Quantitative characterization of proximal femur using ultrasound pulse-echo measurements

(1)

Publications of the University of Eastern Finland Dissertations in Forestry and Natural Sciences

Markus Malo

Quantitative Characterization of Proximal Femur Using Pulse-Echo Ultrasound Measurements

Osteoporosis is a common bone disease leading to increased fragility and fracture probability. However, only quarter of individuals suffering from the disease have received a diagnosis. For effective management of the disease it would be highly important to develop diagnostic tools capable of mass screening of the population at the basic level of healthcare. In this thesis quantitative pulse-echo ultrasound technique for evaluation of proximal femur was investigated and developed towards this goal by means of numerical modelling and in vitro and ex vivo measurements.

45 | Markus Malo | Quantitative Characterization of Proximal Femur Using Pulse-Echo Ultrasound...

Markus Malo

Quantitative

Characterization of

Proximal Femur Using

Pulse-Echo Ultrasound

Measurements

(2)

MARKUS MALO

Quantitative

Characterization of Proximal Femur Using Pulse-Echo Ultrasound

Measurements

Publications of the University of Eastern Finland Dissertations in Forestry and Natural Sciences

No 145

Academic Dissertation

To be presented by permission of the Faculty of Science and Forestry for public examination in the Auditorium SN200 in Snellmania Building at the University

of Eastern Finland, Kuopio, on September, 20, 2014, at 9 a.m.

Department of Applied Physics

(3)

Grano Oy Kuopio, 2014

Editor: Prof. Pertti Pasanen, Prof. Pekka Kilpeläinen Prof. Kai Peiponen, Prof. Matti Vornanen

Distribution:

University of Eastern Finland Library / Sales of publications P.O. Box 107, FI-80101 Joensuu, Finland

tel. +385-50-3058396 http://www.uef.fi/kirjasto

ISBN: 978-952-61-1530-6 (printed) ISBN: 978-952-61-1531-3 (PDF)

ISSNL: 1798-5668 ISSN: 1798-5668 ISSN: 1798-5676 (PDF)

(4)

Author’s address: University of Eastern Finland Department of Applied Physics P.O.Box 1627

70211 KUOPIO FINLAND

email: markus.malo@uef.fi Supervisors: Professor Juha Töyräs, Ph.D.

University of Eastern Finland Department of Applied Physics email: juha.toyras@uef.fi Professor Jukka Jurvelin , Ph.D.

University of Eastern Finland Department of Applied Physics email: jukka.jurvelin@uef.fi

Associate Professor Hanna Isaksson, Ph.D.

University of Eastern Finland Department of Applied Physics Lund University

Department of Biomedical Engineering email: hanna.isaksson@bme.lth.se Reviewers: Professor Sulin Cheng, Ph.D.

University of Jyväskylä Jyväskylä

Finland

email: sulin.cheng@jyu.fi

Professor Mami Matsukawa, Ph.D.

Doshisha University

Laboratory of Ultrasonic Electronics Kyotanabe, Kyoto

Japan

email: mmatsuka@mail.doshisha.ac.jp Opponent: Frédéric Padilla, Ph.D.

LabTau Laboratory Inserm Unit Lyon, France

email: frederic.padilla@inserm.fr

(5)

(6)

ABSTRACT

Bone fractures cause suffering and mortality, but also represent a significant economic burden to the society due to lost work input and direct hospital costs. Osteoporosis increases the risk of fracture through reduced bone mass and changes in the bone microstructure. The current gold standard for diagnostics, i.e., dual energy x-ray absorptiometry (DXA), is available only in specialist healthcare units and therefore, it is not used for mass screening. This might explain in part why as many as 75 % of osteoporotic patients lack proper diagnosis and medication. Thus, it is very important to develop diagnostic tools capable of screening the population at the basic level of healthcare. Quantitative ultrasound (QUS) devices for osteoporosis diagnostics have been available for decades.

Their ability to predict fractures is similar to that of DXA. However, since it is used mainly for measurement of the extremities (heel, wrist), ultrasound has provided only a moderate estimate of bone properties at the most important fracture sites. To further enhance the fracture prediction, QUS measurements should be conducted at more sensitive sites,e.g., at proximal femur. This site is covered by soft tissues, which introduces errors into the ultrasound measurement.

In this thesis, changes in the cortical and trabecular bone tissue elastic coefficients and porosities during aging at the proximal femur were assessed by means of ultrasound microscopy (Study I).

Moreover, novel and traditional ultrasound backscatter parameters were measured from intact proximal femur ex vivo and compared with the bone mineral density and trabecular structure (Study II).

In addition, the ability of the dual frequency ultrasound (DFUS) technique to compensate for errors in bone ultrasound measurement due to soft tissue was evaluated. Furthermore, the error in DFUS arising from non-perpendicular ultrasound incidence at soft tissue and soft tissue - bone interfaces was investigated with numerical simulations (Study III). Moreover, the effect of non-optimal focusing to the soft tissue - bone interface on DFUS based correc-

(7)

tion for attenuation originating from the overlying soft tissues was assessed with numerical simulations and in experimental measurements (Study IV).

The cortical bone tissue elastic coefficient and the porosity were found to increase with age (R² = 0.28−^0.46,p < 0.05−^0.01) (Study I). Furthermore, the elastic coefficient was significantly higher (_p<0.05)in cortical bone than in trabecular bone and the value varied between anatomical locations (Study I). The backscatter parameters measuredex vivoat the proximal femur were significantly correlated with bone mineral density (_R² = 0.45,p < 0.01) and trabecular microstructure (_R² = 0.43,p < 0.01) (Study II). Non- optimal focusing of ultrasound to soft tissue - bone interface (Study IV) and non-perpendicular ultrasound incidence at soft tissue and soft tissue - bone interfaces (Study III) were found to induce significant errors in the QUS measurements as well as in the DFUS estimated soft tissue composition. However, in both studies (III and IV), with optimized ultrasound focusing and incidence at interfaces, the error in QUS parameters was significantly reduced by applying information about the interfering layer thickness and composition, as obtained with the DFUS technique.

