Anatomy Atlases
Cornelius Rosse M.D., D.Sc.,
Professor,
University of Washington
In the sciences basic to medicine, the anatomy atlas is a unique vehicle for representing knowledge in that it is a compendium of images largely devoid of narrative text. An invitation to review three recently published anatomy atlases (included in this issue of the Journal) has induced me to reflect on this phenomenon and several questions came to my mind. In trying to answer these questions, I hope to provide some rationale on which anatomy teachers (myself included) may base our choices when we select the knowledge sources for the courses we teach. I hope, furthermore, that these comments may be of some use to those who are preparing future editions of hard-copy atlases, as well as those who are developing computer-based atlases of anatomy.
What is an anatomy atlas?
The term atlas has a number of meanings. In the current context, however, it refers to a collection of graphical representations of parts of the human body, appropriately annotated and assembled according to some specifiable system. These graphical representations may be in hard-copy, consisting of illustrations prepared by artists through a variety techniques including photography; they may also be rendered by computer graphics, with or without an equivalent hard-copy representation. Since atlases are a mainstay for learning anatomy, there must be a component of anatomical knowledge that is communicated particularly well by an atlas, whether its contents are made available bound between two covers or via an electronic medium.
What component of anatomical knowledge is communicated by an atlas?
As a scientific discipline, anatomy is distinguished from the other biomedical sciences by its primary concern with the physical characteristics of material objects that constitute the human body. These material objects, ranging in size from macromolecules to organs and body parts, qualify as anatomical structures because they have been generated as a result of the coordinated expression of groups of genes (Rosse et al., 1998). Actually, all of the basic and clinical biomedical sciences, not just anatomy, are concerned with these anatomical structures; the non-anatomical sciences, however, focus on the processes in which anatomical structures participate, rather than on the structures themselves. Since understanding of physiological and pathological processes requires familiarity with the anatomical structures that participate in them, students and trainees in most educational programs learn anatomy a step ahead of other subjects. Moreover, diagnostic and therapeutic procedures (ranging from the medical history and physical exam to the interpretation of medical images and the performance of various invasive procedures) also rely on an understanding of anatomy, and therefore training in anatomy, as a rule, precedes clinical training as well.
Learning anatomy entails generating a mental picture - in other words, a spatial understanding - of anatomical structures. Elements of spatial anatomical knowledge include a comprehension of the shape of anatomical structures, their topological features (such as their apices, margins, internal and external surfaces and the modulations of those surfaces), their subdivisions and parts, and also their spatial relationships such as their orientation, location and spatial adjacency (Neal et al., 1998). Although this complex spatial information can be communicated in words, it is captured more effectively in images, which seem to be more readily imprinted in memory than narrative text descriptions. Images, therefore, have long been indispensable components of anatomy texts. Atlases go one step further: they rely almost exclusively on images for communicating the spatial elements of anatomical knowledge.
In disciplines that do not emphasize spatial knowledge of parts of the body, a textbook alone suffices to support a course of study or serve for later reference. In anatomy, an atlas not only plays an integral part in the acquisition of basic understanding: it is also consulted, perhaps more often than a textbook, by clinical specialists (e.g., surgeons, radiologists, radiation therapists) when they need to verify anatomical information relevant to a clinical problem. In fact, atlases have been specifically developed for those clinical specialties that are mainly concerned with the visualization and manipulation of anatomical structures and spaces. Such atlases include, for example, those for estimating bone age based on radiographs of the hand (Greulich and Pyle, 1959), planning surgical procedures (e.g., Dillard and Miller, 1983), applying new techniques for visualizing anatomy (e.g., Fleckenstein et al., 1996), and correlating anatomy with new imaging methods (e.g., Ledley et al., 1977; Cahill and Orland, 1984). As a rule, such special purpose atlases present images of normal and abnormal anatomy at a level of detail appropriate for the clinical specialist rather than the professional student. Nevertheless, atlases developed for the anatomy-intensive clinical specialties confirm the key role played by graphical representations for capturing and communicating spatial knowledge. Therefore we may conclude, first of all, that spatial knowledge is the essence of anatomical information, because it distinguishes anatomy from other biomedical sciences, and secondly, that organized sets of anatomical images have proven to be effective vehicles for representing this knowledge element. In fact, we have come to take these atlases for granted, and can hardly imagine teaching or learning anatomy without them. Can the history of anatomical illustration offer any insights as to how atlases have assumed such an apparently indispensable role for representing and communicating anatomical knowledge?
