Although its origins can be traced back many centuries, biostereo-metrics as a focus of scientific attention is less than 20 years old. It is only 15 years since the first biostereometrics laboratory was established at Baylor College of Medicine in Houston. When SPIE sponsored its first biomedical symposium on quantitative imagery in the biomedical sciences in 1971, there was a mere handful of stereo-metrical ly oriented papers. In 1982, the 3rd International Symposium on Biostereometrics held in San Diego attracted over 70 presentations from 21 countries - a very rapid growth.

A method is described for ordering the fringes in an ordinary moird topogram. It requires no a priori shape information and resolves 'hill-or-valley' ambiguities using local surface reflectance properties. The theoretical and practical constraints of this method are discussed, and analysed examples are presented as experimental evidence.

Three-dimensionnal measurements on human body using a scanning laser beam. The active optical apparatus principle and the image data processing giving three dimensional informations of complex forms is presented. The output is given in terms of one or several files of real coordinates. The basic components of this system are : A light-sheet which is generated by a laser source and, the optical sensors (cameras) with corresponding hard-soft extractor; This 3D sensorial system is especially adapted to partial or total acquisitions of body coordinates. The main advantages are : - Vision and measurement capability of complete accessible contours without shadow areas. - The real time data acquisition and scanning of the object in a few seconds. - The access to distance measurements between significant points. - The presently obtained accuracy is better than 1/1000 in relatives units and lower than one MM absolute. - Physically stuck markers on body are not necessary. - The monochromaticity of the laser light source allows the use of a color filter over the detector (camera) for ambient light rejection. - The fully programmable capability for any use, allows the adaptation to a large variety of particular cases. - The hardware open system offers many options. - The hard-soft tool is designed for auto-calibration operation. - The system offers easy connection to a host computer or a production robot.

A three dimensional television system (VICON) is described. This device is connected to a minicomputer and is used to achieve a biostereometric study of human movements. Three synchronised video cameras enable to capture up to fourty trajectories of reflective markers fixed on anatomical landmarks and illuminated by infrared stroboscopes at fifty fields per second. 3-D trajectories are computed automatically. Displacements, velocities, accelerations and angles data are used to modelize movements.

Two instruments are described which are used for three dimensional measurement of stationary objects. Available computer software for the equipment is discussed. It is designed to run on an IBM Personal Computer and allows the user to specify his own measuring requirements from a library of standard routines. Applications to research in dentistry, medicine and anthropology are also discussed.

A recent need for cranofacial data collection, intended for the design of safety masks, led the authors to conceive a solution using conventional stereophotogrammetry and mirrors. The detailed implementation of the necessary equipment is reviewed here and the determined function and precision characteristics are also presented.

A simple, low cost, and universal automatic 3-D acquisition system was developped using active triangulation. Its main characteristics are: - its speed of acquisition (about 1 minute per object), - the density of its sampling (from 30,000 to 250,000 measures per object), - the structured lattice of the resulting data (distributed on parallel and meridian equidistant lines). From these data, it is possible to use several techniques of surface approximation to compress the information: some of them are presented here, with their respective performances (quality, compression rate, complexity of programming). They are divided into two main classes: - planar approximations, - spline function approximations. We also present several advantages of the resulting data base which is structured like a picture data base. We can take advantage of this structure to develop very fast processing techniques.

A new development is described which allows the Reflex Microscope to be adapted as a stereo plotter for measuring from pairs of photographs. A software suite is being implemented on a microcomputer.

A three-dimensional camera unit has been developed at the Dental School of the University of Zurich for jaw motion recording. Since the system is designed to detect and reconstruct the spatial position of a light source, it can easily be used to digitize an object surface being scanned by a laser beam. This is performed by an additional device which has been developed in order to enlarge the application field of the camera unit as well as to prove the feasibility and find out the limits of such a simple and low-cost quantification system.

Classical nonholographic 3D imagery methods in light microscopy have inherent limitations and drawbacks which are discussed. We propose techniques which we believe are more efficient. They are based on the properties of color film, lenticular screens, and simple optical devices.

A confocal scanning laser microscope (CSLM) has been constructed at the EMBL1 and is applied in the investigation of fixed and of living specimens. Contrary to the conventional microscope the CSLM focuses the light emitted by a point source into an object and detects only the light emitted from a small volume in this object. When operated in either fluorescence or reflection mode the CSLM has a true resolution along the optical axis. This feature is used to record independent images at different levels of height in a fluorescently labelled object or to scan the object in a plane parallel to the optical axis.

