replication of human t a tracheobronchial hollow airway &e...

Copyright 2002, American Industrial Hygiene Association

TH

EOR

ETICA

L&

EXPER

IMEN

TAL

STU

DIES

AIHA Journal 63:141–150 (2002) Ms. #302

AIHA Journal (63) March/April 2002 141

AUTHORSRodney E.

Clinkenbearda

David L. Johnsona*Ramkumar

Parthasarathyb

M. Cengiz Altanb

Kah-Hoe Tanb

Seok-Min Parkc

Richard H. Crawfordc

aAerosols Research Laboratory,Department of Occupational andEnvironmental Health,University of Oklahoma HealthSciences Center, P.O. Box26901, Oklahoma City, OK73190;bSchool of Aerospace andMechanical Engineering,University of Oklahoma, 865Asp Ave., Room 212, Norman,OK 73019-0601;cLaboratory for FreeformFabrication, Department ofMechanical Engineering,University of Texas at Austin,ETC II 5.160, Austin, TX78712-1064

Replication of HumanTracheobronchial Hollow AirwayModels Using a Selective LaserSintering Rapid PrototypingTechnique

Exposures to toxic or pathogenic aerosols are known to produce adverse health effects. The

nature and severity of these effects often are governed in large part by the location and

amount of aerosol deposition within the respiratory tract. Morphologically detailed replica hollow

lung airway casts are widely used in aerosol deposition research; however, techniques are not

currently available that allow replicate deposition studies in identical morphologically detailed

casts produced from a common reference anatomy. This project developed a technique for the

precision manufacture of morphologically detailed human tracheobronchial airway models based

on high-resolution anatomical imaging data. Detailed physical models were produced using the

selective laser sintering (SLS) rapid prototyping process. Input to the SLS process was a three-

dimensional computer model developed by boundary-based two-dimension to three-dimension

conversion of anatomical images from the original National Institutes of Health/National Library

of Medicine Visible Human male data set. The SLS process produced identical replicate

models that corresponded exactly to the anatomical section images, within the limits of the

measurement. At least five airway generations were achievable, corresponding to airways less

than 2 mm in diameter. It is anticipated that rapid prototyping manufacture of respiratory tract

structures based on reference anatomies such as the Visible Male and Visible Female may

provide ‘‘gold standard’’ models for inhaled aerosol deposition studies. Adaptations of the

models to represent various disease states may be readily achieved, thereby promoting

exploration of pharmaceutical research on targeted drug delivery via inhaled aerosols.

Keywords: aerosol, model, rapid prototyping, selective laser sintering, tracheobronchial

*Author to whomcorrespondence should beaddressed.This research wassupported under a grantfrom the University ofOklahoma BioengineeringCenter in Norman, Okla.

Diseases of the lungs and other organscaused by inhalation of toxic or infec-tious airborne particles are among theoldest recorded ailments.(1) Included are

such well-known examples as heavy metal poi-soning, silicosis, black lung, and asbestosis inminers; lung cancer and chronic bronchitis inminers, asbestos workers, and smokers; and tu-berculosis in health care workers and the gen-eral population. Various influences determinethe potential for an airborne particle to be in-haled, deposit in the respiratory tract, come in

contact with susceptible tissues, and exert a tox-ic or pathogenic effect. The location of depo-sition of an inhaled particle in the respiratorytract is determined by the respiratory tract mor-phology, breathing rate, breath volume, andnose versus mouth breathing, as well as by thesize, density, shape, and hygroscopicity of theparticle.(2,3) The types of adverse effects that cansubsequently occur as a result of deposition aredetermined by the particle chemical and physi-cal characteristics, the tissues at the depositionsite or at a remote site to which the particle or

Downloaded By: [University of Oklahoma-Tulsa] At: 22:40 23 December 2008

142 AIHA Journal (63) March/April 2002

THEO

RET

ICA

L&

EXPE

RIM

ENTA

LST

UD

IES

its constituents migrate, and the effectiveness of host defensemechanisms.(4)

