Semiotica 144 (2003), 19-32.
In the reading of radiographic images, verisimilitude ('having the appearance of truth') is a major obstacle to the valid interpretation of Roentgen signs. If the truth value of a triadic sign is taken to be the truth value of its interpretant, then a verisimilar sign is a false sign that has the appearance of being true. The use of verisimilar Roentgen signs in diagnostic arguments (cf. Liszka 1996:57-58, on the anatomy of argument) results in diagnostic error. With the use of false positive signs, a normal patient is 'given' disease by the interpreter. With false negative signs, disease is 'missed'. In this paper, we develop a classification of verisimilar Roentgen signs of fracture. The sources of verisimilitude in these signs will be traced to their triadic structure. We will then examine the semiotic logic used by radiologists to distinguish between verisimilar and true Roentgen signs of fracture.
In a previous study (Cantor 2000) it was shown that conventional radiographic signs may be thought of as interpretants of triadic Roentgen signs. Such interpretants refer to real events in a patient and are independent of their representation. Therefore, in Peircean terminology, Roentgen signs are dicisigns (Peirce 1903a: 275). Since interpretants of dicisigns are statements about real events, they are either true or false. The most common predicates used in such statements are 'normal' and 'abnormal'. Therefore, these interpretants and their corresponding signs may be classified as true normal, true abnormal, false normal and false abnormal. We will say that one Roentgen sign has the appearance of another if the representamen of the first has the appearance of the representamen of the other. Hence, the representamen of a verisimilar sign has the appearance of the representamen of a true Roentgen sign. A false abnormal sign that has the appearance of a true abnormal sign will be called a simulator. A false normal sign that has the appearance of a true normal sign may be called a dissimulator. Hence, a simulator is a verisimilar sign of abnormality and a dissimulator is a verisimilar sign of the normal.
We will first review the elements of Roentgen semiotic grammar as applied to conventional radiography (Cantor 2002). The representamen of a Roentgen sign is a radiographic image i.e. a two-dimensional distribution of brightness intensity. The image ground of a Roentgen sign is a radiograph (film). The object of a roentgen sign is an anatomic event and its object ground is the body of the patient in which the event occurs. The interpretant of a Roentgen sign is an interpretation of the image that represents the object-event and its interpretant ground is comprised of the visual sense and mind of the interpreter. This semiotic grammar may be augmented to include the concept of the ambience of a ground. The ambience of a ground is its contextual complement. Hence, the ambience of the image ground consists of the components of the imaging system other than the film i.e. the radiation source (x-ray tube), devices for image enhancement (fluorescent screens) and the screen-film container (cassette) (cf. Dance 1993: 21,46). The ambience of the object ground is the world outside of the patient that contains the imaging system and any material objects interposed between the radiation source and receptor (cf. Dance 1993:21). The ambience of the interpretant ground consists of all events that may influence the mind of the interpreter. These include the sensory world and other minds. Therefore, each correlate of a Roentgen sign has its own ground and each such ground has its own ambience. In what follows, these concepts will be used to classify the sources of verisimilitude of Roentgen signs of fracture.
Before we can know the verisimilar signs of fracture we must know the corresponding true signs of fracture. This will require a brief review of relevant normal skeletal anatomy and basic concepts of fracture.
The word 'bone' refers both to a material and a structure (Currey 1984: 98, 133). A bone, considered as a structure, may be thought of as a shell of bone material (cortical bone) that is internally buttressed by a porous structure composed of the same material (spongiosa or trabecular bone). The confluent spaces in the spongiosa are filled with soft tissues. The longitudinal growth of a tubular bone occurs at a cartilaginous physis (growth plate). The osseous skeleton is a segmented structure composed of skeletal units (bones) that are held together by sheets and bands of fibrous tissue (joint capsules and ligaments). Motor units (muscles) are attached to the skeletal units by cords of fibrous tissue (tendons). Collectively, these fibrous links may be referred to as the fibrous skeleton.
