-Throughout the development of the Foundational Model of Anatomy (FMA) ontology, one of the use cases put forth has been anatomy education. In this work, we examine which types of knowledge taught to anatomy students can be supported by the FMA knowledge base. We first categorize types of anatomical knowledge, then express these patterns in the form “Given ____, state ____”. Each of the 33 patterns was evaluated for whether this type of knowledge is compatible with the modeling and scope of the FMA.
Knowledge of human anatomy is fundamental to the fields of health sciences. Software applications that support the delivery of healthcare services and training of healthcare providers often incorporate anatomical knowledge, but rarely in ways that are computable and reusable. As researchers seek to make software systems more intelligent, opportunities to draw upon knowledge bases of anatomy will increase. As part of this research agenda it is important to examine whether the needs of specific applications can be supported by available knowledge bases.
This paper categorizes the types of knowledge relevant to student learning in university-level courses in human anatomy, and then examines which types are supported by the Foundational Model of Anatomy (FMA) ontology.
The FMA is both a theory for representing anatomy and a
computational artifact [
This work was supported in part by the National Library of Medicine of the National Institutes of Health under award R21-LM012075. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
The process by which health science students learn anatomy
has traditionally consisted of a combination of cadaveric
dissection, two-dimensional illustrations or photographs, and
text-based descriptions of anatomical relationships. Like most
areas of modern life, computer-based tools have increasingly
been integrated into anatomy education. These include
computer-based interactive atlases, such as the Visible Human
Project, as well as virtual anatomic models that allow students
to rotate and visualize structures and relationships in three
dimensions. These types of computer-based 3D visualizations
can successfully augment more traditional methods of
instruction, resulting in improved understanding and retention of
anatomic knowledge [
If educational applications for learning anatomy make use of common knowledge bases—instead of relying on applicationspecific catalogues of knowledge—benefits will include greater standardization of terminologies, less duplication of effort in constructing knowledge artifacts, and easier implementation of reasoning capabilities. This paper revisits the potential for the FMA to serve as a knowledge base for education in gross human anatomy, three decades after its conception.
To capture types of knowledge relevant to learning human
anatomy in university-level courses, a variety of educational
resources were reviewed. Particular emphasis was given to
structured information presented as tables in atlases and review
guides [
Consider these examples:
• A table describing lymph node groups that provides
information about location, afferent lymphatic structures,
efferent lymphatic structures, and regions of the body
drained (from [
During the process of identifying units of information and developing categories, a category was created if two or more examples of a pattern of knowledge were found within the sampled content. These categories were expressed as assessment questions using patterns in the form “Given ____, state ____” in order to make explicit the prompt and the knowledge to be recalled. An example of a pattern is “Given a structure, state its parts”.
Each category of assessment questions was compared with the modeling and scope of the FMA. Effort was directed toward determining whether the type of knowledge in the assessment questions could be retrieved from the FMA, not determining whether the FMA currently contains the content necessary to answer specific questions.
Five broad categories of anatomical knowledge were identified (see Table 1). Questions were organized into 17 subcategories and expressed through a total of 28 patterns. Table 1 also provides examples of specific questions for each pattern and an assessment of whether the FMA could serve as a source for each type of knowledge.
This analysis shows that the FMA is well-suited to representing knowledge about synonyms of terms, classification of anatomical structures, parts of structures, and connectivity between structures. As expected, the FMA is not a suitable knowledge base for questions about the qualities of anatomical structures (such as morphology or variation within the population).
For the types of knowledge that the FMA can support, several factors may complicate efforts to directly apply the FMA to educational applications:
Part relationships within the FMA tend to be much more granular than those taught in anatomy courses. For example, Figure 1 (top) shows a question about a part of the urinary bladder. In the FMA, this information traverses three part relationships.
Implications: This high level of granularity in the FMA is appropriate for advanced anatomy courses, but may not be a good fit for learners in basic anatomy courses. But just as advanced learners should be able to understand and reason over high-granularity representations to answer low-granularity questions, it is possible that some types of high-granularity representations in the FMA can be converted to low-granularity representations.
Educational materials may focus on general concepts (“ventral and dorsal roots merge to form a spinal nerve”), while the FMA tends to represent knowledge with greater specificity (such as specific ventral and dorsal roots).
Implications: The class hierarchy may provide an avenue for representing knowledge applied to many individual structures. (For example, “Muscle organ” has regional part “Distal tendon”.) However, because properties of class are inherited by all its subclasses, there is a danger that a general anatomical principle will not be true for every subclass.