Just as diseases may be caused by inhaled toxic or pathogenicparticles, diseases may be treated by inhaled pharmaceutical par-ticles.(5) In many therapeutic applications, treatment effectivenessis maximized and unwanted side effects are minimized by directdrug delivery to affected respiratory tract tissues via inhalationrather than indirectly via injection or ingestion. For example,drugs for prophylaxis and treatment of asthma are inhaled frommetered dose inhalers, dry powder inhalers, or nebulizers. Simi-larly, inhalation therapy for pneumocystis carinii pneumonia andother deep lung infections employs nebulized drugs. Certain med-ications for treatment of allergies are marketed in nasal inhalerdevices. Each of these drugs must deposit in a specific region ofthe respiratory tract to exert its therapeutic effect; therefore, meth-ods for ‘‘targeted’’ drug delivery to specific respiratory tract re-gions and tissues are of great interest to the pharmaceuticalcommunity.(6)

The same physiological factors and particle characteristics thatgovern deposition of toxic or pathogenic aerosols govern the re-spiratory tract deposition behavior of inhaled drug particles. Theability to predict aerosol deposition behavior in the respiratorytract is therefore of substantial value in both risk assessment andtherapy applications.(7) However, because ethical considerations se-verely limit the use of human subjects in aerosol deposition re-search, particularly when potentially toxic or pathogenic aerosolsare used, it is necessary to use substitute numerical modeling(computer simulations),(8–11) nonhuman in vivo methods (e.g., liverats, dogs, and swine),(12) or in vitro methods (e.g., excised humanor animal lungs, replica casts, and physical constructs).(13,14)

To this end, physical constructs approximating elements of thelung structure are attractive due to their simplicity and relativelylow cost. Models constructed from plastic or metal have been usedto study particle motion and deposition in airway segments andat airway bifurcations.(15–18) However, these models lack anatomi-cal detail and are necessarily simplistic, representing only an iso-lated fragment of the total lung structure. Morphologically real-istic and detailed replicas of the respiratory tract may be producedwith casting techniques, and numerous techniques and materialshave been used with varying success.(19–23) Current techniques in-volve the injection of a liquid molding material into the airways,curing to form a solid replica of these spaces, and subsequent re-moval of the surrounding tissue by a digestion process.(24–27) The‘‘negative’’ replica thus produced may be used directly to measureairway internal dimensions and evaluate anatomical structures thatmay influence airflow and particle deposition behavior (e.g., larynxstructure, tracheal ribbing, airway branching angles, and carinalridge shapes). Alternatively, the negative may be used as a formfrom which to produce a ‘‘positive’’ hollow cast replica of theairway structure. In the lost wax process a wax negative is coatedwith plaster or a rubber material; then the wax is melted out toleave the hollow cast replica. Detailed replicas may be producedusing this technique, but obviously only one detailed hollow castmay be produced from a given wax negative. A durable positivemold may be produced from a wax negative if the negative istrimmed to represent only a few generations of the airway tree,and the durable positive may then be used to make identical waxnegatives and hollow casts for replicate experiments.(28) Anothertechnique involves coating a durable negative cast with a rubbermaterial that is subsequently cut, peeled away from the core, andthen reassembled. Schlesinger and Lippmann(20) produced such adurable master negative and hollow silicon rubber positives from

a cast pruned to the sixth airway branching. Unfortunately, prun-ing eliminates the smaller airways (less than approximately 3 mmin diameter).

Hollow casts are a preferred method for evaluating respiratorytract particle deposition behavior under well-controlled experi-mental conditions. However, techniques are not currently avail-able that allow replicate deposition studies in identical morpho-logically detailed casts produced from a common referenceanatomy. It would be useful if such a technique could be devel-oped with which an unlimited number of identical hollow modelsof the tracheobronchial portion of the airway tree could be pro-duced. The availability of unlimited numbers of identical modelsrepresenting a common reference anatomy would permit repeatexperiments requiring destructive analysis of the used model, andalso allow comparison of airflow and particle deposition studiesbetween investigators. In this work the present authors combinedhigh-resolution anatomical imaging and precision manufacturingto produce morphologically detailed hollow airway models of thehuman tracheobronchial tree.