If the osseous skeleton is the object of a Roentgen sign, then the fibrous skeleton is in the object ground of this sign. A point of attachment of the fibrous skeleton to the osseous skeleton is called an enthesis. In some individuals, bone may form within the fibrous skeleton at entheses. Such ossifications are called enthesophytes (Niepel and Sit'aj 1979). Futhermore, ossifications that project from joint margins in osteoarthrosis are called osteophytes. Both enthesophytes and osteophytes are commonly referred to as degenerative changes. Later, we will see that some fracture simulators are produced by such degenerative changes.
Radiographic skeletal anatomy
The radiographic image of a bone consists of a continuous bright boundary line (formed by the cortex) that surrounds a gray region (formed by the intra-osseous soft tissues) containing a pattern of fine bright lines (formed by the spongiosa). Enthesophytes and osteophytes project outside of the expected contour of the image of a bone. The physis of a tubular bone presents as a transverse gray line.
Concepts of fracture
Mechanical failure of a material or a structure is an abrupt change of state resulting from a failure to resist imposed stresses. In clinical medicine, mechanical failure of bone may present as either brittle or ductile fracture. In one type of brittle fracture, there is a loss of structural continuity as paired surfaces are formed within a bone. This is referred to as cleavage. (cf. Broek 1986: 33, 47; Burstein and Wright 1994: 177-183).
Tensile stresses applied to entheses may result in avulsion fractures. Indentation (cf. Lawn 1995: 249) is a form of brittle fracture produced by focal compression of a cortical surface. Compression of an articular surface may produce focal depression of the articular cortex or trabecular compaction in the underlying spongiosa. Axial compression of a vertebra may produce a decrease in vertebral body height with a concomitant increase in vertebral body width (a burst deformity or Poisson effect [cf. Burstein and Wright 1994: 111-112]). Oblique-axial compression of a vertebral body produces a tilted superior endplate (a wedge deformity). In ductile fracture, bone undergoes a smooth shape change (cf. Broek 1986: 47-57; Burstein and Wright 1994: 121-123, 180). This occurs most commonly in the tubular bones of children (cf. Ogden 1990: 55-56). Such fractures present as bowing (Euler buckling [cf. Currey 1984; 122-126]) or a torus deformity (local buckling [cf. Timoshenko and Gere 1961: 457-461]).
Fatigue failure refers to crack initiation and growth due to repeated low intensity loading over an extended period of time (cf. Broek 1986: 57-68; Burstein and Wright 1994: 124-126). In its earliest stages, a fatigue crack of bone is microscopic and not detectable by radiography. We are now in a position to describe the typical radiographic signs of mechanical and fatigue failure.
Legisigns of fracture
All radiographic signs of mechanical failure are dependent to some extent upon relative displacement of fragments. Intrinsic radiographic signs of brittle fracture are discontinuities within the image of a bone. Discontinuities of the cortex due to brittle fracture include a gap, a step and a bifurcation. Cortical indentation produces a notch. A discontinuity in the spongiosa due to cleavage presents as a dark line. Depression of articular cortex or trabecular compaction may present as a bright line (or stripe). Shear stress may produce a trabecular fault. Axial compression of a vertebral body produces a burst deformity while oblique-axial compression produces a wedge deformity. Extrinsic radiographic signs of brittle fracture present as fragments displaced outside of the expected contour of a bone. This type of discontinuity will be called a detachment. Ductile fracture presents as a smooth contour deformity. In tubular bones, signs of ductile fracture include bowing and torus deformities.
In Peircean terminology, the above sign types are legisigns and will be referred to as the 'line legisign', 'the step lesigign', 'the burst legisign' etc. Corresponding instantiations of these sign types will referred to as a 'line sinsign', a 'step sinsign', a 'burst sinsign' etc. It must be emphasized that in clinical practice the legisigns of fracture may occur singly or in various combinations. In contrast with mechanical failure, fatigue failure in its earliest stages presents with a normal radiographic image. Hence, the earliest Roentgen sign of fatigue failure is a dissimulator. In later stages of fatigue failure, signs of healing such as a dark line or a bright stripe are interpreted as delayed signs of fracture, which are true abnormal signs.
In this section, simulators of fracture legisigns will be classified on the basis of their triadic structure. Sinsigns of each type of simulator will be cited. For convenience of reference, most of these sinsigns may be found in the comprehensive atlas of Roentgen simulators by Keats (1996). For economy of space, references to specific figures in this atlas will be given by the notation K chapter-figure. When there are numerous sinsigns of one sign type, we will give no more than three examples chosen from different anatomic sites.