3-D dynamic deformation of the trachea during forced expiration is indirectly known through the flow measure. A better knowledge of this deformation would be of great interest for the understanding of the trachea behaviour in such conditions, but an actual in vivo measure is rather difficult. Consequently, a simulated forced expiration experiment is conducted using a silicone trachea model and an isolated human trachea subjected to controlled lowerings of pressure. In a first step, a grating projection technique is used to measure the 3-D contours of these models. A detailed description of the optical device is given. The optical and numerical calibrations of the recording system are then amply described. In a second step, the model outlines are recorded. The grid (picture 1) is projected on to the studied object (picture 2). These two resulting pictures are then used to reconstruct the trachea in a quadrant with a suitable numerical stereographic method. In a third step, two projectors coupled to two cameras are used to allow the acquisition of the whole model surface. A series of distinct recordings are obtained for various pressure values inside the trachea. If the model has deformed, the deviation of each stripe from its previous position or the new object contours can be computed, leading to a quantitative volumic deformation measure. Finally, an actual 3-D deformation analysis of the phantom is forecast by the mean of real-time data acquisition with two TV cameras.

The monitoring of the motion of an arbitrary mandibular point is essential in dentistry in order to extract from typical motion patterns relevant parameters indicating normal or abnormal masticatory function. After first experiences of nonrestrictive jaw motion quantification with the help of a Selspot I we designed a new real-time optoelectronic system. The recording system includes three cameras. Each camera contains a cylindrical lens and a linear optosensor (CCD). Two extra-oral lightweight target frames (2g) containing each three light emitting diodes are fixed on the patients's teeth. The system fires each LED cyclically and computes their three-dimensional coordinates in a reconstruction unit. From these coordinates another processing unit computes the three-dimensional coordinates of an arbitrary jaw point in a head fixed coordinate system. The computed trajectories are drawn on-line with the help of a dedicated 3D viewing hardware on a CRT where they can be directly rotated and zoomed to follow intricate motion details. The system is easy to handle, low-cost, stand-alone, has an excellent linearity and a resolution better than 0.1 mm.

A real-time moire contouring system, which is insensitive to specular reflections, has enlarged depth of field and is used in conjunction with a micropositioning relocation device, has been built for the measurement of wear in dental restorations. Measurements are carried out on accurate polyvinyl siloxane replicas.

In X-ray medical imaging,the promotion of 3D display techniques imposed to improve the spatial and temporal resolution of 3D acquisition-reconstruction devices. One solution is to use a 2D X-ray detector. For one source position,the attenuation measurement of the diverging X-ray beams provides a radiography. The source trajectory defines a set of acquisitions. In terms of illposed problems, we state that a circular trajectory centred on the area to reconstruct can be a good one, provided the source remains far enough from the object. We propose a direct reconstruction algorithm based on an approximated evaluation of the 3D Radon Transform and on the use of its inversion formula. We present the numerical foundations of such a transform method.

A method for the transformation of irregularly distributed data points into a regular grid is presented. In this transformation local approximations of surface patches are used. In order to avoid undesired clipping of the surface area at the margins, a particular criterion discarding unreliable surface patches is applied. As an application, a transformation of rasterstereographic data from a back shape measurement is presented. Approximation theory shows that in the special application of back surface measurement a high accuracy can be acnieved.

The use of lenticular screens has allowed us to produce true 3D synthetic images. The latter, called integrams, provide direct tridimensional images of exceptional quality and allow the observation of hidden parts in a 20° field. This method seems to offer a cost-effective and simple alternative to holographic methods, and represents a promising field for future studies in image synthesis.

The diffraction pattern of a biological cell replica obtained by contrast phase method is Fourier analyzed. In former works, the authors have proposed a theoretical model for the optical transmission function of a biological cell and the analytical Fourier Transform associated to this model was obtained. In the present work both results, analytical and experi mental are compared.

