A Simple Non Invasive ComputerizedMethod for the Assessment ofBone Repair within OsteoconductivePorous Bioceramic Grafts

Francesco Beltrame,1 Ranieri Cancedda,2,3 Barbara Canesi,1 Antonio Crovace,4

Maddalena Mastrogiacomo,2,3 Rodolfo Quarto,2,3 Silvia Scaglione,1 Carmela Valastro,4 Federica Viti1

1Dip. Informatica, Sistemistica e Telematica, Universita degli Studi di Genova, Italy,Viale Causa, 13. 16145 Genoa, Italy; telephone: 0039 010 353 2789; fax: 0039 010 3532948; e-mail: silvia77@dist.unige.it2Istituto Nazionale per la Ricerca sul Cancro, Genova, Italy3Dip. Oncologia, Biologia e Genetica, Universita degli Studi di Genova, Italy4Dip. Emergenze e dei Trapianti di Organi, Universita degli Studi di Bari, Italy

Received 28 January 2005; accepted 28 April 2005

Published online 8 July 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20591

Abstract: Single energyX-ray imaging, due to its lowcostand flexibility, is one of the most used and commontechnique to assess bone state and bone remodeling overtime. Standardized X-ray images are needed to comparesets of radiographs for semi-quantitative analyses oftissue remodeling. However, usefulmathematical model-ing for the analysis of high level radiographic images arenot easily available. In order to propose a useful evalua-tion tool to a wide clinical scenario, we present aninnovative calibration algorithm for a semi-quantitativeanalysis of non-standardized digitized X-ray images. Forcalibration on a unique standardization scale, three timeinvariant regions (ROI) of radiographs were selected andanalyzed. Theaccuracyof thenormalizationmethod forX-ray films was successfully validated by using an alumi-num step wedge for routine X-ray exposures as tool tostandardize serial radiographs (Pearson correlation test:R2¼ 0.96). This method was applied to investigate theprogression of the new bone deposition within ceramicscaffolds used as osteoconductive substitute in largebone defects taking advantage of a large animal model.This innovative image-processing algorithm allowed theidentification and semi-quantification of the bone matrixdepositedwithin the implant. The osteo-integration at thebone-implant interface was also investigated. A progres-sively increasingbone tissuedepositionwithin theporousbioceramic implant and a progressive osteo-integrationwas observed during the 12 months of the trial.� 2005 Wiley Periodicals, Inc.

Keywords: X-ray imaging; engineered bone tissue;calibration; bone formation; osteoconductive grafts

INTRODUCTION

State and structure of bone and its remodeling are typically

assessed by X-ray imaging, according to the radiolucency

properties of the skeleton tissue (Audenino et al., 1996;

Campanacci et al., 1998; Dujardin et al., 1996; Elliott and

Binkley, 2004; Genant et al., 2000; Goitz et al., 1996;

Johnson et al., 1996; Maas et al., 2002; Mazess and Barden,

1988; Sartoris et al., 1992; Wakefield et al., 2004). Bone

changes over time are routinely monitored in cases of osteo-

porosis, bone fractures, and other bone pathologies. The

evaluation of the mineralization progression is also crucial

when large bone defects are treated with a bioengineering/

tissue engineering approach both in animal experimental

surgery and in the clinical practice. In both cases, treated

bones are routinely monitored and sets of radiographs cover-

ing the overall healing period are analyzed and processed in

order to investigate the in vivo bone formation within the

osteoconductive grafts. In fact performance of osteoconduc-

tive grafts should be monitored over time in terms of: (i) new

bone tissue deposition within the implant and (ii) osteo-

integration between host bone and graft (Adani et al., 2004;

Den Boer et al., 2003; Gao et al., 1995; Kon et al., 2000;

Marcacci et al., 1999; Petite et al., 2000; Quarto et al., 2001).