VISIBLE HUMAN DATA

High-resolution axial computerized tomography (CT) and an-atomical imaging data have been developed for both the male

and female human through the National Institutes of Health/National Library of Medicine (NIH/NLM) Visible Human Proj-ectt.(29) The Visible Human Male transverse section axial CT andanatomical images were obtained at 1.0-mm intervals over thelength of the body. The CT scans have resolutions of 512 3 512pixels where each pixel is made up of 12 bits of gray tone, andthe anatomical images are 2048 3 1216 pixels where each pixelis defined by 24 bits of color (there are 3 pixels per millimeter ofhorizontal or vertical image dimension). The CT and anatomicalcross-section images coincide. The Visible Human Female data setis even more detailed, with images obtained at 0.33-mm ratherthan 1.0-mm intervals over the body. The data sets are being usedby more than 1000 licensees in 41 countries and are available fromNIH/NLM at no cost. In this work data from the anatomicalimages is used due to their greater resolution.

Rapid Prototyping

Currently, approximately 20 rapid prototyping techniques areavailable.(30) Those evaluated for this project were stereolithogra-phy (SLA), layered object manufacturing (LOM), fused deposi-tion modeling (FDM), three-dimensional printing (3DP), solidground curing (SGC), and selective laser sintering (SLS). Eachtechnique involves the build-up of a solid model as a series of thinhorizontal cross-sections that are fused together. SLA (3DSystems,Valencia, Calif.) technology directs an ultraviolet laser beam ontoa layer of liquid photopolymer resin spread on an elevated hori-zontal platform inside a vat or ‘‘build chamber.’’ The laser tracesthe cross-section image of the object, fusing the exposed materialin the plane. The support platform is then lowered a distanceequivalent to one layer thickness, the surface is recoated with resinto the proper level, and the process repeats until the object is fullyconstructed. LOM (Helysis, Torrance, Calif.) builds objects usinglaminated sheet material coated with a pressure sensitive adhesive.An infrared laser is used to trace (cut) the edge of the desiredcross-section, and waste areas are traced in a crosshatch pattern tofacilitate later separation from the finished object. Subsequent lay-ers are stacked and bonded by the adhesive using a heated roller.FDM (Stratasys, Eden Prairie, Minn.) is a nonlaser technology that



TH

EOR

ETICA

L&

EXPER

IMEN

TAL

STU

DIES

FIGURE 1. SLS. A thin layer of powder is distributed with aroller in the build plane, and areas representing the cross-sectionof the object being constructed are fused with a laser. Thesupporting platform then indexes one layer thickness downward,and the process repeats for the next cross-section.

uses a heated extruding head to deposit a thin filament of moltenthermoplastic that instantaneously solidifies on being deposited.3DP (Z Corp., Burlington, Mass.) builds objects by depositing abinder material on target areas of a thin powder layer using atechnology similar to ink-jet printing. The piston supporting thepowder bed and the part in progress then lowers one layer thick-ness, a new layer of powder is spread, and the process is repeateduntil the part is completed. SGC (Cubital Ltd, Raanana, Israel) issimilar to SLA in that it uses a photopolymer exposed to a laserto create a model, but differs in that the laser does not trace thedesired cross-section; rather, a negative of the desired cross-sectionmasks the photopolymer so that the slice is created by a xero-graphic process. A mill is then used to trim off excess surfacematerial for accuracy. SLS (Figure 1) (DTM Corp., Austin, Tex.)uses powdered materials instead of the liquid photopolymers in aprocess similar to SLA. An infrared laser beam fuses the powderin the target areas, the supporting platform is lowered and powderis resupplied to the surface, and the process is repeated until thebuild is complete.(31,32)

In each of these processes the computer-controlled apparatusworks from a three-dimensional digital model of the object thathas been digitally ‘‘sliced’’ to provide cross-sectional images at0.07 to 0.10 mm intervals along a chosen axis.(33) The thicknessof individual slices and the object’s orientation in the build cham-ber are selected to optimize the production process by minimizingthe number of ‘‘slices’’ (and hence the operating time), requiredfor the build. Thinner slices produce a smoother model surface,but require more build time. If the object contains straight sidesand no rounded corners, the number of slices can be fewer thanfor a part that is circular or one that has many detailed opposingedges.

