development of biological tissue-equi valent phantoms for...

Indi:lI1 Journal of Experimelllal Biology Vol. 40, May 2002, pp . 53 I -535

Development of biological tissue-equi valent phantoms for optical imaging

R Srinivasan & Megha Singh*

Biomedi cal Enginee ri ng Di vision, Indian Institute o f Tec hnology, Chenna i 600036, India

Received 24 Augu.l'l 2000; revised 18 Jalluary 2002

Optical charac teris ti cs or freshly isolated ti ssues depcnd on thcir color and compositi on. Thc surface backscallered pro­fil e. wh ich account fo r the ti ssue compos iti onal variati on in fre sh cxc ised sheep's heart , lungs, bone and musc le. wcre meas­ured by multi-probe refl ectometer. Optical phantoms were prepared from paraffin wax by mi xing a specifi c combin ati on of wax color materials till the surface backscattered profile of these matched with that of the biologica l tissues. The opt ica l pa­rameters absorption coeffi cient (p.), reduced scattering coefricient ( ~t ,') and an isotropy factor (g) 0 [' these phantoms, are the same as th at of bi olog ica l ti ssues and are obtained by match ing the ir surface backscattered profil es wit h th at as simulated by Monte Carlo proced ure. The maximum and minimum va lues of absorption coeffic icnt are fo r the phallloms of lungs ( I.Ocm- l

) and Illuscle (0.02cm- I), whereas, for scallering coeffi cient thcse va lucs are ['or muscle (2 1.2 cm- I

) and bone (13.08 cm- I).

When a laser beam is incident on a ti ssue, due to mismatch in the refractive index at air-t issue inter­face, part of the beam is specularly retlected, whereas, the remaining part of the beam enters the ti ss ue. Due to spa ti al va ri atio ns in ce llular structure, orientation and refracti ve i dex along the layers beneath , so me photons emerge at the surface as backsca ttered pho­tons, whereas, the res t are either absorbed or transmit­ted after und rgoi ng complex interacti on within the ti ssue medium. The photons, which are backscattered, appear at va ri ous locations from the beam entry point after interacting with the ti ss ue layers. The photons, wh ich appear ::I t the fa rthes t point, ori ginate from the deeper layers and vice versa l

.

Imaging wi th laser rad iat ion is an appeal ing co n­cept and is gaining importance as a non-invasive di­agnosti c moda lity for human ti ssues. This concept holds promise fo r providing methods of ti ss ue charac­teri zation for early detection of ti ssue abnormaliti es, wi th neither ' dverse side-e ffect of those associated wi th X-rays nor with positron emission tomography which requires injec ti on of hazardous contrast agents. Measurement of ultrasoni c scattering is a sensiti ve technique but the detectable changes in acousti c im­pedance in tissues take place at fairly advanced stage of growth of ab normaliti es2

. The change in refractive index is comparat ively more sensitive parameter than the X-ray attenuation coe ffi cient and ultrasonic aco us­ti c impedance of tissues. Laser radi at ions, in parli cu-

*Correspondc nce au thor Ema il : ms ingh@acL! r.iitm.crncl.in FAX: 9 1-44- 235 0509

lar within the opti cal window reg ion (600- 1300nm), penetrate much deeper in so ft ti ss ue and are well ­tol erated in large doses . Th is al so all ows hi gh re­emittance of li ght after deep penetrati on" 4 . For the purpose of diagnosis in so ft ti ssue organsS

, the li ght that is re-emitted (d iffusely refl ected from or transmit­ted through the ti ss ue) cou ld be measured to probe both the structural and functi onal aspects of the ti ssue. These characte ri sti cs have led to the deve lopment of laser transillumination imagi ng technique as a spec ial­ized diagnost ic method6

.