Currently, routine procedures in the clinics and orthopae-

dics are performed on non-standardized images of conven-

tional single-energy X-ray technique, given its low costs,

high availability, and flexibility. However, based on its

intrinsic physical properties and on the multiple variables

involved (exposure setting, beam source-film distance, batch

of film, attenuation coefficients, X-ray filtration) in the X-ray

exam, luminosity and contrast on radiographs of the same

subject generally differ, limiting the assessment of bone and

the measurement of its mineral density.

�2005 Wiley Periodicals, Inc.

Correspondence to: Silvia Scaglione

Contract grant sponsors: The Italian Ministry of Instruction, University

and Research (MIUR-FIRB); The European and the Italian Space Agencies

(ESA & ASI)

Innovative diagnostic imaging techniques, such as dual

energy X-ray absorption (DXA), quantitative ultrasounds

(US), micro-computerized tomography (CT), have been

proposed as useful quantitative approaches for the bone

densitometry (Blake et al., 1997; Elliott and Binkley, 2004;

Genant et al., 2000; Guglielmi et al., 1995; Ibanez, 2003;

Maas et al., 2002; Mazess and Barden, 1988; Sartoris and

Resnick, 1990; Seeman and Martin, 1989; Simonini et al.,

2005; Wakefield et al., 2004).

The use of an X-ray wedge calibration tool has also been

proposed by some researchers groups (Bolander et al., 1990;

De Josselin de Jong et al., 1989; Dornier et al., 2004;

Escarpinati et al., 2002; Haidekker et al., 2004; Martin et al.,

1990; Pearson et al., 2002; Saiki et al., 1993; Symmons,

2004) as a suitable tool to normalize radiographs. Never-

theless up to now in the orthopaedic scenario the assessment

of bone tissue is still monitored by using the single energy

X-ray exam, where no additional device is incorporated

during the routine protocol.

In this paper, we report an innovative calibration algorithm

for the semi-quantitative analysis of non-standardized X-ray

images as an evaluation tool of bone regeneration within

porous biomaterials useful to a wider experimental and

clinical scenario.

This tool has been validated and used on osteoconductive

grafts implanted to replace large bone defects of sheep

within an experimental study. The method was applied to

quantify the progression of the new bone matrix deposition

within the implant during 12 months of observation. The

osteo-integration at the bone-implant interface was also

investigated. Information related to the integrity and perfor-

mance of the implant and to the healing progression were

derived.

MATERIALS AND METHODS

Large Animal Model for Bone Repair

Six-year-old ewes were involved in this study upon proper

approval of competent ethical committees and legal autho-

rities. Surgical procedures were performed essentially as

described in Marcacci et al. (1999) except that an external

fixator type II was used. A large bone segment (4.5 cm of

length) was surgically resected from the central part of the

left tibia of sheep and replaced using a ceramic resorbable

cylinder as osteoconductive graft. In this series of experi-

ments, the scaffoldwas a 100%syntheticCalcium-Phosphate

multiphase biomaterial containing 67% Silicon-stabilized

Tricalcium-Phosphate (Si-TCP) and 33% Hydroxyapatite/

beta Tricalcium-Phosphate (HA/beta-TCP). This scaffold

had a 60%porosity andwas produced byMilleniumBiologix

inc. (Kingston, Ontario, Canada). The external fixator type II

used as orthopaedic device to fix the implant area was

removed 6 months after surgery. The implant region was X-

ray acquired immediately after surgery and thereafter each

month until the animal sacrifice. The anteroposterior (AP)

view was analyzed. To validate the method, an aluminium

step wedge was used, consisting of 11 steps of increasing

thickness (3–33 mm) (EL.SE. S.r.l.).

X-Ray Images Acquisition

The radiograms were acquired with a mobile X-ray unit

(Gilardoni Condensorgil Kv 150� 50 mA s) using Fuji

Medical X-ray Films Super HR-GB 100 NIF (30� 40) and

Fuji G-8 intensifying green fluorescent screens. Film

processing was done automatically with a Cawomat 2000

IR (Cawo).