Rapid prototyping techniques have been employed to modelanatomical subjects and have shown their utility in such applica-tions as presurgical planning structures and solid anatomy repre-sentations.(34–37) Kelley et al. recently used three-dimensional print-ing to produce a soluble nasal cavity negative using CT data.(38)

In their work the negative was dipped in silicone and the starch-based solid ‘‘negative’’ mold was then dissolved away, requiring atwo-step process that resulted in one model from one mold. Todate, no work has been reported in which hollow airway tracheo-bronchial tree models were produced by this two-step techniqueor directly in a one-step process.

A limitation of many of the rapid prototyping systems is therequirement for support structures during some builds.(39) For ex-ample, in building a cylinder oriented with its axis parallel to thebuild platform, slices through the curving walls substantially over-hang one another on the bottom and top portions of the build,and the operating layer is inadequately supported by the layer im-mediately beneath. To provide the required support, the systemsoftware automatically inserts vertical support structures duringthe build. For the horizontal cylinder, support structures wouldbe inserted outside the cylinder to support the lower walls andinside to support the arch of the cylinder ‘‘roof.’’ In the presenteffort to construct a hollow airway human tracheobronchial treemodel, numerous support structures (both outside the tree andinside the airways) would be required, and their postbuild removalwould be extremely problematic. The authors therefore elected touse SLS because support structures are not needed—the unsin-tered powder of previously deposited layers supports subsequentlayers during the build cycle. This approach, if successful, wouldallow modeling of void-space anatomical structures based on im-aging data.

MATERIALS AND METHODS

Three-Dimensional Digital Modeling

Although a three-dimensional computer model in CAD formatcould have been constructed directly from the digitized NIH/NLM anatomical image data using a border-based two- to three-dimensional conversion technique, this step was unnecessary be-cause a three-dimensional model of this type was alreadycommercially available. A three-dimensional point cloud model ofthe tracheobronchial tree has been extracted from the Visible Maleanatomical data set by Visible Productions, Inc. of Ft. Collins,Colo., and a copy of the file in .OBJ and .STL format (compatiblewith rapid prototyping systems) was purchased for this work. Thepoint cloud model included both the interior and exterior surfacesof the conducting airways to at least five airway generations (cor-responding to minimum airway inside diameters of approximately1 mm). The polygonal mesh surface generated from the pointcloud file was scrutinized for defects and edited as necessary usingRhinocerost (Robert McNeel & Associates, Seattle Wash.). Areaswithin the original three-dimensional model that contained erro-neous data points were corrected by deleting facets and data pointswhere necessary. Other portions of the model contained areaswithout facets and data points; in these cases data points wereadded by interpolating between existing data points. Areas withambiguous data points were located and the correct data pointlocations were identified from corresponding Visible Human dataaxial images and added as necessary.

SLS

SLS was chosen for this work because the packed bed of unsin-tered powder provides support for all elements of the tracheo-bronchial tree model during the build, eliminating the need foradditional supporting structures. The powder bed is heated towithin a few degrees of the powder melting point during the buildto minimize the laser energy required to achieve fusion. On com-pletion of the build the powder ‘‘cake’’ is allowed to cool beforeremoving the model.(40) The unsintered powder outside the modelis readily removed, whereas the powder in interior spaces may beremoved by gentle probing, agitation, and the use of compressedair jets.



THEO

RET

ICA

L&

EXPE

RIM

ENTA

LST

UD

IES

FIGURE 2. Accuracy of the digital model. Features present in a cross-section through the digital three-dimensional model (left) were measuredand compared with the same features in the corresponding Visible Male anatomical image (right).

Using a suitable material, a hollow airway ‘‘positive’’ model isproduced that may then be used in experimental work. Prelimi-nary model building was performed at the Laboratory for Free-form Fabrication at the University of Texas at Austin. Productionof the final model was performed at Accelerated Technologies ofAustin, Tex.

Accuracy and Precision Determinations

The accuracy of the three-dimensional computer model was eval-uated by comparing features of the computer model with the samefeatures from the Visible Male data set. The computer model wasdigitally sectioned to obtain a cross-section image that could bematched to the corresponding Visible Male anatomical imagefrom which the original data were derived. Features in the digitalcross-section and the corresponding anatomical image were mea-sured along multiple axes and compared. Three replicate mea-surements were taken along each of three axes and averaged. Mea-surements were performed for 20 randomly selected locations onthe computer model.