The practi cal methods of in vest igating the photon interact ion with ti ss ues (ill sitll ), espec iall y with so rt ti ssue organs such as human breast or thin organs li ke human hand is either to monitor backscattered rad ia-. 6 d I . I t' l ' h 7 g tl on or to etect tle transmlttec co mponent 0 tg t . .

As the scattering coefficient of ti ssues va ri es over a wide ran ge (22 to 4S cm-?, several materi als ha ve been used to fabricate various types of phantoms. Par­affin wax has most frequentl y bee n used l(). Bes ides thi s, Agar gel.J and milk " have also been used. But these are op ti ca ll y not matched with bio logica l ti s­sues . Milk, despite bein g a good scatterer, its color is not at all matching with that of ti ssues and the pene­tration depth of photons in thi s is signi fican tl y hi gher compared to that in so ft ti ssues.

Therefore, for development of optical phantoms, opti ca l matching of these with th at of the actual tis­sues is required. As it is diffi cult to meas ure the opti ­ca l transmi ss ion loss per unit distance of organs under ill vitro and ill vi vo conditions, the surface back­scatterecl profi Ie could be measured by an appropriate

532 INDIAN J EXP BIOL. MA Y 2002

techn ique. By matching this profile with that of paraf­fin wax mixed with right combination of wax colors, ti ssue-equivalent phantoms could be developed. This forms the objective of the present work to develop the ti ssue equivalent phantoms for heart, lung, bone and mu sc le ti ssues. To authors' knowledge such an ap­proach for the development of optica l phantoms has not been reported to date.

Materials and Methods

Laser rejleclOmeter The schematic of laser reflectometer is shown in

Fig. I. This system consists of laser source, optic fibre-probe assembly, photo-detectors, current-vo ltage converter and an ADC interface with a Pc.

Laser source is a compact laser diode modu Ie LDM 135 (Imatronic , UK) of power 2mW operating at 670nm. The optical fibre-p robe assembly consi sts of one input and three output fibres, each of active di­ameter 0.1 cm and 100cm length , arranged in parall el with a centre-to-centre separation of 0.2 cm as shown in the Fig. 2. The probe end is made up of a cy lindri­cal tube of 2.0 cm diameter and 5.0 cm length wi th fo ur holes to accommodate the optical fibres. The coupli ng between the fibres and the refl ectometer is achieved through plug-in adapters, which ensures that the light is incident upon the ti ssue through the input fibre, and the backscattered li ght is picked up by the three output fibres, to be detected by the photodiodes. Pol ish ing the fibre tips minimizes the intensity losses in transmiss ion through the fibres.

The back scattered light co ll ected by the output fibres is converted to proportional current by three high speed, low noi se silicon PIN diodes (BPW34,

pho\odetectors c·v convertors

optical· fibre guide

Fig. I-Schcmatic of laser rcncctoillcter systcm

Siemens). These diodes are mounted and coupled carefully to minimi ze the effect of stray li ght. The current output from three photodiodes is converted into their proportional voltage by three general­purpose operational ampli fiers (TL084, Texas Instru­ments) in the current-voltage configuration, with indi­vidua l ga in control. These outputs are di giti zed using an 8 bit ADC card (PCL-207 , Dynalog) wh ich is in­terfaced to a 386 PC for data acqui sition and process­mg.

The reflectometer was ca librated by placing the probe on a black rubber, which served as the standard surface and the reflectance of thi s was adjusted to zero by the gain control. Thereafter the surface refl ec­tance profiles of all ti ssues with reference to this were measured. To minimi ze the effect of stray li ght all observations were taken in the dark envi ronment. Fur­ther detail s of thi s technique are given elsewhere4

.

Tissue preparation Freshly excised sheep organs heart, lungs, bone

and musc le were procured from a commerc ial butcher and cleaned thoroughl y to remove the dirt and blood, if any. The overl ying soft ti ss ues on the bone surface were gentl y scrapped with a scalpel. The air in lungs was deflated gently. Thereafter, these organs were soaked in physiological saline for abou t 30 min to remove any excess blood. Prior to measurement of the reflectance the ti ssue surface was mopped with blot­ting paper to remove excess sa line and moisture.