The X-ray exams were performed in the two orthogonal

projections: caudo-cranial and medio-lateral, with the

following exposure parameters: Kv 55 and 7.5 mA s, focal

distance 100 cm. For the cranio-caudal projection animals

were positioned on the table in left decubitus with extended

right tibia. (The X-ray cassette was kept in vertical position

with a cassette stand andwas leant on the cranial aspect of the

right tibia). For the medio-lateral projection, patients were

positioned in right decubitus.

X-Ray Images Calibration

Radiographs were scanned using transmitted light, digitized

with 400 dpi of resolution, 256 grey levels. An image analysis

software tool was developed in Matlab 6.5 environment, to

calibrate each digitized radiograph on a unique standardiza-

tion scale, using a reference image belonging to the set,

according to its largest grey levels range.

For each image, three ‘‘time constant’’ regions of interest

(ROI) adapted to the animal bone size were chosen and

analysed: (i) the background of the radiograph, (ii) the

cortical bone area, and (iii) the central bone area (Fig. 1). A

rectangular area was used to select the ROI and, for each

digital radiographic image, three average grey-levels were

obtained and used for the calibration. A quadratic mathema-

tical transfer function was developed to get a new set of

calibrated radiographic images:

yi ¼ Ax2i þ Bxi þ C

where xi and yi correspond to the i-pixel of the image

respectively before and after the calibration on the reference

image; for each radiographs, A, B, and C represent three tra-

nsfer coefficients which are automatically calculated based

on the ROI equalization:

for xi ¼ background of the image to be calibrated )yi ¼ background of reference image

for xi ¼ cortical area of the image to be calibrated )yi ¼ cortical area of reference image

for xi ¼ central bone area of the image to be calibrated )yi ¼ central bone area of reference image

190 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 92, NO. 2, OCTOBER 20, 2005

Some radiographs included an aluminium step phantom

for routine X-ray expositions with 11 3-mm-thick steps. In

these images, three different steps of the step phantom were

considered as ‘‘time constant’’ regions.

After the standardization step, images of the same set

showed the same average grey levels on the background area,

on the cortical bone and on the central bone regions. All

pixels of imagesweremathematically converted according to

the mathematical transfer function (Fig. 2).

Bone Optical Density

The progressive increase of radio-density in the implant area

was measured selecting in each radiograph three constant

areas of the cylindrical implant: one in the central field (the

area confined between the two tangents to the inner cavity of

the implant) and two in the lateral ones (the region confined

between the inner and the external tangents) (Fig. 3).

Background noise was automatically removed from each

image, subtracting after the calibration step the background

grey level from each pixel of all the radiographic images.

The time varying contribution of the biomaterial implan-

ted, in terms of radio-opacity, was also removed according to

its exponential resorption kinetics (50% after 4 months, 70%

after 6months).

After these two steps, for each set of radiographs the

progressive increase of the radio-density of the graft was

shown and associated to the ongoing formation of new bone

tissue within the implant (Fig. 4A).

Osteo-Integration Analysis

X-ray calibrated images were also used to analyze the

physical integration at the interface between the implant and

the host bone. Different lines of interest at the interface

‘‘bone-implant’’ were selected in images obtained at

different time from surgery and for each line the intensity

levels of the pixels belonging to the linewere plotted (Fig. 5).

RESULTS

Validation Model

To validate the normalization method and test its accuracy,

we analysed a set of radiographic images containing an

aluminium calibration phantom with 11 3-mm-thick steps.

For each image (Fig. 6A), three different steps of the phantom

were identified as time-invariant regions during the auto-

matic normalization and a new set of radiographs was

generated (Fig. 6B) and compared to the result of the semi-

automatic normalization method (Fig. 6C).