The accuracy and precision of the SLS process in producing aphysical model from the three-dimensional computer model wasevaluated by producing multiple identical 1.88:1 scale models ofthe trachea and main bifurcation in a single build. The randomerror associated with manual caliper and scale measurements ofthe model was plus or minus a fraction of a millimeter and wasfixed, so a scaled-up model was used to minimize the effect of thiserror (i.e., the coefficient of variation) in the measurement results.The 1.88:1 scale-up was the maximum that could be accommo-dated in the SLS build chamber when constructing the four rep-licate models at once. Two models were sawed into 5 mm-thicksections and labeled, and their cross-sections were digitized as.BMP images by scanning with a Hewlett Packard 5300C 36-bitcolor scanner. A 6-inch ruler was included in each scan to providean absolute scale reference. The segments were measured from theimages in the same manner as described for the computer modeland axial images. All measurements were performed in AdobePhotoshop 5.5

Proof of Concept Model

Finally, a single normal-size (unscaled) copy of the full point cloudtracheobronchial tree model was produced in a Sinterstation 2000(DTM Corp.) at Accelerated Technologies, Inc. of Austin, Tex. Themodel served to demonstrate proof-of-concept and to determinehow many airway generations could be accurately reproduced.

RESULTS AND DISCUSSION

An example comparison between the computer model cross-sections and anatomical images is shown in Figure 2. The ac-

curacy of the computer model relative to the anatomical imageswas extremely good. The root mean square difference in the mea-sured feature dimensions values was 0.07 mm, and the largest dis-agreement was 0.1 mm. These values were near the limit of theprecision of the measurement technique, which was estimated tobe 60.1 mm. The mean and maximum differences as a fractionof the measured value were 60.033 and 0.12%, respectively.

Pairs of corresponding sections cut from the replicate 1.88:1scale trachea models are shown in Figure 3. The accuracy of thephysical model compared with the computer model was excellent,with an average difference of 60.03%. The within-build precisionof the production process was also excellent, with measurementsdiffering by a maximum of 60.1 mm (the limit of measurement).The between-build precision was not evaluated.

The complete tracheobronchial tree physical model (Figure 4)was produced using DuraFormy (DTM Corp.), a polyamidepowder ranging from 15 to 90 mm in particle size. DuraForm waschosen for model construction due to the strength and laboratory-friendly physical properties of the nylon material, including tensilestrength 44 MPa (ASTM test method D638), tensile modulus1600 MPa, tensile elongation at break 9%, flexural modulus 1,285(ASTM test method D790), and impact strength 214 J/m(ASTM test method D256). In addition, DuraForm has a chem-ical resistance to alkalines, hydrocarbons, fuels, and solvents.(41)

The model required 14 hours to produce in a Sinterstation 2000.



TH

EOR

ETICA

L&

EXPER

IMEN

TAL

STU

DIES

FIGURE 3. Tracheal sections. Three pairs of corresponding sections cut from replicate trachea models were compared and found to be identicalwithin the limits of the measurement method.



THEO

RET

ICA

L&

EXPE

RIM

ENTA

LST

UD

IES

FIGURE 4. Complete model as produced. The normal-scale physical model before trimming to remove unusable extremities.



TH

EOR

ETICA

L&

EXPER

IMEN

TAL

STU

DIES

FIGURE 5. Model defects. Minor defects in the physical model included openings in the airway wall. These defects were due to correctable errorsin the three-dimensional computer model rather than the SLS build.

After 12 hours of cooling, which served to strengthen the struc-ture and prevent breakage of the smaller airways, the model wasremoved from the unfused powder cake. Unfused powder wasremoved by blowing out the airways with compressed air, assistedby careful probing of smaller airways with 24-guage aluminumwire or 8-gauge copper wire. The model was durable, yet slightlyflexible, so that no breakage occurred during extensive manipu-lations. The model withstood extensive manipulation withoutbreakage or collapse of even the smallest segments. The surface ofthe trachea, the most extensively handled area, did not appear toerode, soften, or deform in spite of repeated contact with moisthands.