To conclude, measurement of QUS parameters from proximal femur and minimization of the soft tissue related errors with the DFUS technique are possible and warranted. Since the porosity and elastic coefficient were found to vary with age, it would be highly important to investigate these issues also in osteoporotic bones in order to be able to distinguish between aging and osteoporosis related changes in bone with QUS. The ex vivo measurements indicated that the QUS parameters were dependent on the bone mineral density and trabecular structure of intact proximal femurs. Thus, quantitative ultrasound backscatter measurements, supplemented with DFUS correction for soft tissue induced errors, could enable screening for osteoporosis at the level of basic healthcare. How- ever, to reach this, technical development,e.g., use of phased array technique and extensive in vivotesting are needed.

(8)

National Library of Medicine Classification: QT 34.5, QT 36, WE 200, WE 250, WN 208

Medical Subject Headings: Bone and Bones; Bone Density; Femur;

Hip Fractures; Osteoporotic Fractures; Osteoporosis/diagnosis; Ag- ing; Elastic Tissue; Biomechanical Phenomena; Elasticity; Porosity;

Absorptiometry, Photon; Microscopy, Acoustic; Numerical Anal- ysis, Computer-Assisted; Computer Simulation; Ultrasonography;

Ultrasonics

Luokitus: QT 34.5, QT 36, WE 200, WE 250, WN 208

Yleinen suomalainen asiasanasto:luu; luuntiheys; reisiluu; osteoporoosi - - diagnoosi; ultraääni; ultraäänitutkimus; kimmoisuus; huokoisuus;

akustinen mikroskopia; ikääntyminen; röntgentutkimus; fotoniab- sorptiotekniikka; simulointi; numeeriset menetelmät

(9)

To Laura, Luukas and Eevi

(10)

Acknowledgements

What will remain from us?

Hopefully children’s and warm memories to our loved ones.

Possibly a lot of unfinished work.

At the very least - 206 bones.

This study was carried out during the years 2009-2014 in the De- partment of Applied Physics at the University of Eastern Finland.

I would like to express my gratitude to my supervisors for their professional guidance during this thesis project. I would like to thank my principal supervisor Juha Töyräs for the endless enthu- siasm towards research work and for the discussions we have had both about and out with the research topics. Moreover, I am grateful to my second supervisor Professor Jukka Jurvelin for the oppor- tunity to work in his top class research group, Biophysics of Bone and Cartilage (BBC). I also want to thank my third supervisor As- sociate Professor Hanna Isaksson for the tough questions and the push to do critical thinking in research, but also for the understanding and soft values during the different phases of life during this thesis.

I am grateful to the reviewers of this thesis, Professors Sulin Cheng and Mami Matsukawa, for their professional review and en- couraging comments. I would also like to thank Ewen MacDonald for linguistic review.

I would like to express my deepest gratitude to all of my co- authors for their significant contributions to the studies. Partic- ularly, I want to thank Janne Karjalainen, Sami Väänänen, Jukka Liukkonen, Katariina Nissinen, Mikael Turunen, Daniel Rohrbach, Kay Raum, Heikki Kröger, Xioyu Tong, Inari Tamminen, Ossi Riekki- nen, Antti Aula, Mikko Nissi, Erna Kaleva, Jari Rautiainen, Matti Ti- monen, Juuso Honkanen, Tuomo Silvast, Simo Saarakkala, Tuomas Viren, Harri Kokkonen, Roope Lasanen, Chibuzor Eneh, Xiaowei

(11)

Ojanen, Cristina Florea and Viktoria Prantner for the fruitful discussions and their collaboration. Naturally I want to thank every- one working under the BBC group umbrella. It has been a pleasure and a privilege to work in such a stimulating atmosphere. More- over, I want to thank people from SIB labs, Arto Koistinen, Ritva Savolainen and Juhani Hakala, for the help and guidance in the sample preparation and the personnel from the Department of Ap- plied Physics, Jukka Laakkonen, Aimo Tiihonen, Tarja Holopainen and Heikki Väisänen for all kinds of technical support.

I also want to express my gratitude to all my relatives and friends for giving the support and something else to think outside of the work during the past years.

For financial support the strategic funding of University of East- ern Finland, Kuopio University Hospital (EVO grants), Kuopio Uni- versity foundation, Emil Aaltonen foundation, the Finnish Founda- tion for Technology Promotion and National Doctoral Programme of Musculoskeletal Disorders and Biomaterials (TBDP) are acknowl- edged.

Finally, I am grateful to my parents, Irja and Ilkka, and my brother Petri and his family for their continuous support, encouragement and love throughout my life. I owe my deepest gratitude to my beloved wife Laura and our two miracles Luukas and Eevi.

Thank you Laura for your endless love, encouragement, support and understanding during all these years.

Kuopio, August 2014 Markus Malo

(12)

ABBREVIATIONS

3D three-dimensional A/D analog to digital

AIB apparent integrated backscatter ANOVA analysis of variance

BMD bone mineral density

BUA broadband ultrasound attenuation CT X-ray computed tomography CV coefficient of variation DFUS dual-frequency ultrasound DOF depth of field

DXA dual energy X-ray absorption FDTD finite difference time domain FemUS femur ultrasound scanner

FSAB frequency slope of apparent backscatter FWHM full width at half maximum

IRC integrated reflection coefficient MBD mean of the backscatter difference

NAHNES National Health and Nutrition Examination Survey PBS phosphate buffered saline

PDE partial differential equations

PE pulse-echo

PMMA polymethylmethacrylate PZT lead zirconate titanate QUS quantitative ultrasound

RANK Receptor Activator for Nuclear Factor k B RMS root mean square

SAM scanning acoustic microscopy SD standard deviation

SOS speed of sound TOF time of flight TPX polymethylpentene

TSAB time slope of apparent integrated backscatter TT through-transmission

(13)

WHO World Health Organization µCT micro computed tomography

(14)

SYMBOLS AND NOTATIONS AA average attenuation

A amplitude

A(f) amplitude spectrum

Ar area

α attenuation coefficient

b medium dependent coefficient c speed of sound

cp phase velocity

cl longitudinal sound velocity cs shear sound velocity

cm speed of sound in medium cw speed of sound in water

Cii elastic stiffness coefficient to direction ii d distance

Di diameter of transducer E Young’s modulus f frequency

H the term including ultrasound reflections from different surfaces

I acoustic intensity

ki correction factor for compensation of ultrasound reflection at the adipose-lean interface