The optical density increase at the implant region

calculated considering the steps of the calibration phantom

as time-invariant regions was statistically comparable with

the increase determined considering the background area, the

cortical bone area and central bone area as the three time-

invariant regions (Pearson test, R2¼ 0.96).

In the X-ray images normalized with the calibration

phantom, minimal variations (&14%) of the cortical bone

density were detected on the operated bone during the

6 months of observation. Moreover, the histograms of both

healthy long bone and the operated one at the 6th month of

implant were compared (Fig. 7). An almost complete

overlapping of the two histograms confirmed that the status

of the operated leg during the experimental period didn’t

significantly affect the bone optical density.

X-Ray Image Calibration

The accuracy of the normalization method was successfully

assessed with histogram analysis of both cortical and

medullar bone regions (ROI) of X-ray images before and

after the calibration step (Fig. 8). In the original X-rays

images set, both cortical andmedullar bone regions displayed

inhomogeneous gray levels range from the post-operation

radiograph up to the 170th day one, filling almost the overall

range from 0 to 255 pixel gray values. After the calibration,

radiographs were re-displayed at the same gray scale and

displayed the same gray scale distribution from the post-

operation radiograph up to the last one both for the cortical

and medullar bone regions.

Figure 1. Selection of the ROIs to normalize the radiographic set. Three

‘‘time constant’’ ROIs (the background of the radiograph, the cortical bone

area, and the central bone one) were chosen and analysed for each image by

using rectangular areas adapted to the animal bone size.

BELTRAME ET AL.: ASSESSMENT OF BONE REPAIR 191

Quantification of the New Bone Tissue Deposition

The semi-automatic calibration tool was used to estimate the

in vivo deposition of new bone tissue within the osteocon-

ductive graft implanted to repair a large bone defect. The

progressive bone deposition from the time of surgery to the

animal sacrifice was determined separately for the inner and

the external areas of the cylinder.

The on time increase of the optical density (OD) at the

implant region was measured and correlated to the in vivo

deposition of the new bone tissue, since the time varying

contribution of the bioresorbable biomaterial, in terms of its

OD, was quantified for the overall observation window and

removed. In Figure 4B, the average increase of the optical

density for the inner and the external areas of the cylinder

(average and standard error) in the six trial animals was

shown, together with the optical densities of the central and

lateral areas of the untreated bones chosen as reference. After

6months from surgery, the amount of bone tissue detected by

theX-ray analysis reached reference levels, in terms of radio-

opacity. The successful removal of the external fixator at

about the 6th month from surgery, confirmed the good

outcome of the implant in terms of osteointegration and

mechanical properties.

Osteointegration of the Implant

The developed software was also used to investigate the

integration between the ceramic graft and the host ovine bone.

At the beginning of the study, the ‘‘bone-implant-bone’’

segment was formed by three distinct parts physically

separated (Fig. 5A). After 1–2 months, a partial bone

depositionwas shown at the interface, while at the 6thmonth,

the ceramic scaffold was completely integrated within the

bone (Fig. 5B), indicating a total osteointegration at the bone-

implant interface.

DISCUSSION

Comparison of serial radiographs is a standard procedure to

follow the repair progression of large bone defects treated

with implants of osteoconductive scaffolds. Because of the

large number of variable settings during the X-ray exposure,

luminosity and contrast of radiographs of the same subject

but taken at different times generally differ.

In the clinic scenario, several imaging techniques are

available to assess the bone status and its remodeling, either

by using a 2D approach (such as single-energy X-ray, dual-

energyX-ray absorptiometry, ultrasounds) or introducing 3D

imaging techniques (MRI, QCT, micro-CT). Among these,

the dual-energy X-ray absorptiometry (DXA) represents the

gold standard for densitometry (Compston et al., 1995;

Mazess and Barden, 1988), since the two energies enable to

measure the bone density with high accuracy. Nevertheless,

the routine diagnosis exam to scan the bone tissue status still

remains the single-energy X-ray, mainly due to its low cost,

high availability and easy to use.