Radiological assessment and follow-up control of scoliosis, i.e. of a lateral and rotational deviation of the spine, is performed mainly by single plane radiographs. Additional information may be gained from these radiographs by introducing a parametrized vertebral model. By analyzing the radiographic contours according to this model, axial rotation can be determined for any position and orientation of the vertebra. In addition to rotation several other data are determined for each vertebra, such as the tilting angle and the two-dimensional coordinates of the centre. By handling the data as a function of the vertebral location in spine, characteristic curves are generated. In order to find simple shape parameters for these characteristics, a smooth curve has to be fitted to the data points by a least squares approximation. This problem may be solved by a Fourier decomposition of the spinal curves. It appears, that the Fourier coefficients (amplitudes and phases) and some derived shape parameters lend themselves to a medical interpretation, which is consistent with the existing classification of the scoliotic spine.

The body shape of twelve boys of different ages and body types was modelled and growth followed over a period of eight years. The model consisted of sixteen segments and elliptical zones two centrimetres wide constructed from simultaneous front and side views of the subject. Computer graphics for the model were developed which allow for shaded colour surface representation, segment manipulation and image repositioning. Growth affected the segment volumes and produced intersegment differences in the rate of growth, expressed as a proportion of total volume. Head proportion changes were much less than for the other segments. There were differences between individuals in the growth patterns for each segment but the trends for each segment were distinct.

The mechanism by which the cardiac impulse is propagated in normal hearts from its origin in the sinus node to the atrio-ventricular node has not been agreed on fully. We studied the "internodal posterior tract" through the crista terminalis by light microscopy and stereological morphometry. The hearts of 12 Papio cynocephalus were perfused , after sacrifice,with phosphate-buffered formol saline. The regions of the crista terminalis (CT), interatrial septum (IAS), atrioventricular bundle (AVB) and interventricular septum (IVS) were cut off and embedded in paraplast and sectioned (10 4m). The multipurpose test system M 42 was superimposed over the photomicrographs (1,890 points test, ESR = 2%) to the stereological computing. The quantitative results show that the cells from CT were more closely relationed with IAS cells than others cells (IVS and AVB cells). This results are not a morphological evidence to establish the specificity of the "internodal posterior tract". The cellular arrangement and anatomical variation in CT myocardium is very important.

Desuis mon enfance je suis un type soidisant visuel. Cette deviation etait toujours profitable pour moi, pour le clinicien , ainsi que pour le rechercheur scientifique.I1 y a trente ans, que dans l'Atlas Radioanatomique du Thorax nous avons deja utilis6 la conception de trois dimensions. Dans les derni6res edition de cet ouvrage nous avons travaille mke avec la guatrieme, le temps facteur. DePuis cette 6poque-la je me suis ori-en-be vers les quatre dimensions, vers les mouvements dans l'espace et le temps. -:tudiant les mouvements,les mouvements ventilatoires mesure par photogrammetrie le tronc humain.En projetant sur la surface du tronc un reseau quadratique, on pent compter de 3oo-400 points d'information, si un carre est 2 This 2 cm. Simultanement on pent ainsi contraler les deplacement relatifs des differents points. Travaillant avec seriophotographie et ces dernieres annees avec Video, nous avons des documents cDntinuellement en trois dimensions sun la position actuelle en espace de notre modele.On pent mesurer, reproduire la position momentan6e du sujet d'une frequence et dans la quantite desiree. Par exemple on peut les reproduire, une a une, comme une sculpture, en domontrant les changements d'un tronc humain pendant les mouvements respiratoires, comme nous avions fait en 1969 avec le sculpteur Istvan Bencsik en realisant nos differents modeles en inspiration et expiration.flais thooriquement on pent construire tant de sculptures qu'on vent entre les deux positions extremes.La methode est tres utilisable pour l'observation des mouvements sportifs on artistiques, comme le ballet par exemple. / Figure 1./ En etudiant en plusieures series la morphometrie des mouvements respiratoires on pent acquerir une certain experience, qui concerne la phase de respiration, on se trouve un corps humain sculpte, dessine on peint. Come le mimique d'un visage explkue toujours l'6tat affectif d'un individu, le corps humain peut expliquer tart des chases entre les mains d'un artiste doue. Liais nous retournerons encore a cet argument. Avec un objet d'art l'artiste registe un seul moment dune serie de mouvements de son sujet, spit en deux dimensions, s'il est peintre on graveur , dessinateur etc., spit en trois dimensions s'il est sculpteur.i:algre cela , il y a beaucoup d'oeuvresd'art dyna-miques, qui donnent l'impression de mouvement.On pent produire cet effetavec la ROMDOSi-tion des mouvements non simultanees , comme par exemple sur les tableaux de XIX 'si8cle representant des concours hyppiques, on les extremit-ees des chevaux sort distandue an maximum en avant et en arriere; c'est un mode de galop inexistant, mais malgr6, qu'il est faux, it suggere la vitesse.