K bulk modulus λw wavelength k wavenumber Z acoustic impedance ρ mass density ν Poisson’s ratio

p sound wave pressure or statistical difference s tissue interface

vu particle velocity

(15)

θ angle

R correlation coefficient

RC reflection coefficient (intensity) TR transmission coefficient (intensity)

d distance

t time

m medium dependent coefficient

σ stress

ϵ strain

λ first Lamé constant µ second Lamé constant

F force or value of F-test (statistics)

l length

Fl focal length

R_l radius of lens curvature DOF depth of field

x thickness

c speed of sound n number of samples SMI structural model index Tb.N trabecular number Tb.Sp trabecular separation Tb.Th trabecular thickness

∆f(_x) forward difference form

∇f(_x) backward difference form δf(_x) central difference form

∂ partial difference operator

∇ gradient operator

∇· divergence operator u particle displacement

ω angular frequency of the wave ϕ phase angle

(16)

LIST OF PUBLICATIONS

This thesis is based on the following original articles referred to by their Roman numerals:

I M. K. H. Malo, D. Rohrbach, H. Isaksson, J. Töyräs, J. S. Jurvelin, I. S. Tamminen, H. Kröger, K. Raum, “Longi- tudinal elastic properties and porosity of cortical bone tissue vary with age in human proximal femur,” Bone 53(2), 451-8 (2013).

II M. K. H. Malo, J. Töyräs, J. P. Karjalainen, H. Isaksson, O. Riekkinen and J.S. Jurvelin, “Ultrasound backscatter measurements of intact human proximal femurs – relationships of ultrasound parameters with tissue structure and mineral density,”Bone64, 240-5 (2014).

III M. K. H. Malo, J. P. Karjalainen, H. Isaksson, O. Riekkinen, J. S. Jurvelin, J. Töyräs, “Numerical analysis of uncertainties in dual frequency bone ultrasound technique,” Ultrasound in Medicine and Biology36(2), 288-94 (2010).

IV M. K. H. Malo, J. P. Karjalainen, O. Riekkinen, H. Isaksson, J. S. Jurvelin, J. Töyräs, “Effects of non-optimal focusing on dual-frequency ultrasound measurements of bone,”IEEE Trans- actions on Ultrasonics, Ferroelectrics, and Frequency Control58(6), 1182-8 (2011).

The original articles have been reproduced with kind permission of the copyright holders.

(17)

(18)

AUTHOR’S CONTRIBUTION

The publications selected to this dissertation are original research papers on bone ultrasound measurements and experimental and numerical evaluation of the error sources with the dual frequency ultrasound technique. The author has contributed to the study de- sign and carried out all measurements and analyses, except for part of the micro-computed tomography imaging in study II. The author has written the manuscripts of studies I-IV. In all papers the co-operation with the co-authors has been significant.

(19)

(20)

1 Introduction

The number of elderly individuals is growing as the average life ex- pectancy increases. There are several age-related changes occurring in the human body, for example, the human skeleton goes through various developmental and degenerative phases. The skeleton reach- es full maturity and peak bone mass at the age of 25. Thereafter, skeletal degeneration begins,i.e., there is a reduction in bone mass and a deterioration in the bone quality [1–8]. The term bone quality reflects many factors, e.g., bone architecture, turnover, damage accumulation (e.g., microfractures) and mineralization [9].

These changes reduce the mechanical competence of the whole bone. Thus, with age the fracture probability increases [10]. Bone fractures not only lead to increased mortality rate but they are also responsible for cause significant financial expenditures via both direct hospital costs and in the form of lost work capacity [10, 11, 11–

13]. Thus, prevention of fractures is highly important for both the individual and society as a whole.

Osteoporosis is the most common skeletal disease in the elderly.

In global terms, it is estimated that about 200 million people have osteoporosis, 27.5 million of these live in Europe [13, 14]. Osteo- porosis is a systemic skeletal disease characterized by decreased bone mass and deterioration of the bone microstructure. It results in increased bone fragility and on elevated fracture risk [15,16]. The intrinsic properties of bone, e.g., bone material properties, shape and architecture, define the capability of bone to resist fractur- ing [17–19]. However, the fracture risk is also affected by many external factors which can influence the susceptibility to fall, e.g., environmental conditions, individual lifestyle, vision, balance, re- action time and muscle strength [20, 21]. Therefore, the accurate estimation of the fracture risk is a challenging problem.

The gold standard for osteoporosis diagnosis is the measurement of bone mineral density (BMD) with dual energy absorptiom-

(23)

etry (DXA) [16]. The DXA devices used for the osteoporosis diagnostics are generally available only in specialized healthcare centers and are not used for mass screening of the population [22]. This might explain in part why as many as 75 % of the osteoporotic patients are not properly diagnosed and thus fail to receive appro- priate therapy [23]. Thus, it would be very important if one could develop diagnostic tools capable of screening the population at the level of basic healthcare.

Quantitative ultrasound (QUS) has been proposed as being a non-ionizing option for screening for osteoporosis [24–26]. QUS is sensitive to bone elastic properties, density, structure and also it can estimate the inorganic phase that cannot be evaluated by X-ray methods [27]. Although peripheral ultrasound measurement devices have been on the market for decades, they have not achieved any major clinical breakthroughs [28]. There are several reasons (e.g. measurement of extremities) why ultrasound seems to obtain only a moderate estimation of bone properties at the axial skeleton [24, 29]. If one wishes to enhance the fracture prediction at the most serious fracture sites, e.g., the proximal femur, site spe- cific measurements are needed [30–32]. However, bones in the axial skeleton are covered with a substantial soft tissue layer and its thickness and composition vary from patient to patient introducing errors into QUS measurements.