Line legisigns of fracture
Firstness: The basis for verisimilitude in Firstness is artifactual. A film crease is a verisimilar sinsign of the line legisign that presents as a dark line in the image of a bone. The source of verisimilitude of a film crease is in the representamen ground i.e. the film.
Secondness: The sources of verisimilitude in Secondness may be in developmental, degenerative or projectional variation. Developmental skeletal variations such as multiplicity and/or incomplete fusion of ossification centers (primary or secondary) and pseudoepiphyses (cf. Williams and Warwick 1980:241-243, 417-418) may simulate cleavage fracture i.e. may present as dark lines within the image of a bone. The source of verisimilitude of such sinsigns is in the sign object. Dark line sinsigns for primary ossification centers include persistent or anomalous cranial sutures (mendosal suture K1-178), persistent acetabular Y-cartilage and bipartite tarsal bones (talus K7-487). Dark line sinsigns for secondary ossification centers include bipartite epiphyses of tubular bones (distal radius K6-149, lower tibia K7-326, proximal phalanx big toe K7-632) and incompletely fused epiphyses or apophyses (distal radius lappet K6-133, lumbar apophysis K3-387, acetabular rim apophysis K4-79). Dark line sinsigns from pseudoepiphyses occur in short tubular bones of the hands and feet (first metacarpal K6-274, proximal phalanx big toe K7-643). Dark line sinsigns are also produced by bipartite sesamoid bones (patella K7-213, os peroneum K7-406, hallux sesamoid K7-617). Bright line simulators of depressed articular cortex and trabecular compaction include prominent physeal scars and growth arrest lines in the long tubular bones (cf. Ogden 1990: 80-81). Discontinuous entheseal ossifications may also present as dark lines. These are nondevelopmental or acquired skeletal variations and their source of verisimilitude is the sign object. Such simulators usually present as clefts within paravertebral entheseal ossifications. Projectional effects that present as dark lines are produced by the radiographic superposition of certain extra-osseous anatomic structures and bone. Such structures include a skin fold (K7-56), a fat stripe (K5-72) and a gas filled space (K2-77). The sources of versimilitude of such projectional effects are in the object ground (the patient) and in the ambience of the object ground (the relative position of the radiation source and the patient).
Thirdness: The bases for verisimilitude in Thirdness are perceptual and cognitive illusion. A common perceptual illusion that occurs in clinical radiography is the Mach effect resulting from an abrupt change in brightness at an edge produced by the superposition of radiopaque structures. A negative Mach effect originates in the retina of the observer/interpreter and is perceived as a dark line. Hence, the source of verisimilitude of a Mach effect is in the interpretant ground. Sinsigns of the line legisign that are based on the Mach effect may be produced by superposition of the atlas on the odontoid process in the cervical spine (K3-116), superposition of acetabular osteophytes on the femoral neck (K7-53) or superposition of the fibula on the tibia (K7-332). Certain normal structures such as cranial sutures and physes in the immature skeleton, vascular grooves on flat bones and intraosseous canals present as dark lines on radiographs. For example, the temporal artery groove (K1-237), the canal of the middle supraclavicular nerve (K5-100) and the canal of the tibial nutrient artery (K7-302) simulate the line legisign of fracture. The verisimilitude of these sinsigns is a consequence of nonrecognition of the normal. Hence, its source is in the interpretant ground i.e. the mind of the interpreter.
It must be emphasized that the recognition of any simulator is also dependent upon the training of the interpreter, which is in the ambience of the interpretant ground.
Detachment legisigns of fracture
Firstness: A dust particle trapped in a cassette-screen or screen-film interface produces a bright speck in a radiographic image that may simulate a detached fragment. The source of verisimilitude of this artifact is in the ambience of the representamen ground.