What is the history of anatomy atlases?
Although anatomical illustrations have existed for millennia, the anatomy atlas as a free-standing educational resource is only a little over a hundred years old. The use of an organized set of anatomical images for representing anatomical knowledge, however, can be traced back 450 years. It was Vesalius' De Humani Corporis Fabrica published in 1543 (Cushing, 1943; Saunders and O'Malley, 1950), which first combined a systematic and comprehensive graphical representation of the human body with narrative text . An earlier attempt undertaken with a similar objective by Leonardo da Vinci, in collaboration with the anatomist Marcantonio della Torre, was abandoned, like so many of the artist's grandiose projects, long before its completion. Until Fabrica, anatomical knowledge was communicated through text, essentially devoid of illustrations, and was dominated by the "infallible" Galen. The revolutionary step taken by Vesalius was to communicate a coherent body of knowledge through graphical representations of anatomical structures as they could be observed in a systematically dissected human body. Vesalius' decision to take this approach is the more remarkable because his training in anatomy was rooted in the Galenical mentality. He was even persuaded to participate - with some reluctance - in the perpetuation of Galen's doctrines. While he was working on Fabrica, he contributed to the first Latin translation of Galen's works from the Greek (and in doing so, he made a number of corrections, with much trepidation out of fear of offending his own teachers). Omnia Opera Galeni was published a year before Fabrica and was maintained through successive editions, with Vesalius' chapters included, up to 1625. Until Fabrica appeared, anatomical illustrations had been distributed as so called "fugitive sheets", containing one or two illustrations for guiding such procedures as venepuncture and minor surgery performed by barber surgeons and bath attendants. Vesalius himself produced such "fugitive sheets", some of which survive as the Tabulae Sex (six plates).
Much of the text of De Humani Corporis Fabrica relates directly to the illustrations or plates; it describes the plates and their intended use. That it has become virtually inconceivable to describe anatomy purely by text, devoid of figures, is largely attributable to the influence exerted by Fabrica. With the benefit of 450 years of hindsight, we may, therefore, credit Vesalius with demonstrating the effectiveness of illustrations for representing anatomical knowledge and thereby raising awareness of the importance of the spatial components of this knowledge.
Although this awareness would not be articulated explicitly for a long time to come, it gained expression through the collections and museums of anatomical specimens that were assembled during the age of Enlightenment and through much of the nineteenth century, as well as through the spectacular volumes of anatomical illustrations that are associated with leading anatomists of this period throughout Europe. Several are remembered today by the anatomical structures that still bear their names, but few of us have browsed through the rare book collections of our university libraries and marvelled at the beauty of the "tomes" they have authored. The creation of these books was as much motivated by the desire to leave a legacy as by the artistic opportunities offered by new graphical media, such as the engraved copper plate and lithography. Today it is hard to imagine these books (some of them justifiably claiming to be atlases) in the hands of students; indeed, they were probably used primarily by scholars.
The origin of the anatomy atlas as we know it today is associated with the establishment and differentiation of the "natural sciences" during the latter half of the 19th century, when the universities of France, Austria and Germany, dominated by powerful personalities, vied with each other to assume the central position in medical science. Anatomy institutes were established at one university after another. Their status enhanced by discoveries in histology and embryology, they claimed responsibility for the teaching of anatomy, which became an independent subject in the curriculum. Anatomical collections were energetically boosted, but the specimens in these museums could not meet the needs of the growing number of students, for whom a substantial part of the curriculum was devoted to the study of anatomy and the systematic dissection of the body. Although the textbooks written by the great anatomists of the period (Jacob Henle, Henry Gray, Joseph Hyrtle, Jean Cruveilhier) contained many illustrations (sometimes drawn by their authors), several atlases also began to appear. Filled with comprehensive and realistic graphical representations, they were intended to supplement the illustrated texts and to assist students in the dissection of the cadaver. The atlases by Carl Toldt, Johannes Sobotta and Werner Spalteholz are representative of this first generation of anatomical knowledge sources specifically intended for the anatomy curriculum. They soon appeared in English language editions and made their way across the Atlantic, where they were instrumental in exerting a predominantly German influence on academic medicine as it was evolving in America. The Toldt atlas was particularly popular here in the early part of the century.