The speed of sound in bone depends on its physical density and elastic properties. For example, when evaluating the cortical bone thickness, it is important to know the age dependent variation of factors affecting the speed of sound. With aging, the trabecular microstructure within the bone undergoes changes. Common find- ings include a reduction in the trabecular network connectivity and the number of trabeculaes, but also thinning and a change in shape of single trabeculae. Further, these changes may be measured in vivo by dual energy absorptiometry as a decrease in the areal bone mineral density. However, the potential of ultrasound backscatter to detect these changes has not been investigated in intact proximal femurex vivo. It would be very important to determine whether the

(24)

trabecular microstructure and bone mineral density can be evaluated at the most serious fracture site, i.e., proximal femur with ul- trasound backscatter measurements before undertaking extensive in vivoevaluation of this technique.

A dual frequency ultrasound (DFUS) technique has been developed for the measurement of soft tissue thickness and composition [33]. The layered and non-homogenous structure of soft tissue causes attenuation and scattering thus distorting the ultrasound signal measured from bone. If one incorporates information about the soft tissue covering the bones, the error caused by the soft tissue in the bone QUS parameters may be minimized. This technique has been tested in vitro and applied in vivo [33, 34]. However, its sensitivity to the error sources that are present in vivo have not been fully characterized. Since bones are often located under a thick layer of soft tissue, focused ultrasound transducers need to be applied to maximize the signal to noise ratio. Unfortunately, due to individual variation in soft tissue thickness, it is not always possible to conduct the measurement at the optimal focal distance from the bone surface. Moreover, non-perpendicular ultrasound incidence at tissue interfaces cause distortion and refraction to the propagating ultrasound pulse. The non-perpendicular ultrasound incidence at tissue interfaces and non-optimal focusing may introduce errors when analyzing the thickness and composition of the soft tissue with the DFUS method. Thus, clarification and the elimination of these sources of error would be very advantageous.

The present study aims to fill these gaps in our knowledge.

The changes in the porosity and elastic coefficient of cross-sectional bone samples obtained from the femoral neck and shaft have been evaluated using scanning acoustic microscopy (SAM). Furthermore, the sensitivity of ultrasound backscatter in the estimation of bone mineral density and trabecular structure has been determined in the proximal femurex vivo. Finally, numerical simulations and experimental measurements have been applied to examine the error sources related to application of the DFUS method.

(25)

(26)

2 Bone

2.1 SKELETON

The human skeleton can be divided into the axial skeleton, including the vertebrae and pelvis, and the appendicular skeleton, including all the long bones. The skeleton consists of four types of bones:

long bones, e.g., tibia and radius, short bones, e.g., phalanges, flat bones, e.g., ribs, and irregular bones, e.g., vertebrae. Bones pro- vide mechanical support for the body and enable locomotion. They function as levers and transform the forces from the muscles into movements. Bones protect our vital organs,e.g., the brain, but they also store and release minerals such as calcium and phosphorus.

Moreover, the inside of bones is the location of red or yellow bone marrow. Red marrow produces red and white blood cells, and platelets, hence it has a key role in hematology and the immune system. Yellow marrow primarily acts as a fat storage site [35–37].

Skeletal bone mass increases rapidly during adolescence to reach its peak value when the individual is around 25 years of age [1, 3].

Thereafter, the bone mass starts to slowly decline. In women, after the menopause, the reduction in bone mass occurs rapidly [4, 38].

2.2 STRUCTURE AND COMPOSITION Structure

Bone tissue can be divided into cortical and trabecular bone based on its structure at the macroscopic level. The division can also be based on the maturity of the bone tissue. Woven and lamellar bone represent newly developed and mature bone, respectively. In woven bone, the collagen fibril network is randomly organized and the osteocyte and water contents are higher than in fully mature bone [39, 40]. Woven bone undergoes a rapid rate of deposition and turnover, and its mineralization pattern is irregular. During maturation of lamellar bone, its structure including the collagen

(27)

fibers and hydroxyapatite crystals, becomes organized. Lamellar bone exists both in trabecular and cortical bone. Lamellar bone is stiffer than woven bone, due to its highly organized structure. Fur- thermore, the mechanical properties of woven bone are isotropic whereas lamellar bone is mechanically anisotropic [35]. According toWolff’s Law, the bone adapts its internal architecture and external shape to resist forces in the main mechanical loading direction [41].

A schematic illustration of bone structure is presented in Figure 2.1.

Cortical bone

Cortical bone is compact (approximately 5 to 10 % porosity) and dense (1600-2000 kg/m³) and it forms the outer layer of the bones [42, 43]. The human skeleton is mainly (80 - 90 %) composed of cortical bone [35, 44, 45]. The cortex is thick especially in the dia- physis of the long bones, providing maximum resistance to torsion and bending. In the epiphyisis, the thin cortex is supported by the underlying trabecular structure enabling high deformation during loading. Cortical bone is formed of packed osteons and interstitial tissue. The osteons are connected by the Haversian system [36, 37].

In comparison to the porous trabecular bone, cortical bone has a slow turnover rate and lower metabolism.

(28)

Figure 2.1: A schematic illustration of cortical and trabecular bone structure, showing the lamellar structure, Haversian system, peri-, and endosteum. (Figure is modified from [35]).

Trabecular bone

Trabecular bone consists of a network of trabeculae and it is highly porous. The typical diameter of a trabeculae is 100 to 200 micrometers [15]. The volumetric density of trabecular bone is much lower than that of cortical bone. Thus, the surface area of the trabecular bone is about twenty times higher than that of cortical bone of a

(29)

similar mass. However, the density of the calcified matrix is approximately the same in cortical and trabecular bone. Trabecular bone is present particularly in the vertebrae, and at the ends of the long bones. The volume between the trabeculae is filled with bone marrow. Since the cells in the trabecular bone are located on the surface of the trabeculae, they are in proximity or in direct con- tact with the bone marrow, blood vessels and cells. This together with high surface-to-volume area facilitates a high rate of metabolic activity and good sensitivity to remodeling under mechanical loading [4, 35].

Composition

Bone is a composite material consisting of an organic and an inorganic phase. Approximately 70 % of the weight of the bone is mineralized material, i.e., crystalline hydroxyapatite which makes the bone hard and stiff, and enables it to resist compression. Be- tween 5 % to 8 % of the weight of the bone is water, with the remaining part being organic material consisting mostly of collagen type I (90 %) and a variety of non-collagenous proteins [46].