Secondness: Developmental variations that simulate detachments include accessory ossification centers, accessory skeletal elements and incompletely fused apophyses. Accessory ossification centers may occur between cranial sutures (sutural bones K1-186), in epiphyses (ulnar styloid K6-154) or metaphyses (lower fibula K7-325) and in the patella (K7-201). Numerous accessory skeletal elements (ossicles) occur in the wrist, ankle and midfoot (cf. Kohler and Zimmer 1993: 79-82, 778-781). Nonfused apophyses that simulate avulsions occur at vertebral endplates (K3-196), at the end of vertebral processes (K3-290) and at the base of the fifth metatarsal (K7-557). Degenerative variations such as incompletely ossified enthesophytes or osteophytes that simulate detachments commonly occur outside of vertebral endplates (K3-209) and near apophyses of the appendicular skeleton. The source of verisimilitude of detachment sinsigns produced by accessory ossification centers and incompletely fused apophyses is in the object. The source of verisimilitude of accessory ossicles and entheseal ossifications is in the object ground.
Thirdness: The cognitive basis for misinterpretation of simulators of detachment is a lack of knowledge of the normal anatomy of the immature skeleton, the normal accessory ossicles and the range of degenerative variation in the fibrous skeleton. In these cases, the source of verisimilitude is in the interpretant ground.
Deformation legisigns of fracture.
In a sinsign of a deformation legisign of fracture, the contour of the image of a bone differs from what is expected i.e. the normal. The deformation legisigns of fracture include the step, angulation, notch, burst, wedge, bowing and bifurcation deformities.
Firstness: Since radiographic imaging systems are designed to provide images with recognizable contours (i.e. faithful representations), the source of verisimilitude in deformation sinsigns on standard images cannot be in Firstness.
Secondness: Developmental variation due to discordant ossification or growth rates may produce step (bathrocephaly K1-162, lower end fibula K7-319, upper end radius K6-102), angulation (odontoid process K3-104, xyphoid process K5-155), burst (atlas of cervical spine K3-66, vertebral body K3-411) or wedge (cervical vertebral body K3-193, thoracolumbar vertebral body K3-408) deformities. Entheseal ossifications (femoral neck K7-49) and projection (sacrum K3-459, spiral groove of scaphoid K6-238) may also produce simulators of the step legisign of fracture. The bowing legisign of ductile fracture may be produced by projectional exaggeration of the normal curvature of a tubular bone and an apparent cortical bifurcation is a projectional effect that sometimes occurs in a metatarsal neck. Hence, the sources of verisimilitude for simulators of deformation may be in the object, the object ground or the ambience of the object ground.
Thirdness: The cognitive basis for the misinterpretation of deformation simulators is nonrecognition of a normal radiographic contour. Legisigns of fracture deformation that may be simulated by normal radiographic contours include the notch (cervical apophyseal joint surface K3-259, femoral condylar groove K7-164), bowing (fibular shaft K7-287, inferior vertebral endplate K3-422), and trabecular fault (glenoid neck K5-71). The source of verisimilitude for these simulators of fracture deformation is in the interpretant ground.
In this section we will describe two ways in which simulators are distinguished from true signs of fracture. These involve discrimination by exclusion and discrimination by opposition.
Discrimination by exclusion
Locative exclusion is a mental operation that is used to distinguish simulators from true signs of fracture. By locative exclusion, a sinsign of a given legisign of fracture is recognized as a simulator when its representamen refers to an object or condition outside of bone. Therefore, it cannot be a true sign of fracture. Such sinsigns may be called locative excluders.