The use of conventional X-ray imaging approach, how-

ever, permits neither a precise evaluation of bone status nor a

quantitativemeasurement of the bonemineral density. In this

Figure 2. X-ray serial images calibration. Set of radiographic images (panel A) belonging to the same animal were calibrated according to the normalization

tool described (see Materials and Methods section) and re-calculated on a unique standardization scale (panel B). For each image, three time constant regions

were selected (background, cortical bone and central bone) and used to normalize radiographic images towards the one chosen as reference. During the

calibration process, pixels of all images were converted according to the coefficients obtained for each image from the implemented mathematical function.

192 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 92, NO. 2, OCTOBER 20, 2005

context, two different approaches may be followed: (i) to

introduce a standard calibration method of radiographs, by

using ametal wedge calibrator for X-ray films, (ii) to develop

an innovative procedure of image processing, which does not

require the use the wedge calibrator at the time of the

radiograph exposure.

In the attempt to provide a larger population of

orthopaedics and clinicians with a simple semi-quantitative

method to monitor in vivo bone formation by using the most

common 2D structure image technique, we developed a new

semi-automatic tool to normalize and compare non-standar-

dized X-ray films. This tool has been validated and used on

osteoconductive grafts implanted to replace large bone

defects of sheep within an experimental study. We have

assumed that, within each X-ray images set belonging to the

same animal monitored monthly for 1 year, the background

region of the radiographic films and the healthy bone tissue

Figure 3. Panel A: Selected fields of the osteoconductive implanted graft.

The central fieldwas defined as the area confined between the two tangents to

the inner cavity of the implant parallel to theX-ray beam; the lateral fieldwas

defined as the region confined between the inner and the external tangent of

the circular section. Panel B: Selection of the ROIs at the implant region.

Based on the size of the animal bone, three rectangular area were used to

select in each X-ray image the central region of the implant and the two

lateral regions.

Figure 4. Panel A: Time evolution of the optical density in the implant

area. The progressive increase of the optical densities in the lateral and

central fields (dashed line and solid line respectively) of the implant is shown

for a serial of radiographs. After subtraction of the background noise and of

the optical density attributable to the residual ceramic biomaterial in the

implant area, the trend lines (3rd degree polynomial function) show a

progressive deposition of new bonewithin the grafts.Panel B: Average bonedeposition in the implants performed in all sheeps (n¼ 6). Average values

and their standard error are shown together with the optical density levels of

the central (black line) and lateral areas (grey line) of the untreated bones

chosen as reference. The average OD values were plotted by a trend line

corresponding to a 3rd degree polynomial function. The external fixator was

removed between the 170th and the 200th day after the implant.

BELTRAME ET AL.: ASSESSMENT OF BONE REPAIR 193

region confining the implant remain constant during the

overall healing period.

These areas therefore were used to normalize the overall

set of radiographs on the same grey scale. A quadratic

mathematical transfer function was chosen, since three

boundary conditions were identified.

In order to keep this assumption truthful and to avoid

artifacts due to inflammation or periosteal responses to the

surgery, the calibrationwas done selecting regions of bone far

from the implants.

The accuracy of the normalization method for X-ray films

was successfully validated by using an aluminium step

wedge as tool to standardize serial radiographs.

This result opens awide scenario for the use of our method

in the orthopaedics, radiological, and clinical fields where,

while the X-ray exposure remains one of the most common

diagnostic tools and, the comparison of not-standardized

films for the assessment of bone turnover in case of osteo-

porosis, bone fractures, and other bone pathologies is a

routine procedure.

Figure 5. Analysis of the osteo-integration at the bone implant interface. Two radiographic images of the same implant taken at different timeswere analysed

(panelA: 20 days after surgery;panelB: 6months after surgery). The pixel intensity levels of three selected lines of interestwere shown. In panelA, a separation

of the implant from the host bone was evident. On the contrary, 6 months after surgery (panel B), the graft was well integrated with the bone surrounding the

implant.