Educational materials often make use of assumptions and unwritten knowledge. Making this knowledge explicit, as required by the FMA, can introduce an expected level of complexity. As show in Figure 1 (bottom), answering a question about the passage of air through the nose and into the pharynx using the FMA requires that the nasal cavity is explicitly recognized as a part of the nose. A medical student has tacit knowledge that movement of air through the respiratory system (at the level of gross anatomy) takes place within tubes and cavities, and would immedicately recognize that this question refers to air-filled spaces—even if he or she did not conceive of “nasal cavity” as an anatomical structure.
Implications: Directly translating the FMA content into educational contexts is largely inappropriate because it risks directing students’ attention toward modeling details of the FMA, rather than on building upon their existing understanding of anatomy. However, it may be appropriate to use explicit FMA representations as a supplement to less-detailed representations as a way to help students construct and deepen their knowledge of anatomy.
As noted in previous work to test the FMA against anatomy
examination questions [
Implications: The precision of relationships used in the FMA may be helpful in stimulating students to think more deeply
CATEGORIES OF ASSESSMENT QUESTIONS FOR ANATOMICAL KNOWLEDGE Category of assessment question Category 1: Representations and real anatomy Understanding visual and semantic representations and their relation to real anatomy 1a. Cadaver “pin test” Given a structure marked within a cadaver, state the corresponding anatomical term. 1b. Translating between visual representations and verbal representations Given a visual representation of a structure, state the corresponding anatomical term. [And reverse: Given an anatomical term, identify the structure in a visual representation.] 1c. The language of anatomy Given a directional term, state the definition. [And reverse.] • Superficial: toward the surface • Distal: away from the center Given a plane, state the definition. [And reverse.] • Median: separates right lateral and left lateral regions at midline • Transverse: separates superior and inferior regions Given an anatomical root word, state the definition. [And reverse.] • Brachial: of the arm • Orbital: of the eye Given a structure, state a synonym.
• Pharyngotympanic tube: Eustachian tube • Nostril: naris Suitable for FMA No No No No No Yes, synonyms are provided.
Category 2: Classification Understanding how categories are used to describe anatomy, as well as characteristics of members of categories 2a. General vs. specific Given a specific structure, state the type of structure to which it belongs.
• Elbow joint: synovial joint • Frontal bone: flat bone • Lateral meniscus: cartilage Given a type of structure and a defining characteristic, state the specific structure.
• Nerve that innervates the foot and leg: sciatic nerve • Fluid in the lymphatic system: lymph • Joint that is the largest and most complex in the body: knee joint 2b. Cardinality Given a type of structure, state how many are present in the (canonical) body.
• Permanent teeth: 32 • Major calyces per kidney: 2–3 • Layers of meninges surrounding brain and spinal cord: 3 Category 3: Canonical structure Understanding the location, composition, and demarcation of structures 3a. Whole-part relationships Given a structure, state its parts.
• Mandible: left ramus, right ramus, body of mandible • Lymph node: cortex, medulla • Cortex of lymph node: superficial cortex, paracortex Given a region of a structure, state the indicated part of that structure.
• Lowest portion of the brainstem: medulla oblongata • Triangular divisions of the medulla of the kidney: renal pyramids Yes. Available in the class hierarchy No, unless encoded through class hierarchy or other relationships.
No, although some information may be implied through the class hierarchy.
Yes. Available in regional and constitutional part hierarchies.
No, although some information may be available in definitions. Given a structure, state the types of tissues that compose it.
• Skin: epidermis, dermis • Nasal cartilage: hyaline cartilage 3b. Regional location of structure Given type of structure and region of the body, state the specific structures of that region. [And reverse.] • Muscles of the neck: longus capitis, longus colli, rectus capitis anterior, … • Foramen of the skull: right/left mental foramen, right/left infraorbital foramen, … • Lymph node groups of head and neck: submental, submandibular, occipital, … 3c. Spatial relationships among structures Given a structure and a spatial relation, state the associated structure(s).
• Spinal cord passes through: foramen magnum • Femoral triangle contains: femoral vessels, femoral nerve, lymph nodes • Subarachnoid space contains: cerebrospinal fluid • Femoral artery bisects: femoral triangle • Serous pericardium surrounds: heart • Annular ligament surrounds: radial head • Deltopectoral triangle has superior boundary: deltoid • Ribcage is superficial to: lungs Given two structures, state the structure positioned between them. [And reverse.] • Between the visceral and parietal layers of the peritoneum: peritoneal cavity • Between the lungs, immediately anterior to the heart: thymus • Dividing the right and left sides of the nasal cavity: nasal septum Given a structure (artery, vein, or nerve), state the structures it encounters along its course.