Discrimination by opposition
The previously described legisigns of fracture were obtained from Roentgen sinsigns by the mental operation of hypostatic abstraction (cf. Peirce 1868:2, Liszka 1996: 12, 71). Hypostatic abstraction directs the attention of the interpreter to one type of representation while neglecting all others. Using this concept, we have identified a relation between legisigns that is used in Roentgen interpretation which we will call hypostatic equivalence. By hypostatic equivalence, we mean the recognition by the interpreter of the equivalence of two legisigns, one obtained by hypostatic abstraction from a perceived image (a sinsign) and the other recalled from visual memory. In clinical practice, the problem of verisimilitude arises when the interpreter believes that there is a relation of hypostatic equivalence between a legisign of a false abnormal sinsign and the memory of a true abnormal legisign. To avoid diagnostic error, the interpreter must be able to distinguish a perceived verisimilar sinsign (a simulator) from the memory of a true abnormal sinsign, both of which instantiate the same legisign. Such semiotic discrimination is based on the mental operation of prescissive abstraction which directs the attention of the interpreter to a single quality of a representation while neglecting all others (cf. Peirce 1903a: 270, Liszka 1996: 12, 71). In Roentgen semiotics, a quality of a representation is a visual feature of an image. Simple visual features that are distinguished by prescissive abstraction will be called distinctive features in analogy with phonic distinctive features (cf. Cantor 2000). A distinctive visual feature may be thought of as the representamen of a distinctive qualisign. The mental operation of prescissive opposition is commonly used to discriminate between simulators and true abnormal signs. By prescissive opposition we mean the mental operation that directs the attention of the interpreter to two distinctive features, one prescinded from a perceived image and the other from the visual memory of the interpreter. Prescissive opposition applied to a Roentgen sinsign forms a binary opposition of distinctive features. In turn, this induces a binary opposition of corresponding distinctive qualisigns. Hence, a sinsign is a simulator if prescissive opposition applied to the sinsign yields a perceived distinctive qualisign that is in opposition to the memory of a true distinctive qualisign. Such binary oppositions of distinctive qualisigns will be called prescissive discriminators.
There are two types of binary opposition between visual distinctive features which we will refer to as the specific and the general. This typology is based on the concept of markedness as defined by Roman Jakobson for use in linguistics (cf. Cantor 2000). A binary opposition may be thought of as an asymmetric binary relation between marked and unmarked terms. In Roentgen diagnosis, abnormality is marked and normality is unmarked. In the context of this paper, verisimilitude is marked and truth is unmarked. In the specific type of binary opposition, the marked term consists of more or less of a feature that is shared with the unmarked term. In the general type of binary opposition, the marked term consists of the presence or absence of a feature that is not shared with the unmarked term. Since binary oppositions of distinctive features induce binary oppositions of their corresponding distinctive qualisigns, there are specific and general types of prescissive discriminators. All of the above concepts will be illustrated by clinical examples in a later section.
The Roentgen diagnosis of fracture may be divided into three stages. Diagnosis begins with detection which is followed by localization and identification (cf. Cantor 2002). Before detection can occur, the mind of the observer must be prepared by the acquisition of a sense of the normal. This sense of normality is based on the Peircean concept of habit (cf. Peirce 1878: 129 and 1905: 336-337). A visual habit is an expectation that is acquired through training or experience. Fulfillment of such an expectation is experienced as an emotion of satisfaction. Habit formation is accompanied by the storage in memory of perceptual judgments (cf. Peirce 1903b: 155 and 1903c: 191) that are considered to be veridical by a consensus of the community of observers. Detection of a true or verisimilar sign of abnormality is a mental event that involves habit-breaking. The term habit-breaking refers to the mental event in which an expectation of normality is unfulfilled (in Peircean terminology, 'passive surprise' [Peirce 1906:384]). This experience is accompanied by an emotion of surprise that evokes the next stage in the diagnostic process. After a radiographic sign has been detected, its object must be localized in order to be identified. Localization is the recognition of a spatial relation between the object of a sinsign and a known anatomic structure e.g. the relation of being 'inside' or 'outside'. In the previous section, we have seen how the mental operation of locative exclusion may be used to distinguish between verisimilar and true signs of fracture. Only after the object of a sinsign is localized within a bone can it be identified as a fracture. Identification ultimately involves discrimination between verisimilar and true signs of fracture. In this case, discrimination is based on the mental operation of prescissive opposition which was described in the previous section.
In this section, we will show by clinical examples how verisimilar and true sinsigns of the same legisign are distinguished by exclusion and opposition. Examples of this type of differential diagnosis will be given for the line, detachment and deformation legisigns of fracture.
A line legisign may be dark or bright relative to the normal intra-osseous soft tissue gray level.