The field of biostereometrics provides us with three dimensional measurements (x,y,z coordinates) by utilizing stereopairs of either photographs or radiographs. The recent rapid development of sophisticated equipment and advanced computergraphics methods give an impetus for applications of biostereometrics to biomedical imaging. The combination of these two areas, that is, biostereometrics and advanced computergraphics, can be used to enhance our ability to make instantaneous (within 1/30 second), interactive surgical simulations and representations, and continuous, comparative, superimposed, quantitative displays. These procedures are applicable for serial observation and diagnosis and treatment planning for craniofacial anomalies, dermatologic lesions, gingival inflammatory processes and skeletal anthroplogy.

In order to be practically useful, gait analysis in patients with motor handicaps should provide information on joint loads, forces and moments. For accurate joint force estimates, mass and inertial properties of the limb segment must be known. A program for the determination of segmental mass and inertial properties was therefore set up, using stereophotogrammetry for the evaluation of segmental body volumes. The methodology using two stereo pairs of cameras is described. A 15 segment body model was defined with its segmental boundaries and its segmental anatomical axis systems.

The purpose of this work is the knowledge of the upper limb kinematics for various final tasks requiring effort or precision. The aim is to integrate these results in a C.A.D. system for human engineering studies. The description of these movements proceeds from the 3-D trajectories of anatomical landmarks delivered by the VICON system. Such a representation using fixed orthogonal reference system x, y, z, does not lead to a simple analysis of the gesture. So from these data we compute a set of angular parameters which are in closer relation with the real kinematics of the upper limb. We obtain this result by the introduction of pertinent intermediate reference systems, related to each rotation degree of freedom. The result exhibit typical patterns of the temporal evolution of the angular parameters related to the task assigned to the subject. The collected data constitute a computerized catalogue of movements included in ERGODATA system.

The ergonomic analysis of a control or a supervision workstation for a vehicle or a process, necessitates to take into account the biomecanical visuo-postural system. The measurements, which are necessary to do, must give informations about the spatial direction of the limbs, the dorsal shape, eventually the eyes direction, and the postural evolution during the working time. More, the smallness of the work station, the backrest and sometime a vibratory environment made use specific, strong and small devices wich do not disturb the operator. The measurement system which we propose is made of an optical device. This system is studied in relation with the french "Institute de Recherche pour les Transports" for an ergonomic analysis of a truck cabin. The optical device consists on placing on the body of the driver on particular places materializing specially members and trunck joint points, some drops which reflect the infra-red raies coming from a specific light. Several cameras whose relative positions depend on the experiment site, transmit video signals to the associated treatment systems which extract the coordinates (Xi, Yi) of each drop in the observation scope of any camera. By regrouping the informations obtained from every view, it is possible to obtain the spatial drop position and then to restore the individual's posture in three dimensions. Therefore, this device doesn't enable us, in consideration of the backrest, to analyse the dorsal posture, which is important with regard to dorsal pains frequency. For that reason, we complete the measurements by using a "curvometer". This device consists of a flexible stick fixed upon the individual back with elastic belts, whose distorsions (curvature in m-1) are measured, in the individual's sagittal plane, with 4 strain gauges pairs; located approximately at the level of vertebra D1, D6, D10 and L3. A fifth measurement, concerning the inclination (in degree) of the lower part of the stick, makes it is possible to represent at any time the dorsal shape in the space. Some examples of real postures, measured by these complementary electromecanic and video devices, will be presented.

The model described is based on a combination of simple mathematical concepts, experi-mental 3-D data and a set of intercorrelations. The basis of this model can be considered as an evolution from all the others derived from Squire's concepts. It allows to realize global volumes of the left and right reach envelopes from the modeled reach surfaces that it generates. These volumes are useful to study man-machine interactions using C.A.D. systems.

A close-range three-dimensional photogrammetric system developped for studying human knee bones is discussed. A stereophotographic facility of two non-metric cameras with syn-chronized shutters, a stereoplotter with digitizing capabilities and a micro-computer with appropriate software are capable of obtaining digital data all around the bones that constitute the human knee. Graphic representation of any part of the object by using digital information as well as perspective and stereoscopic representation of such desired part of the object is now routinely possible with an accuracy of less than ±0.3 mm.