• Internal iliac artery: passes over pelvic brim and descends into pelvic cavity 3d. Connectivity between structures Given a structure and type of connectivity, state the associated structure(s).
• Scapula articulates with: clavicle, humerus • Via the coronal suture, the frontal bone articulates with: right/left parietal bones • Carpometacarpal joint of thumb connects: trapezium and metacarpal of thumb • Anconeus has origin: lateral epicondyle • Anconeus has insertion: lateral side of olecranon, upper ulna • Anconeus has innervation: radial nerve • Right subclavian trunk drains into: right lymphatic duct • Occipital artery has origin (or source): external carotid Given two or more structures, state the structure they join or merge to form.
• Ventral and dorsal roots merge to form: spinal nerves Given a structure, state the two or more structures it branches, bifurcates, or divides into.
• After exiting the vertebral column, each spinal nerve divides into: dorsal ramus, ventral ramus, meningeal branch, communicating rami • Trachea bifurcates into: right and left main bronchi 3e. Clinical regions and landmarks (points, lines, borders) Given a region or structure, state the associated clinical regions.
• Abdomen: epigastrium, umbilical region, suprapubic region, right and left lumbar regions … Given a structure, state the associated landmarks. [And reverse.] • Points of the skull: right and left euryon, right and left coronale, right and left auriculare, … • T2 (second thoracic vertebra): superior border of scapula 3f. Morphology Given a structure, describe its form.
• Duodenum: c-shaped part of the small intestine • Vertebral foramen of cervical vertebra: triangular space • Mandibular alveoli: sockets (for teeth) Yes. Available in the constitutional part hierarchies.
Yes, if a region has been represented. For example, classes such as “Musculature of hand” have members that are individual muscles.
Most, using relationships such as surrounds, lateral to, contains.
No No Yes, using relationships such as articulates with, has origin, has insertion, drains into.
An alternative modeling scheme using branches and tributaries (as regional parts) is employed.
An alternative modeling scheme using branches and tributaries (as regional parts) is employed.
Yes, if modeled as regional parts.
Some. For example, the class hierarchy contains subclasses of “Anatomical point of skull”. Other landmarks may be captured using the scheme 3D structures are bounded by 2D surfaces, bounded by 1D lines, bounded by 0D points.
No, unless available in definition. Category 4: Variation in structure Understanding variations in human anatomy 4a. Sexual dimorphism Given a structure, describe the morphological differences between female and male structures. • Sacrum: female sacra tend to be wider, shorter, and less curved than male sacra • Pelvic inlet: circular in females, heart-shaped in males 4b. Anatomical variation Given a structure, describe common variations.
• Branches from the aortic arch: in the most common variant, the left common carotid artery arises from the brachiocephalic trunk (instead of the aortic arch itself) • Sternalis: a muscle parallel to margin of the sternum, present in less than 10% of population Category 5: Developmental anatomy Understanding structural changes during gestation and early childhood 5a. Development of structures Given a structure, state the structure(s) it develops into or becomes part of. [And reverse.] • Urogenital ridge: pronephros, mesonephros, metanephros • Neural tube: brain, spinal cord 5b. Germ layer origins Given a structure, state the germ layer it developed from.
• Kidney: intermediate mesoderm • Epithelium of gastrointestinal track: endoderm 5c. Developmental homologues in males and females Given a (male/female) structure, state the developmentally homologous (female/male) structure. • Ovary: testis • Cowper’s gland: Bartholin’s gland • Three primary brain vesicles: 4th week • Implantation: about 7 days 5d. Timing of developmental events Given a structure, state the stage (or time interval) at which it is present. [And reverse.] No No created.
No Some. The relationships derives, matures into, and transforms into have been used in recent work.
Some. The relationship germ origin has been Some. The relationship developmental stage of has been used in recent work about anatomical relationships, but may not be directly relevant to the needs of students in basic anatomy courses.
This work helps to make explicit ways in which the FMA knowledge base could (and could not) support learning within a university-level anatomy course. The work will assist developers of educational applications in identifying types of anatomical knowledge, as well as recognizing opportunities for making use of a knowledge base such as the FMA.
Many thanks to James Brinkley and José (Onard) Mejino of the Structural Informatics Group at the University of Washington for supporting M.D.C.’s earlier work with the FMA. We also thank Kate Mulligan for discussions concerning anatomy education.