The organic component provides tensile strength and elasticity to the bone and enables the bone to deform and resist stretching and twisting [47]. The bones store the majority of the minerals in human, i.e., they represent a store from which calcium, phosphorus, sodium and magnesium ions can be released into the extracellular fluid [35].

2.3 REMODELING

The skeleton undergoes remodeling throughout the lifespan. In fact, most of the skeletal system is replaced approximately every 10 years [48]. This remodeling is achieved by resorption of bone matrix by osteoclasts and its replacement by osteoblasts [49].

Bone resorption by osteoclasts

Osteoclasts dissolve the old bone by secreting acids and enzymes.

(30)

Osteoclasts are formed from the same precursor cells as the white blood cells [48]. Osteoclasts become activated when Receptor Ac- tivator for Nuclear Factorκ B (RANK) receptors (in osteoclast pre- cursors) are stimulated by RANK ligand (secreted by osteoblasts).

Furthermore, osteoprotegerin binds RANK ligand and in this way it regulates osteoclast activation [50]. When an osteoclast has finished bone resorption, it undergoes apoptosis.

Bone formation by osteoblasts

When old bone has been removed by osteoclasts, the empty space is replaced with new bone produced by osteoblasts. Osteoblasts are formed from marrow precursor cells, which are also capable of differentiating into fat cells. The osteoblasts produce proteins that form the organic matrix and control the mineralization. Osteoblast receptors are sensitive to several hormones,e.g., estrogen, parathy- roid hormone and vitamin D. Moreover, they communicate with other cells,e.g., osteoblasts secreting RANK ligand [35, 48].

Osteocytes and lining cells

When osteoblasts have finished forming new bone, a part of them become trapped in the matrix where they differentiate into osteocytes. The cells remaining at the surface of the matrix differentiate into bone lining cells while the remaining osteoblasts undergo apoptosis. Osteocytes are arranged circularly around osteon and are connected through canaliculi which have diameters of a few micrometers [51] (Figure 2.1). They are able to signal and activate both lining cells and osteoblasts. Lining cells are flat shaped cells that cover the bone surfaces. The lining cells possess receptors for hormones and growth factors and may become activated to initi- ate bone remodeling when necessary [52]. The remodeling process complies withWollf’s Law and, thus, the bone adapts to the domi- nant direction of mechanical loading.

(31)

2.4 AGING AND OSTEOPOROSIS

During adolescence, the activity of osteoblasts is greater than that of the osteoclasts, but with increasing age, the osteoblast activity becomes reduced. This may lead to a deterioration of the bone microarchitecture, i.e., an increase in porosity and a thinning of cortex and trabeculae, and changes in the bone material properties [4, 8, 53–56]. Thus, with aging, the bone structure weakens, which reduces the mechanical strength of the bone and increases the fracture risk [57, 58].

Figure 2.2: Two-dimensionalµCT section of normal and osteoporotic trabecular bone. The white color illustrates the trabecular bone structure of a cylindrical bone sample extracted from a human proximal tibia.

Osteoporosis means - porous bone in latin. The osteoporotic changes in the skeleton appear as a reduction in the bone mineral density in the whole skeleton,i.e., a deterioration in the bone macro- , and microarchitecture, such as an, increase in cortical porosity, a thinning of the trabeculae and cortices, and a decreased number of trabeculaes (Figure 2.2) [59]. These changes increase the risk of fracture since they weaken the structure of the remaining bone [4, 31, 32, 60, 61]. As the trabecular bone has a large surface to volume ratio, it is a sensitive reflection of the appearance of osteoporosis. There may be various reasons for the skeletal degeneration, but the overall result is that the osteoclasts remove more bone than

(32)

the osteoblasts can produce. Approximately 40 % of women and 15 % of men over 50 years of age will suffer one or more fragility fractures during their remaining lifetime [38]. On average, the an- nual declines from the peak bone mass in areal bone mineral density at the femoral neck are 0.4 percent for men and 0.5 percent for women [62].

Diagnostics of osteoporosis

The diagnosis of osteoporosis is based on DXA measurement of the bone mineral density in the proximal femur or spine [16]. Accord- ing to the World Health Organization (WHO) and the International Osteoporosis Foundation recommendations, osteoporosis diagnosis is based on the T-score [63,64]. The T-score is defined as the number of standard deviations (SD) above or below the mean bone mineral density of a young adult reference population. The age-adjusted relative increase in fracture risk is approximately doubled for each SD decrease in BMD [31, 32, 61]. A hip BMD value greater than 1 SD below the young adult reference population mean (T-score ≥ –1) is considered as normal. A hip BMD value between 1 SD and 2.5 SD below the young adult reference population mean (T-score <

–1 and > –2.5) is referred to as being osteopenic,i.e. low bone mass.

A hip BMD value of 2.5 SD or more below the young adult reference population mean (T-score ≤ –2.5) is diagnosed as osteoporosis, and in addition if there have been one or more fragility fractures, then the diagnosis is severe osteoporosis,i.e., established os- teoporosis [16]. Currently the recommended reference database is the National Health and Nutrition Examination Survey (NHANES) III which contains femoral neck measurements from 20-29 year old women. Although there has been discussion and conflicting reports about whether the same cut-off values for osteoporosis may be applied for the male population, the present recommendation is to use the same database and criteria for men and women [65].

Screening of osteoporosis with DXA in population is not recommended and instead, screening should be targeted to those patients belonging to a risk group [22, 61, 66]. In order to identify these

(33)

patients, a Fracture Risk Assessment Tool FRAX) has been developed by WHO [67]. The FRAX tool is based on individual patient models that integrate the risks associated with several clinical factors. The FRAX tool can be used with the national osteoporosis guidelines by which patients can be categorized into three treatment groups: lifestyle advice and reassure, measure BMD and treat.

These thresholds are provided only to support clinicians when con- sidering which patients would benefit from further treatment.

The best prediction of fracture seems to be obtained by site specific measurements, e.g., the best prediction for hip fracture is ob- tained by the DXA measurements of BMD from the proximal femur.