Human knee joint movements during walking were analyzed in time and in the three dimensions of space by photogrammetry using highspeed cinecameras and direct linear transformation. Triangular models for the thigh and shank were used instead of the usual stick diagrams. The results are demonstrated in a normal subject and in a typical patient with a movement disorder (spastic diplegia). High accuracy of measurements was achieved. The positioning of the shank and thigh is better demonstrated by such modelling than by stop motion images of the life subject.

The ankle is a complex structure allowing foot mobility while providing stability. In an attempt to improve the knowledge of the kinematics of the ankle, an approach incorporating both experimental and analytical techniques was developed. Stereophotogrammetry combined with the Direct Linear Transformation (DLT) technique, was used to quantify the spatial displacements of the foot. Four motorized cameras were fixed on a baseboard 0.62 m from a support frame so as to obtain two stereopairs, one medial and one lateral. For a pair, the cameras were 0.52 m apart and maintained a convergent angle of 21.5°. The support frame was designed to fix the tibia while allowing foot motion. A device comprised of 76 markers, 38 of which were visible to each pair of cameras was used for the calibration. The spatial position of each marker was measured to a precision of 0.05 mm whereas their computed spatial position using the DLT technique was accurate to 0.4 mm. For the experiment, two embalmed cadaver legs and feet, amputated at midshank and of normal appearance were used. After a partial dissection, three pin markers were embedded into each of the medial and lateral sides of the talus permitting the calculation of its center of rotation. Each foot was photographed in 5 positions at 10° intervals, ranging from 30° of plantarflexion to 10° of dorsiflexion. An analytical model was developed to spatially describe the rotation of the foot about the ankle. The model calculates the plane of motion and the orientation of the axis of rotation relative to the sagittal, frontal and transverse planes. These were found respectively to be for foot one: 100°, 86°, 15° and for foot two: 91°, 69°, 21°.

The study of the human female breast for the presence of neoplastic lesions has been the major focus of our research. Close-range stereogrammetry has been used to produce digital as well as graphic representations (contour maps) of the breast under examination. Previous study of these maps has been used to evaluate single contours for the presence of aberrations (deviations from the convex hull) and to correlate these with clinically documented masses. The purpose of this paper is to report, in part, the development of a measurement strategy to quantify breast volume and volume distribution comparisons within our subject pool. More specifically, the data acquisition techniques and algorithms employed in defining the boundary between the breast and thorax of the subject are described. Special emphasis has been placed on the determination of validity and reliability of the boundary placement by the stereoplotter technicians. In addition, the application of this measurement process to documentation of the presence or absence of total volume and volume distribution differences existing in normal breasts and those containing benign or malignant lesions are discussed.

The feasibility of a system which is capable of automating production of posterior dental restorations has been investigated. Data acquisition can be accomplished in a clinical setting using standard equipment to obtain stereoscopic views of the prepared tooth, adjacent and opposing teeth, and the jaw in motion. Data from the patient is digitized with a very high resolution digitizer. Stereophotogrammetric reconstruction and kinematic analysis establish the three-dimensional envelope in which the restoration must function. Ideal tooth profiles have been digitized for each of the fourteen unique teeth. Each tooth profile is represented by B-spline or Bezier curves and surfaces. Utilizing local influence properties of these curves and surfaces, the ideal profile can be modified to provide proper function and fill the available space, yielding a morphologically correct restor-ation which meets the unique design requirements for a specific patient. Spline curves and surfaces also ensure that the surfaces are well behaved, without discontinuities. This property facilitates generation of tool paths for numerically controlled machining. The latest version of Control Data Corporation's ICEM-DDN package, with its interface with the newly developed graphics programming language (GPL) is used to create those tool paths. Special consideration is given to unique requirements of mating internal and external sur-faces as well as to high precision, small size manufacturing.

This paper gives a brief historical survey of the development of medical applications in photogrammetry. It discusses in detail the point reached by methods and instruments now available in photogrammetry, together with their advantages and drawbacks in medical applications. It concludes with suggestions for making better use of the potential offered by medical photogrammetry and for closer interdisciplinary collaborat ion between photogrammetry and medicine.