Even though all these atlases claimed to support dissection, they were organized strictly according to organ systems, a separate volume of each atlas being usually devoted to one or more related systems. Of these early German classics, only Sobotta's atlas is current in this country (Staubesand, 1990); selected plates from it were rearranged by Professor Carmine Clemente in 1975 according to body parts, in order to support a regional approach to dissection (Clemente, 1975). Such a deliberately regional approach was pioneered by J.C. Boileau Grant, whose atlas first appeared in 1943 (Grant, 1943); to my knowledge, it was the first anatomy atlas that had its origin in North America.
Despite the almost universal streamlining of anatomy instruction in North America in recent times, a change resulting from reorganizations of the preclinical curriculum, a number of new anatomy atlases have appeared during the latter part of the 20th century, and joined updated editions of the established classics. Among the challenges that face the authors of today's atlases are the problems of defining the overall organization and the scope of such works. Given the great variety of anatomical and clinical images and the diversity of graphical techniques for rendering them, perhaps the greatest of these challenges is the distinction that should be made between illustrations that are best suited for a textbook and those that should be assembled in an atlas, particularly if these two knowledge sources are to complement one another.
Should there be a difference between textbook and atlas images?
The first American edition of Gray's Anatomy, corresponding to the eighth British edition (Gray, 1878), contained 522 illustrations ("engravings on wood"); the current 38th edition (Williams, 1995) has over 2,800 figures which make use of all kinds of graphical methods, ranging from some of the original engravings to computer-rendered graphics. However, it is not only such systemically organized reference books that are extensively illustrated. Textbooks specifically developed for today's anatomy courses also contain many figures. [Two examples: both Moore (1992) and Rosse and Gaddum-Rosse (1997) contains over 700 figures.] Yet the courses that rely on such books almost universally require an anatomy atlas as well. The respective authors usually declare that neither the textbook nor the atlas is intended as a self-sufficient resource for learning. The question arises, therefore, as to whether there should be a distinction between the kinds of graphical representations of anatomy that are designed for a textbook and those intended for an atlas.
Opinions regarding this question are likely to vary. Indeed, some textbooks borrow most of their illustrations from an atlas (e.g., Moore, 1992), suggesting that distinctions need not be made. In answering this question, one should perhaps consider the educational intent of these two kinds of sources of anatomical knowledge if they are indeed to complement one another, as their authors apparently have intended.
Whether textbooks of anatomy take a systemic or a regional approach, they describe distinct anatomical entities (organs, their parts, spaces, compartments, organ systems, tissues systems and their respective subdivisions, etc.). Some provide embryological explanations, functional rationalizations and clinical examples to assist the student in conceptualizing an entity or a related group of entities. Because of the very nature of anatomy, this cognitive task is hard to perform without a mental image of the entities in question. The textbook should, therefore, provide the template for such a mental image. This means that textbook illustrations should narrowly pertain to the entity which is being described, they should be simple, schematic rather than realistic, and readily "imprintable" on the student's memory. As the student builds a graphical mental model during the learning process, the student should be able to sketch these illustrations from memory (however inept he or she is at drawing). A simple and standard palette of contrasting colors should be used to aid distinctions between the components or subdivisions of the entities.
Images are indispensable for illustrating many of the developmental, functional and clinical applications of anatomy on which a textbook relies; the narrative component of the textbook is the ideal vehicle for presenting such images of applied anatomy in the appropriate educational context. Therefore, a textbook can readily accommodate a variety of illustrations without appearing fragmented; its text can "glue" the diverse images of applied anatomy to the template images of the anatomical structures themselves. Textbook images are, therefore, critical for providing the foundation for thinking about, or reasoning with, anatomical entities.