Dark line legisigns. A film crease that extends outside of the image of a bone is immediately distinguished from a fracture by locative exclusion. In contrast, a film crease contained within the image of a bone may be distinguished from a fracture by prescissive opposition. The discriminators employed in this operation are curvature (a crease is usually curved and a fracture straight) and edge density gradient (a crease usually has unsharp margins while a fracture has sharp margins). The opposition curved/straight is of the general type and the opposition unsharp/sharp is of the specific type. Occasionally, a film crease does not exhibit appreciable curvature or unsharpness and the differential diagnosis is indeterminate. Incomplete fusion of accessory ossification centers produces dark line sinsigns that may simulate fracture. In epiphyses and apophyses and in the skull, an accessory synchondrosis or suture produced by incomplete fusion of an accessory ossification center is distinguished from a fracture by its characteristic location and by distinctive features such as rounded (as opposed to sharp) or sclerotic (as opposed to non -sclerotic) bone edges and smooth (as opposed to pointed) corners. The opposition rounded/sharp is of the specific type while the oppositions sclerotic/non-sclerotic and smooth/pointed are of the general type. Incomplete fusion of canonical ossification centers and sesamoid partitions may be distinguished from fractures by the same binary oppositions. Pseudoepiphyses are distinguished from fractures by their characteristic location in the non-epiphyseal end of metapodial bones and their circumferential (as opposed to full-thickness extent) in addition to the previously mentioned edge discriminators. Linear discontinuities of entheseal ossification are distinguished from fracture by their rounded (as opposed to sharp) edges with smooth (as opposed to pointed) corners. Linear projectional artifacts that extend outside of the image of a bone are immediately distinguished from fractures by locative exclusion. Of the linear projectional artifacts that lie entirely within the image of a bone, some may be distinguished by their characteristic location and others by contour and edge discriminators. Occasionally, it may be difficult to distinguish projectional simulators by prescissive opposition and additional imaging may be required. Frequently, a dark line confined within the image of a bone is an optical illusion created by the Mach effect. In such cases, the extraosseous origin of the dark line is recognized when it conforms to the edge of a superimposed radiopaque structure. In such cases, discrimination is by locative exclusion. Vascular grooves or canals are distinguished from fractures primarily by their characteristic locations and predictable orientations. However, prescissive opposition yields other discriminators including optical density (vascular grooves and canals are less dark than fracture lines) and edge density gradient (vascular grooves and canals have rounded as opposed to sharp edges). Note that the optical density discriminator is of the specific type.
Bright line legisigns. In some individuals, a dense physeal scar presents as a transverse bright line or stripe simulating a trabecular compaction fracture or healing fatigue fracture, particularly in the lower end of the tibia. Such bright lines are distinguished from fracture by their characteristic location and their edge density gradient (sharp as opposed to unsharp).
An artifact produced by a dust particle may be distinguished from a cortical avulsion by locative exclusion if the simulator occurs in a non-entheseal location. Discrimination by prescissive opposition is based on optical density (such artifacts are brighter than bone) and shape (particles are irregular rather than flat). Accessory ossification centers, ossicles and incompletely fused apophyses are distinguished from detached fragments both by their characteristic locations and by prescissive opposition. The most useful discriminator that distinguishes between a developmental ossification and a detached fragment is cortical continuity (a continuous as opposed to interrupted cortical boundary). The cortical continuity discriminator is of the general type. Incompletely ossified enthesophytes may be distinguished from detached fragments by prescissive opposition using the edge density gradient (rounded as opposed to sharp edges) discriminator.
A developmental step at a cranial suture or a physis is distinguished from a displaced fracture by its rounded (as opposed to sharp) and sclerotic (as opposed to nonsclerotic) edges. A step on the femoral neck produced by entheseal ossification is distinguished from a displaced fracture by cortical continuity (as opposed to discontinuity) in the underlying bone. The developmental Poisson deformity is distinguished from fracture by the absence (as opposed to presence) of either a step or bifurcation in the vertebral cortex. The thoracolumbar vertebral wedge deformity is distinguished from fracture by its characteristic location in the spine, tilted (as opposed to non-tilted) inferior endplate and the absence of a cortical step.
In this study, we have examined the phenomenon of verisimilitude as it occurs in the Roentgen diagnosis of fracture, which is a common problem in clinical radiology. We first presented a phaneroscopic classification of true Roentgen signs of fracture. We then used the Peircean Categories as a basis for a comprehensive classification of verisimilar Roentgen signs of fracture. We have seen that discrimination between simulators and true signs of fracture is based on mental operations of exclusion and opposition. It remains to be seen whether these mental operations form a basis for all Roentgen differential diagnosis.