Figure 6. X-ray serial images calibration either with the semiautomatic tool or with the aluminium step wedge. As validation, radiographic images were

calibrated either using the aluminium step wedge (panel B), or our normalization tool (panel C). Results of the two normalization techniques were displayed

selecting as reference image the radiograph acquired at the 50th day of implantation (panel A).

194 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 92, NO. 2, OCTOBER 20, 2005

Figure 7. Cortical and medullar bone histograms. Cortical and medullar bone ROIs were selected both in the original radiographic set and in the normalized

one, and histograms of each ROIwere displayed. The first column (A) refers to the cortical bone ROI of original X-rays, from the day 0 radiograph (first row) up

to the 170th day one (last row), the column (B) refers to the cortical boneROI ofX-ray images normalizedwith our semi-quantitativemethod, the third and fourth

columns (C and D) show the histograms of the medullar bone ROI of original and normalized radiographs respectively.

BELTRAME ET AL.: ASSESSMENT OF BONE REPAIR 195

Figure 7. (Continued)

196 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 92, NO. 2, OCTOBER 20, 2005

The tool allowed the investigation of the bone regenerating

ability of a Calcium-Phosphate multiphase osteoconductive

hollow cylinder graft (67% Si-TCP) and 33%HA/beta-TCP)

implanted in sheep animal models. This scaffold presents

specific osteoconductive properties and a structure permis-

sive to vascular invasion. In agreement with our results in a

parallel series of experiments in which the scaffold was

stabilized with a neutralizing plate (Mastrogiacomo et al.,

submitted), a good osteointegration and mechanical perfor-

mance was also obtained in this series of experiments where

the scaffold was stabilized by external fixation. A significant

progressive increasing of optical density was shown at the

implant region and cross-correlated to the neo-bone matrix

deposition. After a linear growth of the optical density during

the first months of observation, physiological levels of bone

optical density were reached at the 6th month from surgery

and maintained on time, as shown by plateau behavior in

Figure 4B.

The possibility of under-evaluating bone formation at the

implant region when using a bioresorbable scaffold was

avoided by introducing within the semi-quantitative tool a

mathematical function plotting the time-resorption values of

the biomaterial.

In principle, the variation of the optical density detected at

the implant area is proportional to the physical mineral

density of the tissues analysed, expressed in g/cm3, as it’s

customary in bone densitometry. In case a precise assessment

of a quantitative bone mineral density indices is required,

either for comparison studies from different groups or for

final diagnosis of skeletal diseases, a proper calibration

standard, such as DXA, quantitativeUS, or tomography axial

computerized (TAC) should be introduced.

Parallel to the continuous bone tissue depositionwithin the

implant during the healing period, a good integration

between the ceramic graft and the host bone enclosing the

graft was also observed. A good osteointegration is an

important prerequisite for an optimal mechanical stability of

the implant. This method should provide to the orthopaedics

a solid tool to determine the correct time window for the

fixator removal and eventually improve the recovery period

of the patients.

In conclusion, we have developed a semi-automatic

algorithm to normalize serial radiographic images related

to the orthopaedic field. The systemwas validated and used to

quantify the continuous deposition of mineralized new bone

within the implant region in a tibia sheep model. Possibly,

this procedure could be extended to wider experimental and

routine orthopaedic scenarios to study bone turnover and

repair.

The authors thankMilleniumBiologix Inc, Kingston, Ontario, Canada

for providing samples of ceramic scaffolds before they were available

on the market.

References

AdaniR,DelcroixL, InnocentiM,Marcoccio I, TaralloL,CelliA,CerusoM.

2004. Reconstruction of large posttraumatic skeletal defects of the

forearm by vascularized free fibular graft. Microsurgery 24:423–429.