Glaucoma is a disease characterized by increased ocular pressure with optic nerve atrophy and loss of visual field, or side vision, which eventually leads to blindness. The optic disc becomes more cupped with increasing progress of the disease. We have devised a system for the use of the Donaldson simultaneous stereophotographic fundus camera to obtain optimal reproducibility and sensitivity of measurements from optic disc stereophotographs. Stereomodel deformations caused by eye-to-camera optical variables (photographic magnification, working distance, relative orientation between the camera optical axis and the optic disc, and position of the optic disc image in the film frame) were evaluated. We developed an analytical correction method for stereomodel deformations and evaluated the impact on measurement of optic disc cupping. We have evaluated the reproducibility of new techniques for measurement of cup parameters such as volume, depth, and area of the cup relative to the disc area; slope of the cup floor, wall, and rim; and cup shape using volume profiles in which contour areas were plotted of the cup depth from the top to the bottom.

Modeling and simulation methods of man-machine systems are developed at the laboratory by interactive infography and C.A.D. technics. In order to better apprehend the morphological variability of populations we have enriched the 3-D model with a parametric function using classical anthropometric dimensions. We have selected reference, associate and complementary dimensions : lengths, breadths, circumferences and depths, which depend on operator's tasks and characteristics of workplaces. All anthropometric values come from the International Data Bank of Human Biometry of ERGODATA System. The utilization of the parametric function brings a quick and accurate description of morphology for theoretic subjects and can be used in C.A.D. analysis.

The lecturer describes the methods and the means to design in relation with anthropometrics datas the driving cab of modern SNCF vehicles such as the SYBIC universal locomotive and the TGV-Atlantique power car vehicle is to operate in the end of the decade.

An intelligent system for automatic research of optimal postures in ergonomics has been developed at the Laboratoire d'Anthropologie Appliquee. A dynamic programming algorithm allows the automatic graphic research for an eight degrees of freedom bidimensional model. The different range of motions and acceptable ergonomic intervals are used for each articulation. This method is going to be integrated in the C.A.D. system of ERGODATA.

Computer Aided Design is able to perform workstation's conception. An ergonomic data could be complete this view and warrant a coherent fiability conception. Complexe form representation machines, anthropometric data and environment factors are allowed to perceive the limit points between humain and new technology situation. Work ability users, safety, confort and human efficiency could be also included. Such a programm with expert system integration will give a complete listing appreciation about workstation's conception.

An analysis of the external and internal curvatures of spine was carried out on a sample of nine males from biostereometric measurements for different imposed postures. The results concerning the modifications of the external shape of the curves are used for the 3-D human body modeling in C.A.D. applications.

The mechanisme of the breathing pump-system changes with age. Morphometrical informations are presented by means of Photogrammetry.The effectiveness of the diaphragm contraction in relation to the ribcage was observed with trunc-surface deforrnations.The thorax-wall of new-borns and sucklings are pliable:the diaphragm action sipps the ribcage. Growing childrens thorax can resist against this sipping effect, but practically they have no thoracal, only abdominal breathing.Later in childhood the thoracal breathing type developps individually. Introduction Since 1974 I have the honour to participate with my papers on several Biostereometrics Session. At that time I had 15 years of Photogrammetry study of the breathing movements behind me. Based on Photogrammetrical documentations I worked with manual geometry, constructing my designes,hoping, that I can elaborate a model-method and to grope around the photogrammetrical possibilities in the study of the mechanism of the human breathing pump-system. Using a ruler and a pair of compasses, with steady obstinacy I arrived to realise three abstractions:Spatial; spatiotemporal analysis informations and the velocity or the acceleration factors, this last one being the most sensitive. Recently I made the-first steps to study four dimensional information, with two time factors.

The objective and reproducible definition of a body-fixed coordinate system is of crucial importance for a meaningful interpretation of many biostereometric measurements. A method for defining such a coordinate system using anatomical landmarks is presented. On the back surface many landmarks are characterized by their peculiar curvature. Therefore, the landmarks may be objectively detected by an analysis of the distribution of back surface curvatures.

A method for measurement of left-right asymmetry of the back shape is presented. The procedure is based on an analysis of the surface curvature distribution. As a result, a "sym-metry line" (line of least asymmetry) on the back and an asymmetry function is obtained. According to the evaluation procedure the data are essentially independent of a reference coordinate system. The results are to be compared with radiographic findings of the scoli-otic spine.