Several of the smallest airways had to be trimmed back with arazor knife to find the lumen. Trimming was continued to removesegments without an intact lumen or areas with obvious structuralerrors such as malformed airways or airways with wall openings(Figure 5). In each case these defects were found to be due tocorresponding defects in the computer model rather than due toerrors in the SLS build; the model defects were thus repairableusing available software.

After trimming, a minimum of five airway generations wereachieved for the entire model (Figure 6), and six generations wereachieved in several locations. Airways smaller than 1 mm in di-ameter were produced in many locations, though most segments



THEO

RET

ICA

L&

EXPE

RIM

ENTA

LST

UD

IES

FIGURE 6. Tracheobronchial tree model after trimming. A minimum of five airway generations was achieved, with six generations in some areas.Minimum airway diameters were slightly less than 1 mm, but most were 1–2 mm.

were 1–2 mm in diameter. Airway wall thicknesses were generallyless than 1 mm.

The model surface texture, though quite smooth, neverthelessretained the finely ribbed texture typical of rapid prototyping pro-cesses (Figure 5). Studies are currently under way to characterizethe surface roughness and to explore postbuild techniques forsmoothing the surface, such as by coating. Work is also in progressto create a model from the more detailed 0.33-mm sectionedNIH/NLM Visible Female data set.

Rapid prototyping holds great promise for precise and accu-rate modeling of anatomical structures. It has a number of majoradvantages over older casting-based techniques, particularly re-garding its ability to utilize computerized axial tomography,magnetic resonance imaging, or other noninvasive imaging datafrom living persons as well as the cadaver-derived anatomical sec-tioning data used in this work. Although at present the high-resolution anatomical imaging data may be the most detailed andtherefore the best available for modeling fine structures such assmall airways, improvements in the noninvasive technologies maysoon allow in situ modeling for study of various disease condi-tions such as restricted or occluded airways or major bloodvessels.

SUMMARY AND CONCLUSIONS

An SLS rapid prototyping technique was used to manufacturemorphologically detailed models of human tracheobronchial

tree structures. A commercially available three-dimensional pointcloud computer model of the NIH/NLM Visible Male tracheo-bronchial tree was adapted for use with SLS, and the resultingphysical models were evaluated to characterize the accuracy andprecision of the method in this application. Airway structuresshown in digital cross-sections of the three-dimensional file wereessentially identical to the same structures shown in correspondinghigh-resolution photographic images of anatomical sections fromthe original Visible Male data set. Similarly, replicate models ofsegments of the trachea and main bifurcation produced in a singlebuild were found to be identical within the limits of the measure-ment. A morphologically detailed actual-size tracheobronchial treemodel was produced with intact airways to at least five branchingsand airway diameters slightly less than 1 mm. A number of minordefects in the physical model were identified and determined tobe due to correctable errors in the digital three-dimensionalmodel.



TH

EOR

ETICA

L&

EXPER

IMEN

TAL

STU

DIES

The technique developed in this work provides a means of pro-ducing an unlimited number of identical copies of a morpholog-ically detailed tracheobronchial airway model suitable for variousresearch applications. The model is based on the widely used an-atomical data generated under the NIH/NLM Visible HumanProject. Excellent results were obtained in this proof-of-concepteffort using the relatively low-resolution 1-mm section male ana-tomical data, and it may be expected that the higher-resolution0.33-mm section data for the female will yield even more detailedmodels. Hollow airway models extending beyond six branchingsmay be achievable when this data has been extracted and con-verted to an SLS-compatible format. In addition, although SLSrapid prototyping was used for tracheobronchial tree modeling inthis work, the technique can be directly applied to the productionof other anatomical structures such as the nasopharyngeal passag-es, major blood vessels, bones, and so forth, so that the modelingmethod has potential for widespread application in multiple fieldsof bioengineering research and education. For aerosol depositionresearch in particular, it is the only method capable of producingan unlimited number of identical copies of replicate human airwaymodels including the tracheobronchial tree, for which currentmodels are inadequate.

It is anticipated that rapid prototyping manufacture of respi-ratory tract structures based on reference anatomies such as theVisible Male and Visible Female may provide ‘‘gold standard’’models for inhaled aerosol deposition studies. Adaptations of themodels to represent various disease states may be readily achieved,thereby promoting exploration of pharmaceutical research on tar-geted drug delivery via inhaled aerosols.