The atlas should enhance these mental constructs by representing reality, which includes spatial and structural complexity. This does not mean that, in learning a new structure, the student might not look at it first in an atlas. In such a case, however, the student should find, on turning to the textbook, that the simple image of the same structure communicates the structure's defining anatomical characteristics. It should make the realistic atlas image more comprehensible and, by merging it to a template, more meaningful. If, on the other hand, learning starts with the textbook, the atlas image should help the student to reinforce, modify and augment the mental template he or she has already constructed from the textbook. The nature of the atlas images should, therefore, be dictated by the kinds of enhancements the atlas should offer.
This means that the first requirement for an atlas, as for a textbook, is to have an explicit educational agenda - one which must, in this case, be accomplished entirely through images. In today's academic environment, merely supporting dissection does not suffice. The atlas should be a source of learning, complementary to the textbook. By its very nature, its greatest potential is to promote spatial understanding of anatomy.
Since the purpose of anatomy courses is to promote the learning of anatomy, rather than that of clinical medicine, the only justifiable role of clinical examples in an anatomy textbook is to enhance the understanding of dynamic, spatial anatomy. In order to accommodate such applied anatomy, however, descriptions of spatial relations in narrative text have, of late, become increasingly shortchanged. The graphical content of the atlas, therefore, assumes particular importance: it should explicitly provide for gaining a spatial understanding of anatomy. The composition of individual atlas images, as well as of single and facing pages, should be planned with a deliberate educational strategy. Whether the graphical representations are generated by medical artists, photography or medical imaging, clarity and sufficient contrast between different structures in the images must be a key requirement. Experience in teaching is needed, coupled with aesthetic judgement, to hit the appropriate balance between selectivity, desirable repetition and the density of the information that can be absorbed from one image. Disorienting and distracting the student must be avoided. These objectives require that the author carefully consider the educational intent of each image and each page. What is the student supposed to learn from it?
We may, therefore, conclude that both the textbook and the atlas must promote spatial knowledge of anatomy, but textbook figures should narrowly target anatomical entities through graphical means and simplify - in other words, edit - true spatial complexity. The objective of atlas images, on the other hand, should be to capture this complexity true to reality and employ judiciously selected graphical and artistic means for editing and clarifying it.
What criteria should be used for selecting an atlas?
It is fair to say that there have never been so many anatomy atlases and textbooks to choose from as there are now. Teachers of anatomy are faced with the dilemma of deciding as to which of these hard-copy sources should provide the bases for their courses, and they often delegate this responsibility to the students themselves by recommending several texts and atlases to choose from. A teacher cannot delegate such a responsibility to the student, however, if the teacher is motivated by a well formulated educational philosophy and if learning in the course is to be guided by a set of explicitly stated objectives. If that is indeed the case, the selection of an atlas as a resource for a course must be based upon an examination of the educational objectives of the atlas and a determination as to what extent these objectives synergize with the educational approach of the course as a whole. The same pertains to the textbook; indeed, the textbook and the atlas cannot be decided on independently, if they are to complement one another.
In synergy with one another, textbook, atlas and hands-on experience with exploring the dead and living body (through dissection and living anatomy exercises, respectively), should assist the student in generating a personalized mental model of the human body, one which can support anatomical reasoning. Anatomical reasoning (a requirement for clinical reasoning) is the cognitive process that relates manifestations of normal and abnormal function to anatomical entities and seeks to explain these manifestations in terms of the attributes of different anatomical structures (Rosse, 1995). Thus, successful completion of a course of study in anatomy should change the way a student thinks, which is a qualitatively superior educational accomplishment than the number of anatomical or clinical facts he or she knows.
If the objective is to promote cognitive skills rather than factual knowledge, then selection of the textbook must take precedence. Does the textbook teach, or does it only present anatomical facts? In other words, does the textbook provide a coherent body of knowledge which has an identifiable logical foundation? Can the student deduce from the book why anatomy is the way it is, and why structures and spaces are named the way they are? Will the student be able to think spatially in terms of the graphical representations provided by the text? Will the text and its figures facilitate the generation of mental images of discrete anatomical entities?