Precise three dimensional measurements were made on a normal, dried adult vertebral column, using the recently developed reflex microscope. Preliminary conclusions are drawn from these measurements regarding the geometry of the lumbar and thoracic vertebrae. Subsequent measurements were made on a number of unrelated fifth lumbar and first sacral vertebrae and comparisons were made with the findings of the "master model".

The volume of the spine was measured in 9 embryos from 8 to 31 mm crown-rump length (complete series of sagittal sections). Spine morphometry was performed by planimetrical point counting of horizontal projections on a 5 mm square grid. Total spine volume was integrated by multiplication of the thickness by the area : V = t ET=1 Si. The integrated volumes (including the base of the skull around the foramen magnum) were aligned on the diagram semi-logarithmic volume v.s. linear crown-rump length. The correlation between the spine volume and the total weight of the spine is very high (r=0,94 ; p<0,01). The spine growth of the embryos during the post-somitic phase corresponds to the general laws for this period (particularly acceleration in the second half), one of the most interesting for morphometry during the uterine life. The vertebral morphology is perfecting between the two extremities of the observed period with a nearby adult disposition at 31 mm. Datation of the 9 embryos was determined from their C-R length by means of the table published by MOORE and al. (1981) from 641 staged embryos of the Carnegie Institute, thus establishing their stage with actual admitted criteria. There was no important variation of the unique spine curvature during the postsomitic phase. Linear measurements of maximal width on the entire embryos and their spine were determined from the total number of sections and those where the vertebrae could be observed. The two widths reported to crown-rump length, drew linear curvatures with a slight irregularity, emphasized by plotting the values of three indices. These variations could be dued to the imprecision in section thickness and to the individual variation during the spinal growth for this embryonic phase. In contrast with linear data, the morphometrical method applied to the spine volume has proved to be very effective in quantitative studies for embryos of the post-somitic phase.

A convolution-backprojection algorithm for fully 3-D reconstruction from cone beam projections was developed and implemented on an ADAC 3300 nuclear medicine computer system interfaced to a rotating scintillation camera with a pinhole collimator. As an approxima-tion, a space-invariant filter was applied to each projection independently before back-projection. The resolution of the system was found to be 7.8 mm at the center of field of view and 10.5 mm in the periphery, as opposed to 18 mm for a parallel hole, all purpose collimator. Results with phantom studies demonstrated that the technique provided images with improved spatial resolution.

We show the feasibility of an image-analysis and computer graphics procedure, allowing to visualize the cerebral blood vessels. The methods which have been used (photogrammetry and computer graphics) are described, illustrated on real cases and then discussed. The future ways of application of this technique in the medical field are presented.

Three-dimensional (3-D) reconstructions of bone and soft-tissue anatomy can be obtained by data processing of bidimensional images from Computed Tomography. These 3-D data, using C.A.D. technique may provide optimal information for the design of individual prosthesis, for the simulations of several procedures in pre-planning surgery and for the evaluation of complex anatomical relationships.

An algorithm is presented for surface recognition and display of anatomic humain structures obtained from CT data and focused to medical planning and treatment. The algorithm utilizes an efficient new method for the computation of surface normal vectors and a new gradient operator for the surface detection. Some examples of skull structures are given.

Functional neurosurgery such as stereotactic biopsies, interstitial irradiation, chronic depth electrodes, neuro-stimulator or drug injection, requires the implantation of cerebral probes through the labyrinth of arteries and veins. Current methods used to determine the zones free from blood vessels that can be retained as safe passages for the probe to avoid internal cerebral hemorrhages either lack precision or introduce certain limits. This paper presents a newly developed method aimed at facilitating dependable implantation of cerebral probes. It results from an inter-disciplinary research and is based on applying stereophotogrammetric techniques in radiography. An apparatus permitting stereo-vision of a floating line has been developed. It permits to simulate probe implantations. Tests carried out on an actual human cerebral blood vessels network, under severe experimental conditions, indicate that nearly 100% of the probes can be implanted with the help of the device in all directions without risks of hemorrhages.

A computer and graphics system for 3D surgical planning based on CT and MR scan images is presented. The system allows display of anatomy, production of patient anatomical replicas, and the design of custom implants. It also allows integration of 3D information across scan modalities for optimal use of scan data.