Measurement of BMD with DXA is quite reproducible (coefficient of variation 2 %). This is very important when evaluating the ef- fectiveness of the medical treatment or conducting follow-up studies [68]. Although DXA is a reproducible tool for measuring the areal BMD, it only provides indirect information on the bone mechanical properties and gives no information about the microarchitecture or the organic composition of the bone. Moreover, changes in the composition of the overlying soft tissue affect the determined values of BMD implying that the method for accounting for the soft tissue related factors is not optimal [69]. Furthermore, it has been claimed that the variation in bone marrow composition may introduce error to the value of BMD measured [70].

(34)

3 X-ray assessment of bone

Since the discovery of X-rays in 1895, X-ray imaging has been widely used in clinical diagnostics. Plain X-ray imaging can be used to detect bone fractures and subchondral sclerosis related to osteoarthri- tis, and the dual energy X-ray absorptiometry (DXA) is used for the diagnostics of osteoporosis.

3.1 DUAL-ENERGY X-RAY ABSORPTIOMETRY

Dual-energy X-ray absorptiometry (DXA) measures the areal bone mineral density (aBMD, g/cm²). There are two main types of DXA devices. There are peripheral devices which can be used for the measurement of the heel or the radius, and there are whole body DXA scanners, which are more commonly used. In whole body scanners, the X-ray source, the collimator and the detector are im- plemented into a C-arm. The DXA devices are based either on pencil or fan beam technology. The ’Dual Energy’ refers to fact that X-ray radiation with two different energy levels, such as, 38 and 70 keV is being used [71]. As the patient is scanned with two different energy levels, two different attenuation profiles are obtained. The areal BMD,e.g., from the proximal femur or lumber vertebra, can be evaluated by determining the soft tissue thickness and composition adjacent to the bone and compensating for its effect in the bone measurement. The typical regions of interest in the proximal femur are illustrated in Figure 3.1.

(35)

Figure 3.1: A screenshot from BMD measured by DXA in the proximal femur ex vivo.

The left side of the figure shows the analyzed regions and the shape of the femur. The right side of the figure shows an illustration of the average BMD in females as a function of age (cyan represents±1 SD limits) in which the black dot corresponds to the measured femoral neck BMD value.

The radiation dose obtained from a whole body DXA examination is small (4 - 30µSv), corresponding to approximately four daily amount of natural radiation [71, 72]. The duration for a total body scan with a fan and pencil beam technology is less than 10 and 20 minutes, respectively [71], whereas a hip measurement only takes less than a minute.

3.2 COMPUTED TOMOGRAPHY

Computed topography (CT) is based on X-ray attenuation projec- tion images of an object from multiple angles of rotation (rotation of at least 180 degrees). These images are further processed by math- ematical algorithms to provide a three dimensional (3D) graphic reconstruction of the imaged object. With clinical CT devices, the typical isotropic voxel size in the images varies from 0.5 x 0.5 x

(36)

0.5 mm³ to 5 x 5 x 5 mm³. By increasing the radiation dose, the isotropic voxel size can be decreased to 300 x 300 x 300µm³. With a dedicated peripheral CT one can achieve, an isotropic resolution of 82 x 82 x 82 µm³ (Scanco XtremeCT) [73]. However, the imaging time is approximately 3 minutes per centimeter. With the latest developments in cone beam techniques, the new peripheral CT device (Planmed Verity) is capable of capturing a 130 mm x 160 mm field of view with a 200 x 200 x 200 µm³ isotropic voxel size in 17 seconds [74]. For comparison, the isotropic resolution with a mod- ern desktopµCT is around 1 - 200 µm³. The field of view in µCT devices is limited to the millimeter scale and, thus, they are capable of imaging only small animals andin vitrosamples. Furthermore, the resolution comes with the cost of increased radiation dose and prolonged imaging time. Thus, it can take from tens of minutes to hours if one wishes to obtain high resolution images with a µCT.

However, due to the nature of the objects being imaged, this is usually not a problem. In order to determine trabecular structure reliably, a resolution of at least 28 µm is desirable [75]. At the mo- ment, there is no clinical device capable of reliably estimating the trabecular microstructurein vivo.

The high resolution 3D CT-images can be utilized to calculate many geometrical parameters describing the imaged structure. Typ- ically from cortical bone, its thickness and volumetric porosity are evaluated, whereas from trabecular bone other parameters, e.g., bone volume to total volume (BV/TV) and trabecular thickness, shape and connectivity are evaluated [6, 8, 55, 76, 77]. While DXA provides information on areal BMD, CT determines the volumetric bone mineral density (vBMD) and the 3D geometry of the bone [72].

This information is extremely valuable since it has been shown to provide a more reliable estimation of the fracture risk than BMD measurements made by DXA alone [19, 78, 79]. However, the radiation dose utilized in a CT measurement may be a hundred times higher than that needed for DXA measurements, which limits the former’s use in screening for osteoporosis. It has been proposed that DXA measurement alone would allow estimation of the bone’s

(37)

3D shape with the BMD value and some index of its mechanical strength [80]. However, this approach requires patient specific finite element models, but would provide information on the fracture load and what factors affect the fracture susceptibility [81].

10mm

Figure 3.2: A three-dimensional reconstruction of a cubioid bone sample harvested from the trochanter of a male cadaver. In the sample, dense cortical bone lies over the porous trabecular bone. The isotropic voxel size of theµCT imaging was 34 x 34 x 34µm³.

(38)

4 Ultrasound assessment of bone

Quantitative ultrasound (QUS) techniques have been used for the assessment of bone for decades [28]. The ability of QUS to predict fractures is approximately the same as can be achieved with DXA [82–85]. Ultrasound is sensitive to many bone - related properties, i.e., density, structure, composition and mechanical proper- ties, and it provides also information on bone organic phase [86–91].

However, the interaction between the bone matrix and ultrasound wave field is complex and, therefore, it is challenging to relate ultrasound parameter values to bone properties. Thus, for decades major efforts have been made to gain a comprehensive understanding of the issues related to bone ultrasound measurements [28, 92].

QUS does hold the potential to become an alternative to X-ray based techniques for mass screening of the population to identify those individuals with an elevated risk of suffering fractures [25, 84, 85, 93].