Audenino AL, Zanetti EM, Calderale PM. 1996. Radiograph processing for

quantitative assessment of bone remodelling. Med Eng Phys 18:382–

389.

Blake GM, Gluer CC, Fogelman I. 1997. Bone densitometry: Current status

and future prospects. Br J Radiol 70 Spec N:S177–S186.

Bolander ME, Gardiner J, Coyle J, Joyce M, Himmel HH, Thornton BC,

Ruttimann UE. 1990. Detection and measurement of simulated early

rheumatoid lesions of the hand using digital subtraction radiography.

Invest Radiol 25(6):708–713.

CampanacciM,MercuriM,GasbarriniA, Campanacci L. 1998. Thevalue of

imaging in the diagnosis and treatment of bone tumors. Eur J Radiol 27:

S116–S122.

Compston JE, Cooper C, Kanis JA. 1995. Bone densitometry in clinical

practice. Br Med J 310:1507–1510.

De Josselin de Jong E, Ten Bosch JJ, Noordmans J. 1989. Optimised

microcomputer-guided quantitative microradiography on dental miner-

alised tissue slices. Phys Med Biol 32(7):887–899.

Den Boer FC, Wippermann BW, Blokhuis TJ, Patka P, Bakker FC,

Haarman HJ. Th M. 2003. Healing of segmental bone defects with

granular porous hydroxyapatite augmented with recombinant human

osteogenic protein-1 or autologous bone marrow. J Orthop Res 21:

521–528.

Dornier C, Dorsaz-Brossa L, Thevenaz P, Casagni F, Brochut P,Mombelli A,

Vallee J. 2004. Geometric alignment and chromatic calibration of serial

radiographic images. Dentomaxillofac Radiol 33(4):220–225.

Dujardin FH, Bocquet G, Ertaud JY, Thomine JM. 1996. Quantitative

assessment of cortical bone remodelling from routine radiographs of

total hip arthroplasty. Med Eng Phys 18:489–494.

Elliott ME, Binkley N. 2004. Evaluation and measurement of bone mass.

Epilepsy Behav 5(Suppl 2):S16–S23.

Escarpinati MC, Costa Vieira MA, Schiabel H. 2002. Computer technique

for digital radiographic images correction based on the digitizer

characteristic curve. J Digit Imaging 15(Suppl 1):228–230.

Gao TJ, Lindholm TS, Kommonen B, Ragni P, Paronzini A, Lindholm TC.

1995. Microscopic evaluation of bone-implant contact between

hydroxyapatite, bioactive glass and tricalcium phosphate implanted in

sheep diaphyseal defects. Biomaterials 16:1175–1179.

Genant HK, Gordon C, Jiang Y, Link TM, Hans D, Majumdar S, Lang TF.

2000. Advanced imaging of the macrostructure and microstructure of

bone. Horm Res 54(Suppl 1):24–30.

Figure 8. Healthy and operated long bone histograms.After the 6thmonth

of implant, the animal was sacrificed and both legs were X-ray acquired. The

histograms of X-ray images showing the healthy long bone and the operated

one were shown (gray and dark respectively). No significant variation in

terms of gray scale values was shown, confirming that no significant bone

density changing due to unloading situations during the experimental period

was displayed.

BELTRAME ET AL.: ASSESSMENT OF BONE REPAIR 197

Goitz HT, Goitz RJ, Watson JT, Schurman JR, Roth HJ. 1996. Orthopedic

implants: A guide to radiographic analysis. Curr Probl Diagn Radiol 25:

113–168.

Guglielmi G, Gluer CC, Majumdar S, Blunt BA, Genant HK. 1995. Current

methods and advances in bone densitometry. Eur Radiol 5(2):129–139.

Haidekker MA, Stevens HY, Frangos JA. 2004. Computerized methods for

X-ray-based small bone densitometry. Comput Methods Programs

Biomed 73:35–42.

Ibanez R. 2003. Bone mineral density measurement techniques. An Sist

Sanit Navar 26(Suppl 3):19–27.