ACKNOWLEDGMENTS

The authors wish to express their appreciation to the staff ofVisible Productions, Inc. and Accelerated Technologies, Inc.

for their advice and assistance in data file adaptation and modelproduction.

REFERENCES

1. Rose, V.E.: History and philosophy of industrial hygiene. In S.R.DiNardi, editor, The Occupational Environment—Its Evaluation andControl, pp. 3–18. Fairfax, Va.: AIHA Press, 1997.

2. Raabe, O.G., M.A. Al-Bayati, S.V. Teague, and A. Rasolt: Re-gional deposition of inhaled monodisperse coarse and fine aerosol par-ticles in small laboratory animals. Ann. Occup. Hyg. 32(suppl 1):53–62 (1988).

3. Heyder, J., J. Gebhart, G. Heigwer, C. Roth, and W. Stahlhofen:Experimental studies of the total deposition of aerosol particles in thehuman respiratory tract. Aerosol Sci. 4:191–208 (1973).

4. Costa, D.L., and E.S. Schelegle: Irritant air pollutants. In D.L. Swiftand W.M. Foster, editors, Air Pollutants and the Respiratory Tract,pp. 119–145. New York: Marcel Dekker, 1999.

5. Atkins, P.J., N.P. Barker, and D. Mathisen: The design and devel-opment of inhalation drug delivery systems. In A.J. Hickey, editor,Pharmaceutical Inhalation Aerosol Technology, pp. 155–186. NewYork: Marcel Dekker, 1992.

6. Gonda, I.: Targeting by deposition. In A.J. Hickey, editor, Pharma-ceutical Inhalation Aerosol Technology, pp. 61–82. New York: MarcelDekker, 1992.

7. Cohen, B.S., R.G. Sussman, and M. Lippman: Factors influencingdistribution of airflow in a human tracheobronchial cast. Resp. Physiol.93:261–278 (1993).

8. Soong, T.T., P. Nicolaides, C.P. Yu, and S.C. Soong: A statistical

description of the human tracheobronchial tree geometry. Resp.Physiol. 37:161–172 (1979).

9. Horsfield, K., G. Dart, D.E. Olson, G.F. Filley, and G. Cumming:Models of the human bronchial tree. J. Appl. Physiol. 31:207–217(1971).

10. Tawhai, M.H., P.J. Hunter, and A.J. Pullan: Generation of an an-atomically based three-dimensional model of the conducting airways.Ann. Biomed. Engin. 28:793–802 (2000).

11. Rudolf, G., R. Kobrich, and W. Stahlhofen: Modeling and alge-braic formulation of regional aerosol deposition in man. J. Aerosol Sci.21(suppl 1):s403–s406 (1990).

12. Yeh, H.C., R.F. Phalen, and O.G. Raabe: Factors influencing thedeposition of inhaled particles. Environ. Health Persp. 15:147–156(1974).

13. Pedersen, C.J., J.W. Watson, and A.C. Jackson: Technique for mak-ing hollow central airway casts. J. Appl. Physiol. 55:254–257 (1983).

14. Ferron, G.A.: Deposition of polydisperse aerosols in two glass modelsrepresenting the upper human airways, J. Aerosol Sci. 8:409–427(1977).

15. Guichard, J.C., A. Zerrari, and M.Y. Jaffrin: Etude experimentaledu depot de particules fibreuses dans un modele bronchique rigide.J. Aerosol Sci. 19:73–85 (1988).

16. Myojo, T.: Deposition of fibrous aerosol in model bifurcating tubes.J. Aerosol Sci. 18:337–347 (1987).

17. Myojo, T.: The effect of length and diameter on the deposition offibrous aerosol in a model lung bifurcation. J. Aerosol Sci. 21:651–659 (1990).

18. West, J.B., and P. Hugh-Jones: Patterns of gas flow in the upperbronchial tree. J. Appl. Physiol. 14:753–759 (1959).

19. Phalen, R.P., H.C. Yeh, O.G. Raabe, and D.J. Velasquez: Castingthe lungs In Situ. Anat. Rec. 177:255–264 (1973).

20. Schleisinger, R.B., and M. Lippmann: Particle deposition in castsof the human upper tracheobronchial tree. Am. Ind. Hyg. Assoc. J.33:237–251 (1972).