Prepara tion of Phantom In thi s study, paraffin wax (Type- II) , a white crys­

talline odourl ess hyd rocarbon was used since it is eas­il y available and can be cast into any des ired shape with uni fo rm di stribution of coloring materi al. For prepara tion of a phan tom, measured quantities of

output3

mm

Ti ssue t Fig. 2-Schclllatic of optic fibrc - probe asscmbly , consisting of one input and thrcc output fib rcs.

SRINIVASAN & SINGH: BIOLOGICAL TISSUE-EQUIVALENT PHANTOMS FOR OPTICAL IMAGING 533

coloring materials (Ganapathy Color Company, Chennai, India) were added to 100 ml of melted wax, and stirred gently so that these were uniformly dis­tributed in the medium. Thereafter this was poured into a glass cylinde of diameter 3.0 and 10.0 cm length, and allowed to cool at the room temperature. After solidification this was removed from the cast cy linder and its surface reflectance measured.

Monte Carlo simulation technique This is a stochastic technique applied to randomly

occurring radiation transport processes. The scheme of the simulation is shown in the Fig. 3. A laser beam enters the tissue surface orthogonally. The tissue has infinite extensions in depth and width. The origin of the tissue base coordinate is at the tissue surface and at the center of the la er beam (0,0,0). The path length within the ti ssue is determined by its absorption ()la)

and reduced scattering coefficients ()ls'>. Part of the incident beam after multiple scattering inside the me­dium is emi tted as backscattered photons at various locations on the surface. The back scattered fractions calcul ated at various locations from the beam entry point are normalized. Further detail s of the procedure are given elsewhere l

.

The simulation procedure was carried out in two steps . In the first step the ti ssue medium was consi­dered as isotropic i.e., anisotropy coefficient g = 0. Then diffuse reflectance profiles were computed for various combinations of )ls (scattering coefficient) and )lao The simulated profile was matched with the ex­perimental profile and jJ.a value for the tissue was ob-

Incident y. .. 11 Diffusely ohoton .. " reflected

S ... " photon <. .... ........... (Q,.9, . .9.~:W.IJ.~J ........... r. .~>

Tissue 1 ./

WT2 == (J-lS/J-lI) 0 ,2 0

C O2 = [ (~a/~Jx! (~S/~l)] 'b Di=(~a/~I)(~S/~I)i-1

, i-I WTi=(~sf~l)

o zv Fig. 3-Scheme of MOnle Carlo simulation. Wt- weight at i th

scallering; D-absorbed dose during i th scallering.

tained. In the second step, )la was kept constant and the diffuse reflectance profile for various combina­tions of g and ~ls' was computed. By obtaining the best fit of the theoretical curve with that of the experimen­tal one the optical parameters )ls' and g of the materi al was obtained. One million photons were considered in the simulation procedure and each photon is tracked till its weight has reduced to 5%.

Data collection and analysis The surface reflectance profiles of heart , lungs,

muscle and bone were measured by holding the probe head perpendicular to the ti ssue surface in the dark environment. This corresponds to the max imum va lue compared to the values obtained at other inclinations. Prior to each measurement the probe was placed on the black rubber and it is adjusted to zero value. The reflectance values were expressed as normali zed in­tensity values with respect to the max imum input power measured by placing the source fiber directly at the photodetector. These normali zed backscattered intensity (NB!) va lues were plotted against rad ial dis­tance from the source entry point.

Simi larly , the back scattered profi Ie of the phantom was measured by placing the probe on its top fl at sur­face. To achieve the matching of the ti ssue experi­mental back scattered intensi ty profiles, phantoms with varying color compositions were prepared . The combination showing the best-fit with experimental one was selected. By similar procedure phantom of each tissue was prepared and its backscattered profi Ie measured.

