001;81:434-439. [PubMed] 13. Panchaphongsaphak B, Burgkart R, Riener R. BrainTrain: brain simulator for medical VR application Studies in health technology and informatics 2005;111:378-384. [PubMed] 14. Smith B, Ceusters W, Klagges B, Köhler J, Kumar A, Lomax J, et al. Relations in biomedical ontologies Genome Biology 2005;6(5):R46. [PubMed] 15. OBO Consortium [homepage on the internet]. The OBO Foundry; c2006 [updated 2006 April 24; cited 2007 Mar 20]. OBO Relations Ontologyhttp://www.obofoundry.org/ro 2005. accessed Mar 20, 2007. 16. Bittner T, Donnelly M, Smith B. Individuals, Universals, Collections: On the Foundational Relations of Ontology Formal Ontology in Information Systems. . Proceedings of the 3rd International Conference - FOIS 2004, Turin, Italy, November 4–6, 2004. Amsterdam etc: IOS Press; 2004. pp. 37-48. 17. Rector A, Rogers J, Bittner T. Granularity, scale and collectivity: When size does and does not matter J Biomed Inform 2006;39(3):333-349. [PubMed] 18. Bittner T, Donnelly M. A theory of granular parthood based on qualitative cardinality and size measures. In: Bennett B, Fellbaum C, editors. Proc Fourth Int Conf on Formal Ontology in Information Systems, FOIS06; 2006. 19. Johansson I, Smith B, Munn K, et al. Functional anatomy: A taxonomic proposal Acta Biotheoretica 2005;53(3):153-166. [PubMed] 20. Niggemann J. Analysis and representation of neuroanatomical knowledge Appl Artif Intell 1990;4:309-336. 21. Grenon P, Smith B, Goldberg L. Biodynamic ontology: applying BFO in the biomedical domain Studies in health technology and informatics 2004;102:20-38. [PubMed] 22. Basic Formal Ontology [homepage on the internet]. Saarbruecken, Germany: IFOMIS Institute for Formal Ontology and Medical Information Science, Saarland University; c2006 [updated 2006 Sep 28; cited 2007 Mar 20]. Basic Formal Ontology (BFO)http://www.ifomis.org/bfo/ 2004. accessed Mar 20, 2007. 23. Simons P. Parts: A Study in OntologyOxford: Clarendon Press; 1987.
|