Moreover, ultrasound instrumentation can be manufactured to have a small size and the devices themselves are inexpensive. However, currently the clinical use of ultrasound is limited. This is partly because the diagnostic criteria for osteoporosis, as defined by the World Health Organization (WHO), is based on areal bone mineral density obtained by DXA [63, 64].

4.1 BASIC PHYSICS OF ULTRASOUND

Ultrasound is defined as a propagating mechanical wave with a frequency higher than can be detected by the human ear (20 kHz).

The wave propagation is based on displacement of medium particles from their resting positions which induces displacements of the neighboring particles. When the particle becomes displaced from

(39)

the equilibrium position, the restoration forces together with the inertia of the particles results in oscillation. The fundamental equations describing wave propagation are presented in Table 4.1. In the present thesis, only longitudinal waves were investigated. However, there are also several other wave modes,e.g., transverse (shear) and surface (Rayleigh) waves. The longitudinal wave can propagate in all types of material (solids, liquids and gases), since the energy is transferred through compression and expansion of the medium. In transverse waves, the displacement of particles occurs in the direction perpendicular to the direction of the wave propagation. The surface wave is a combination of the longitudinal and transverse waves, resulting in an elliptic orbit in particle displacement [94].

4.2 ACOUSTIC PROPERTIES OF TISSUES

All materials have characteristic acoustic properties, which may be defined by the acoustic impedance and attenuation coefficient.

Acoustic impedance is dependent of the volumetric mass density of the material and the speed of sound. Speed of sound is dependent on Young’s modulus, Poisson’s ratio and volumetric mass density, which are temperature dependent (Table 4.2). The acoustic properties of bone tissue depend on its collagen and mineral contents [88]. Moreover, the bone structure affects the acoustic properties. In highly porous materials,e.g., trabecular bone, a slow and fast wave can be detected [98–104]. The fast wave is considered to result from the displacement component in-phase in the mineralized tissue and marrow, whereas the slow wave is a result of the components oscillating out of phase [99]. Furthermore, for dis- persive materials, both group and phase velocities may be determined [105–109]. Phase velocity is defined individually for each frequency component whereas the group velocity is described for the envelope of the wave [110]. Attenuation quantifies the loss of energy as the wave propagates through the medium. Attenua- tion results mainly from scattering, reflection, absorption and beam

(40)

spreading [95]. These phenomena will be elaborated in the following sections.

(41)

Table 4.1: Basic equations describing the propagation and behavior of planar sound waves in medium and at the interface of mediums [27, 95–97]

Parameter Equation

Particle displacement [m] u=_u₀_sin(_ωt−ϕ) Angular frequency of particle [rad/s] ω=_2π_f

Wavelength [m] λ_w= ^c^p

f =_c_p_T

Wavenumber [m^-1] k= ^2π_λ

w = _c^ω

p

Acoustic impedance [rayl] Z=_ρc_l

Longitudinal sound velocity in isotropic elastic solid [m/s]

c_l=

√ E(1−ν)

ρ(1+ν)(1−2ν)

Longitudinal sound velocity in fluids [m/s] c_l=

√Ka

ρ

Shear sound velocity in isotropic elastic solid [m/s]

cs=

√ E ρ(1+ν)

Acoustic intensity [W/m²] I=_2Z^p² Sound wave pressure for plain waves [Pa] p=_ρc_l_c_u

Snell’s law [-] ^sinθ_c ¹

1 = ^sinθ_c ²

2 =^sinθ_c ³

3

Reflection coefficient (intensity) [-] R_C= (^Z_Z²^cosθ¹^−Z¹^cosθ²

2cosθ1+Z1cosθ2)² Transmission coefficient (intensity) [-] T_C=_(Z ^4Z¹^Z²^cos²^θ¹

2cosθ1+Z1cosθ2)²

Attenuation law [-] p(_d) =_p₀_e^−αd_andI=_I₀_e^−2αd Attenuation coefficient [dB/cm] α=_{b f}^m

f = frequency,cp = phase velocity,T= period,ϕ= phase angle,E= Young’s modulus,ν= Poisson’s ratio,ρ= density,Ka= adiabatic bulk modulus,cu= particle velocity,d= distance,t= time,b= medium dependent coefficient and m= medium dependent coefficient. θ1andθ2are the angles of the incidence and refraction, respectively. Subscripts 1, 2, and 3 refer to the first and second medium, and shear wave, and l and s to the longitudinal and shear sound velocity.

(42)

Table 4.2: Material properties can be described with the following equations [27, 111, 112]

Elastic stiffness coefficient (Pa) (orthotropic) C_ii=_ρc²_li

Young’s modulus (Pa) E= ^σ

ϵ = ^µ(_3λ+_2µ) λ+_µ

Bulk modulus (Pa) K= ^E

3(₁−2ν) =_λ+² 3µ

Poisson’s ratio (-) ν=−_ϵ ^ϵ^lateral

longitudinal

= ^λ

2(_λ+_µ)

Stress (Pa) σ= ^F

Ar

Strain (-) ϵ= ^∆l

l₀

First Lamé constant (Pa) λ= ^Eν

(₁+_ν)(₁−2ν) Second Lamé constant (shear modulus) (Pa) µ= ^E

2(₁+_ν)

C= Elastic stiffness coefficient,ρ = mass density, c= speed of sound,E = Young’s modulus,σ= stress,ϵ= strain,λ= First Lamé constant,µ= Second Lamé constant,K= Bulk modulus,ν= Poisson’s ratio,F= force,Ar= area,

∆l= change is the length andl0= original length. Subscriptslandirefer to longitudinal and in the direction of orthogonal symmetry, respectively.

4.3 ULTRASOUND SCATTERING AND REFLECTION

Scattering arises when the proceeding wave encounters inhomo- geneity in the medium density or elastic properties,i.e., a scatterer [96]. Furthermore, roughness of the interfaces induces scattering.