If such a resource can be found, or a study guide (syllabus) can be provided that meets these requirements, then the choice of the atlas must be dictated by the quality and comprehensiveness of the spatial information it presents. Will it help the student to visualize what the textbook or study guide intends to teach? Will it enhance that visualization and make it more real? The learning process is greatly enhanced by reiteration of the same information presented in different and interesting ways, so that familiar elements can be recognized and used as springboards to further insights. Can such synergism be perceived between the textbook and atlas?
If the course relies on dissection, does the atlas prepare the student for extending the mental visualization of anatomy to a hands-on exercise? The precious resource of the cadaver would realize much greater benefits for the student if he or she commenced each exercise with a reasonable understanding of the descriptive anatomy of the part to be dissected. Therefore, it would be a measure of the educational value of an atlas if the teacher could formulate explicit objectives for what the student should learn from given pages of an atlas, in preparation for a dissection.
More specific criteria must also be considered, but should not overshadow the general educational objectives. Will the atlas serve as an ideal, which the student should aspire to achieve with his or her cadaver? Does the atlas show those anatomical structures and spaces, and their relationships to one another, that are hard for a student to dissect, yet it is important to understand? If the selected textbook is deficient in explanatory sketches and clinical images, will the atlas compensate? Are such ancillary images in the atlas anchored to the main body of images in a complementary rather than a distracting way? Is the presentation of the image-based information consistent and logical? Are successive stages of display (dissection) presented logically so that the images actually teach? Will continuities be evident to the student? Are there different views, perspectives and contexts provided for important structures and clusters of structures to enhance spatial understanding of anatomy?
Finally, is the atlas beautiful? Is it a pleasure to handle? Does it promote in the student an admiration for the complexity and beauty of the human body? Will the student enjoy learning from it and be proud to own it? This enjoyment and pride must derive not only from the physical presentation of the atlas, but also from the educational value the student can obtain from it. The educational value of the atlas will depend on the specified role the course of study assigns to it in relation to other knowledge sources.
What is the future of hard-copy anatomy atlases?
If anything, the number of hard-copy anatomy atlases has increased on the market, rather than diminished, since computer-based atlases have made their appearance. Anatomy teachers, however, are beginning to structure their courses around these electronic resources. It will be interesting to know to what extent, and to what purpose, hard-copy atlases continue to be used in parallel with electronic resources.
Many of the computer-based anatomy programs rely on graphical representations of the human body that have been (or can be) rendered in hard copy, and several of the atlases are made available in parallel, through both media. Interactivity with these images is different in the electronic medium than in a bound atlas; on the other hand, hard-copy atlases are more portable and, until substantial changes are implemented in the cadaver laboratory environment, they are a more convenient knowledge source at the dissecting table.
The potential and promise of computer-based representations of spatial knowledge of anatomy, however, lie in those aspects of rendering that cannot be duplicated in hard-copy. These aspects of representation include three-dimensional computer graphics models of anatomical structures and the kinds of interactivity which allows spatial manipulation of the rendered objects. The application of computer graphics for displaying anatomical structures in 3D for supporting anatomy education was largely pioneered in the 1980s (Brinkley et al., 1989; 1997), and has provided a motivation for the Visible Human project of the National Library of Medicine (Spitzer et al., 1996). The latter project has spurred on an unprecedented level of activity in anatomical imaging, which will undoubtedly exert a tangible effect on the ways students can gain insights into the three-dimensional structure of the human body.