Johnson KD, Frierson KE, Keller TS, Cook C, Scheinberg R, Zerwekh J,

Meyers L, SciadiniMF. 1996. Porous ceramic as bonegraft substitutes in

long bone defects: A biomechanical, histological, and radiographic

analysis. J Orthop Res 14:351–369.

Kon E, Muraglia A, Corsi A, Bianco P, Marcacci M, Martin I, Boyde A,

Ruspantini I, Chistolini P, RoccaM, Giardino R, Cancedda R, Quarto R.

2000. Autologous bone marrow stromal cells loaded onto porous

hydroxyapatite ceramic accelerate bone repair in critical-size defects of

sheep long bones. J Biomed Mater Res 49:328–337.

Leffler SG, Vaccaro JP. 2000. Bone mineral density assessment. Med Health

R I 83(4):123–126.

Maas M, Poll LW, Terk MR. 2002. Imaging and quantifying skeletal

involvement in Gaucher disease. Br J Radiol 75(Suppl 1):A13–A24.

MarcacciM, Kon E, Zaffagnini S, Giardino R, RoccaM, Corsi A, Benvenuti

A, Bianco P, Quarto R, Martin I, Muraglia A, Cancedda R. 1999.

Reconstruction of extensive long-bone defects in sheep using porous

hydroxyapatite sponges. Calcif Tissue Int 64:83–90.

Martin RB, Papamichos T, Dannucci GA. 1990. Linear calibration of

radiographic mineral density using video-digitizing methods. Calcif

Tissue Int 47(2):82–91.

Mazess RB, Barden HS. 1988. Measurement of bone by dual-photon

absorptiometry (DPA) and dual-energy X-ray absorptiometry (DEXA).

Ann Chir Gynaecol 77(5-6):197–203.

PearsonD, Cawte SA, Green DJ. 2002. A comparison of phantoms for cross-

calibration of lumbar spine DXA. Osteoporos Int 13(12):948–954.

Petite H, Viateau V, Bensaid W, Meunier A, de Pollak C, Bourguignon M,

Oudina K, Sedel L, Guillemin G. 2000. Tissue-engineered bone

regeneration. Nat Biotechnol 18:959–963.

Quarto R, Mastrogiacomo M, Cancedda R, Kutepov SM, Mukhachev V,

Lavroukov A, Kon E, Marcacci M. 2001. Repair of large bone defects

with the use of autologous bonemarrow stromal cells.NEngl JMed344:

385–386.

Saiki T, Yumoto E, Okamura H. 1993. Quantitative evaluation of X-ray

opacity of themaxillary sinus using a digital disposition system. Nippon

Jibiinkoka Gakkai Kaiho 96(11):1907–1914.

Sartoris DJ, Resnick D. 1990. Innovative approaches to noninvasive bone

densitometry. Crit Rev Diagn Imaging 30(1):19–39.

Sartoris DJ, Holmes RE, Resnick D. 1992. Coralline hydroxyapatite

bone graft substitutes: Radiographic evaluation. J Foot Surg 31:

301–313.

SeemanE,Martin TJ. 1989.Non-invasive techniques for themeasurement of

bone mineral. Baillieres Clin Endocrinol Metab 3(1):1–33.

Simonini G, Cimaz R, Falcini F. 2005. Usefulness of bone ultrasound

techniques in pediatric rheumatic diseases. J Rheumatol 32(1):198–

199.

Symmons R. 2004. Digital photodensitometry: A reliable and accessible

method for measuring bone density. J Archaeol Sci 31:711–719.

Wakefield RJ, Conaghan PG, Jarrett S, Emery P. 2004. Non invasive

techniques for assessing skeletal changes in inflammatory arthritis:

Imaging technique. Curr Opin Rheumatol 16(4):435–442.

198 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 92, NO. 2, OCTOBER 20, 2005