The wavelength and the dimensions of the scatterer determine the scattering phenomenon. If the scatterers are much smaller than the wavelength, the phenomenon is called Rayleigh scattering. When

(43)

the scatterer is smaller than the wavelength, the scattering pattern can be described as monopole radiation or dipole radiation in the case of changes in the elastic properties or volumetric mass density, respectively. When the wavelength and the size of the scatterer are similar, the scattering pattern becomes more complex [113]. In the case of trabecular bone, the scattering has been estimated by Faran scattering and weak scattering models [113, 114]. Typically in Faran solution, the scattering is evaluated by investigating theka where a characterizes the radius of the spherical scatterer andk is the wavenumber (Table 4.1). As the ultrasound pulse encounters an interface of two media layers having different acoustical impedance values (Z) with one layer being thicker than the wavelength, then reflection and transmission phenomena take place (Table 4.1). The greater the difference in the Z values of the materials, the greater is the reflection coefficient (RC) and the smaller is the transmission coefficient (T_C). The reflection and transmission coefficients are also dependent on the angle of incidence according to Snell’s law (Table 4.1). As the particle bindings in soft tissues are weak, only longitudinal waves can propagate. Thus, the angle of the incident wave is equal to the angle of the reflected wave and the wave proceeding through the interface becomes refracted. However, when there is an interface between soft tissue and bone (which has strong particle bindings), also shear waves are encountered (Figure 4.1.).

(44)

Incident longitudinal wave Reflected longitudinal wave

Refracted longitudinal wave

2 1 1

Liquid Liquid

a)

Incident longitudinal wave Reflected longitudinal wave

Refracted longitudinal wave

3 1 1

2

Refracted shear wave Liquid

Solid

b)

Incident longitudinal wave

Leaky surface wave

Liquid Solid

1 1

Reflected longitudinal waves

c)

Figure 4.1: Snell’s law. a) Refraction and reflection of incident longitudinal wave in a liquid-liquid interface. b) Refraction and reflection of incident longitudinal wave in liquid- solid interface. c) An illustration of leaky surface wave formation.

When the angle of incidence is increased to, or it is greater than the critical angle, the wave will not propagate through the interface.

Instead, total reflection takes place. Thus, an leaky surface wave is formed that will propagate in parallel to the surface of the interface and will reflect longitudinal waves back into the soft-tissue (Figure 4.1.). This phenomenon can be measured by an axial transmission technique enabling the evaluation of cortical bone propertiesin vivo [115, 116].

4.4 ULTRASOUND ABSORPTION

Absorption is dependent on the frequency content of the sound wave and density, viscosity and temperature of the media (Table 4.1). The absorbed energy of the acoustic pulse typically is con-

(45)

verted into heat. The attenuation coefficient is the sum of absorption, scattering, reflection and beam spreading, which are also closely related to the structure and viscous properties of the media. The ex- ponential decrease in the ultrasound energy along the propagation can be quantified by the attenuation coefficient (Table 4.1). In acous- tically isotropic materials, the acoustic properties are similar in all directions. However, biological tissues are commonly anisotropic and this needs to be considered when interpreting ultrasound measurements of bone.

4.5 ULTRASOUND APPLICATIONS

Some polar materials, e.g., lead zirconate titanate (PZT), change their dimension when a voltage difference is applied on different sides of the material (electrostriction). Moreover, the voltage difference between the sides of the material can be measured, when the dimensions of the material are changed (piezoelectric effect). These material properties are utilized in ultrasound transducers when an electrical pulse is converted to a mechanical pulse (transmission), and when the mechanical pulse is converted back to an electrical signal (receiving). There are various types of ultrasound transducers which can be matched to the specific applications and needs.

Flat faced ultrasound transducer is technically simple and it is the most widely utilized. The transducer may also be focused, which means that the ultrasound energy is directed to a certain location through focusing either with on an acoustic lens or by the application of phased-array technology (Figure 4.2).

(46)

Depth of field

Di FWHM

F_L Lens

R_L

Figure 4.2: A schematic view of a focused ultrasound transducer. Focal length (F_l), depth of field (DOF), the diameter of the transducer (D_i), radius of lens curvature (R_l), and the beam width at half maximum amplitude (full width half maximum, FWHM) are indicated.

Table 4.3: Equations describing the characteristics of an focused ultrasound transducer [95, 97].

Focal length [m] F_l= ^R^l

1− ^c^l cm

Full width of the half maximum [m] FW HM= ^λF^l D_i = ^c^m^F^l

f D_i Depth of field [m] DOF=_7λ(^F^l

D_i)²=₇^c^m f (^F^l

D_i)²

Fl = focal length,Rl = radius of lens curvature,cl = speed of sound in lens, cm= speed of sound in medium,λ= wavelength,Di= diameter of transducer andf = transducer center frequency.

The basic equations describing the focused transducer properties are summarized in Table 4.3. In the pulse-echo (PE) geometry, one transducer is used to transmit and receive the ultrasound signal. All the measurements in this thesis were conducted with PE geometry. In the through transmission (TT), the geometry trans- mitting transducer is placed on the other side of the object and the

(47)

receiving transducer records the transmitted signals on the other side of the object (Figure 4.3).

Figure 4.3: Transmission and PE geometries. TOF denotes time of flight.

Quantitative ultrasound for evaluation of bone

The majority of the clinical quantitative ultrasound (QUS) devices that are intended for use in the diagnostics of osteoporosis measure peripheral sites, e.g., calcaneus, with TT geometry. Although, also radius and phalanges can be measured [117]. Most commonly, the speed of sound and broadband ultrasound attenuation are determined [24, 28, 84, 118]. The used techniques to determine speed of sound vary between manufacturers and in an attempt to avoid manufacturer dependent variation in speed of sound, a form of standardization has been proposed [119, 120, 120, 121]. Definitions describing the calculation of various QUS parameters are presented in Tables 4.4 and 4.6.

Technical and signal processing development in the field of ultrasound has been adapted also in the QUS evaluation of bone

Quantitative characterization of proximal femur using ultrasound pulse-echo measurements

Markus Malo

Quantitative Characterization of Proximal Femur Using Pulse-Echo Ultrasound Measurements

Markus Malo

Quantitative

Characterization of

Proximal Femur Using

Pulse-Echo Ultrasound

Measurements

Quantitative

Characterization of Proximal Femur Using Pulse-Echo Ultrasound

Measurements

To Laura, Luukas and Eevi

Acknowledgements

Contents

1 Introduction

2 Bone

3 X-ray assessment of bone

10mm

4 Ultrasound assessment of bone