21. Briant, J.K., and M. Lippman: Particle transport through a hollowcanine airway cast by high-frequency oscillatory ventilation. Exp. LungRes. 18:385–407 (1991).

22. Eisman, M.: Lung models, hollow flexible reproductions. J. Appl.Physiol. 29:541–543 (1970).

23. Frank, N.R., and R.D. Yoder: A method of making a flexible castof the lung. J. Appl. Physiol. 21:1925–1926 (1966).

24. Fang, C.P., B.S. Cohen, and M. Lippman: Aerosol tracer study ofgas transport to 0.1-cm airways by high-frequency ventilation in ahuman lung airway cast. Exp. Lung Res. 18:615–632 (1992).

25. Patra, A.L., and E.M. Afify: An experimental study of velocity dis-tribution in a human lung cast. J. Biomech. Engin. 105:381–388(1983).

26. Tompsett, D.H.: A new method for the preparation of bronchopul-monary casts. Thorax 7:78–88 (1951).

27. Pederson, C.J., J.W. Watson, and A.C. Jackson: Technique formaking hollow central airway casts. J. Appl. Physiol. 55:254–257(1983).

28. Cheng, Y.S., S.M. Smith, and H.C. Yeh: Deposition of ultrafineparticles in human tracheobronchial airways. Ann. Occup. Hyg.41(suppl 1):714–718 (1997).

29. National Institutes of Health: Fact Sheet: The Visible Human Pro-jectt. [Online] Available at http://www.nlm.nih.gov/pubs/fact-sheets (Accessed 1999).

30. Karapatis, N., J. Van Griethuysen, and R. Glardon: Direct rapidtooling: A review of current research. Rapid Prototyping J. 4(2):77–89 (1998).

31. Williams, J.D., and C.R. Deckard: Advances in modeling the effectsof selected parameters on the SLS process. Rapid Prototyping J. 4(2):90–100 (1998).

32. Miller, D., C. Deckard, and J. Williams: Variable beam size SLSworkstation and enhanced SLS model. Rapid Prototyping J. 3(1):4–11 (1997).

33. Xu, F., H.T. Loh, and Y.S. Wong: Considerations and selection of



THEO

RET

ICA

L&

EXPE

RIM

ENTA

LST

UD

IES

optimal orientation for different rapid prototyping systems. RapidPrototyping J. 5(2):54–60 (1999).

34. Barker, T.M., W.J. Earwaker, and D.A. Lisle: Accuracy of stereo-lithographic models of human anatomy. Australasian Radiol. 38:106–111 (1994).

35. Berry, E., J.M. Brown, M. Connell, et al.: Preliminary experiencewith medical applications of rapid prototyping by selective laser sin-tering. Med. Eng. Phys. 19:90–96 (1997).

36. Bouyssie, J.F., S. Bouyssie, P. Sharrock, and D. Duran: Stereolith-ographic models derived From X-ray computed tomography, repro-duction accuracy. Surg. Radiol. Anat. 19:193–199 (1997).

37. Chang, K.H., P.S. Tang, and Q. Sun: Solid model construction

from discrete section contour data. Journal of Computer-Aided Design[Manuscript in review]

38. Kelly, J.T., A.K. Prasad, and A.S. Wexler: Detailed flow patterns inthe nasal cavity. J. Appl. Physiol. 89:323–337, (2000).

39. Wang, X.: Calibration of shrinkage and beam offset in SLS process.Rapid Prototyping J. 5(3):129–133 (1999).

40. Bugeda, G., M. Cervera, and G. Lombera: Numerical predictionsof temperature and density distributions in selective laser sinteringprocess. Rapid Prototyping J. 5(1):21–26 (1999).

41. DTM Corp.: DuraFormy polyamide and DuraFormy GF proper-ties. In DTM Corporation Material Fact Sheet, p. 1. Austin, Tex.:DTM Corp., 1998.


replication of human t a tracheobronchial hollow airway &e...

Documents