These developments in anatomical knowledge representation may be regarded as the most recent step in the continuous striving toward more effective methods for communicating a component of biomedical knowledge that is unique to anatomy. Computer-based approaches in knowledge representation offer new and distinct challenges for both the spatial and symbolic domains of anatomical information, as well as for linking these two domains to one another (Brinkley and Rosse 1998). It is as yet too early to tell the extent to which these new methods will replace or synergize with hard-copy resources of anatomical knowledge. As in the case of traditional anatomy atlases, in the last analysis, the viability and usefulness of the computer-based resources will be judged by their effectiveness for promoting the cognitive skills required for anatomical reasoning. The requirements for realizing such an objective are complex, and their discussion is better delegated to a separate communication
References
Brinkley, J.F., J.S. Prothero, J.W. Prothero and C. Rosse 1989 A framework for the design of knowledgebased systems in structural biology. In Proceedings of the 13th Annual Symposium on Computer Applications in Medical Care (SCAMC 89). L.C. Kingsland (ed.). Los Alamitos (CA): IEEE Computer Society Press, pp. 6165.
Brinkley, J.F., S.W. Bradley, J.W. Sundsten and C. Rosse 1997 The Digital Anatomist information system and its use in the generation and delivery of Webbased anatomy atlases. Computers Biomed. Res. 30:472503.
Brinkley, J.F. and C. Rosse 1998 Requirements for an online knowledgebased anatomy information system. J. Am. Med. Inform. Assoc. In press.
Cahill, D.R. and M.J. Orland 1984 Atlas of Human CrossSectional Anatomy. Philadelphia: Lea & Febiger, 139 p.
Clemente, C.D. 1975 Anatomy: a Regional Atlas of the Human Body. Philadelphia: Lea & Febiger.
Cushing, H. 1943 A Biobibliography of Andreas Vesalius. New York: Schuman's, 229 p.
Dillard, D.H. and D.W. Miller 1983 Atlas of Cardiac Surgery. New York: Macmillan, 258 p.
Fleckenstein, J.L., J.V. Crues III and C.D. Reimers 1996 Muscle Imaging in Health and Disease. New York: Springer-Verlag, 461 p.
Grant, J.C.B. 1943 An Atlas of Anatomy. Baltimore: Williams & Wilkins.
Gray, H. 1878 Anatomy, descriptive and surgical. Philadelphia: Henry c. Lea, 983 p.
Greulich, W.W. and S.I. Pyle 1959 Radiographic Atlas of Skeletal Development of the Hand and Wrist. 2nd ed. Stanford: Stanford University Press.
Ledley, R.S., H.K. Huang and J.C. Mazziotta 1977 CrossSectional Anatomy An Atlas for Computerized Tomography. Baltimore: Williams & Wilkins, 330 p.
Moore, K.L. 1992 Clinically Oriented Anatomy. 3rd Ed. Baltimore: Williams & Wilkins, 917 p.
Neal, P.J., L.G. Shapiro and C. Rosse 1998 The Digital Anatomist Structural Abstraction: a scheme for the spatial description of anatomical entities. J. Am. Med. Inform. Assoc. In press.
Rosse, C. 1995 The potential of computerized representations of anatomy in the training of health care providers. Acad. Med. 70:499505.
Rosse, C. and P. GaddumRosse 1997 Hollinshead's Textbook of Anatomy. 5th ed. Philadelphia: Lippincott Raven, 902 p.
Rosse, C., J.L. Mejino, B.R. Modayur, R. Jakobovits, K.P. Hinshaw and J.F. Brinkley 1998 Motivation and organizational principles for anatomical knowledge representation: The Digital Anatomist Symbolic Knowledge Base. J. Am. Med. Inform. Assoc. 5:1740
Saunders, J.B. deC. M. and C.D. O'Malley 1950 The Illustrations from the Works of Andreas Vesalius of Brussels. Cleveland: The World Publishing Company, 252 p.
Spitzer, V., M.J.Ackerman, A.L. Scherzinger and D.Whitlock 1996 The Visible Human male: a technical report. J. Am. Med. Inform. Assoc. 3:11830.
Staubesand, J. (ed.) 1990 Sobotta Atlas of Human Anatomy. 11th English Ed. Baltimore: Urban & Schwarzenberg, Volume 1, 406 p.; Volume 2, 384 p.
Williams, P.L. (ed.) 1995 Gray's Anatomy: the Anatomical Basis of Medicine and Surgery. 38th Ed. New York: Churchill Livingstone, 2092 p