The spatial distribution of ultimate bone strength in trabecular bone at the proximal tibial epiphysis was calculated from x-ray computed tomography (CT) images. The distribution obtained was displayed in pseudo 3D plots generated by a standard computer graphics routine. Clinical examples in normal and malaligned knees exhibit remodelling of bone strength in accordance with Wolffsl theorem. It is found that pseudo 3D plots of bone strength patterns may yield valuable informations in clinical practice as well as in basic biomechanical research.

A sequence of two-dimensional image slices (such as obtained from computerized tomography) may be used to create three-dimensional visualizations of human anatomy. One limitation of three-dimensional representations from such data may be that the image resolution within each slice (typically 0.4 to 1.4 mm) is large com-pared to the size of features in the ana-tomy under study. Adjacent slices may be separated by even larger distances, typically from 1 to 10 mm. Many methods have been developed for interpolating between the coarsely sampled pixel values to generate a three-dimensional surface, with varying degrees of realism.

A new technique for making a three dimensional map of the optic nerve head is expected to have a major impact on the way in which glaucoma is diagnosed and treated. The new technique, contour photography, allows the health of the optic nerve head to be objectively evaluated every six months during the patient's routine office visit. In contour photography, a set of parallel lines of light are projected into the patient's eye and the back of the eye is photographed using a standard camera that is available in almost all ophthalmologist's offices. The three dimensional information is encoded in the positions of the photographed lines, and is decoded by treating each stripe as the intersection of a plane of light with the fundus. At present, a trained human observer identifies the edges of the stripes. In order to decrease the data extraction time, several automated edge detection algorithms were examined for their suitability in the analysis of contour photographs, and the best was extensively evaluated using Monte Carlo simulation. The accuracy and reproducibility of the edge position estimate in images with various amounts of film grain noise were measured for many values of the edge detector parameter and of the signal parameter, its modulation, m. When normalized by the amount of film grain noise, the relationship between reproducibility and l/m was found to be linear over the range of parameters likely to be encountered in contour photography. The accuracy was found to be independent of the amount of film grain noise, and linearly related to 1/m. By estimating m for each edge, the accuracy could be treated as a correctable systematic error of the edge detection process. A sample calculation which used parameter values that are likely to be found in contour photography showed that the automated edge detection process would be expected to produce a random variation in the measurement of the depth of the optic nerve head surface whose standard deviation is 0.5 micrometers, or 0.05 percent of the deepest part of the surface.

In the case of serial sections observed by the means of an electron microscope, it is possible to rebuild an image of an object, using the local intelligence of an image work-station without the need of a powerful computer. We will explain the basic principles of a program that we have written and explain its further developments.

A system is described which can be used to create a three-dimensional model of a neurone from the central nervous system. This model can then be used to obtain quantitative data on the physical and electrical pro, perties of the neurone. Living neurones are either raised in culture, or taken from in vitro preparations of brain tissue and optically sectioned. These two-dimensional sections are digitised, and input to a 68008-based microcomputer. The system reconstructs the three-dimensional structure of the neurone, both geanetrically and electrically. The user can a) View the structure fran any point at any angle b) "Move through" the structure along any given vector c) Nave through" the structure following a neurone process d) Fire the neurone at any point, and "watch" the action potentials propagate e) Vary the parameters of the electrical model of a process element. The system is targeted to a research programme on epilepsy, which makes frequent use of both geometric and electrical neurone modelling. Current techniques which may involve crude histology and two-dimensional drawings have considerable short camings.

2D images can be projected one after the other on any sensitive material covered by a lenticular sheet. Unlike holography,this process can reproduce all colors,but only inside the "lobe angle". Outside this angle,the order of the images seems to start once again from zero,producing some kind of flipping over. Integraphy is not new,but little used. The greatest interest of two-step integraphy is that the user can take a series of ima-ges himself,making use of about any kind of standard photographic camera,the exposure time, f-stop,light,etc. remain standard also. The films (B&W or diapositive) should be processed by the user,but remain uncut. The task of synthesis,color processing of the 3D copy,supplying of the lenticular sheet,cementing and cutting will be ensured by ourselves. This technique is an extension of all the scientific processes already providing images.

A technique is presented which offers the possibility of automatic measurement of the 3-D shape of the external surfaces of the human body. It is computationally simple and is suitable for implementation on a microcomputer.