Monte Carl o simulation scheme (Fig. 3), was im­plemented to obtain the di ffu sely refl ected surface profiles for a range of reduced scatteri ng and absorption coefficients, and anisotropy factors. The actual measurements were compared with these and the best matching curve was selected. The theoretical values of the )l/, )la and g thus obtained were the val­ues of these parameters for the tissue and its phantom.

Results Figure 4 shows the surface emiSSion profile of

freshly excised sheep organs heart, lungs, muscle and bone. The normalized backscattered intensity (N B!) decreases with tffe increase of distance from the beam entry point. The change is maximum for muscle tissue and minimum for heart. Bone and lungs show similar pattern but the NBI is higher near the entry point and less at farther positions compared to that of the heart.

534 INDiAN J EXP BIOl, MAY 2002

The tissue equivalent phantoms were developed and their NBI compared. Fig. 5 shows an example of the matching of the backscattered profile of heart ti s­sue and its phantom. By similar procedure the ti ssue equi va lent phantoms for lungs, muscle and bone were obtai ned .

4~-------------------------------'

3

'::2 co Z

o

o Bone

o Heart

t1 lung%

.&. Muscle

~------,-------,-------,-------~ 2 3 4 5 6

Distance from bea m e ntry po int

Fig. 4-Normali zed back scattered intensity profiles of bone, heart, lu ngs and muscle.

4~----------------------------~

o H e a rt o MCS

3

o 2 3 4 5 6

Di stance from beam entry poin t (mm)

Fig. 5-Variation of the normali zed backscattered in tensit y of the heart ti ssue and theoretica l c urve as obtai ned by Monte Carlo simula ti on

4~------------~ ,--------------,

3

2

o~~~~~~~~ __ ~~~ 2 3 4 5 62 3 4 5 6

Fi g. 6 - Nonnali zed backscattered in ten sity variation of the heart lungs, bone and muscle tissues and their phantoms along with best-fit curves as obtai ned by Monte Carl o simulation.

Depending upon the composition of the ti ssue, the optical parameters vary. Fig. 6 show the NBI curves for these ti ssues. The Monte-Carlo simulated curves show the best fit with that of the ti ssue and its equiva­lent phantom. Thus the values of optical parameters as ass igned to Monte-Carlo simulation are the same as that of the biological tissues and their equivalent phan toms. Table I shows that these parameters, which vary depending on the composit ion. The maximum and minimum values of absorption coefficient are for

the phantoms of lungs ( 1.0cm- t) and muscle

(0.02cm- t), whereas, for scattering coefficient these

are for muscle (2 1.2 em- I) and bone (13.08 em- I) , re­spectively .

Figure 7 shows the photograph of these phantoms. The variation in optical properties is not only attrib­uted to the color at the sUlface but its di stribution throughout the phantom, which are optica lly matching with that of ti ssues. The proportion of the colors added to wax is shown in Table 2. For muscle equi va­lent optical phantom three colors were used, whereas, for heart only red co lor was used. This is primarily attributed to the variation in the structu re of these ti ssues.

Discussion The structural components that influence the li ght

absorption and backscattering in ti ssues are color, compos ition and bl ood contents. From our measure­ment studies, it is found that the muscle tissue--bas the maximum scattering coeffic ient and thi s correlates well with its structure. Since skeletal muscle is a highly structured ti ssue it backscatters more when

Table I - Optical parameters of sheep organs as de termined by laser re flectometry and Monte Carlo simu lation

Ti ssue

Bone Heart lungs Muscle

Reduced scattering coefficient

(cm- I )

13.08 13.8 17.9 2 1.2

Absorption Ani sotropy coefficient factor

(c m- I )

0.08 0.8 0.18 0.78

1.0 0.75 0.02 0.74

Table 2 - Color combinati ons of ti ssue-equi valent phan toms

Tissue-equi valent optica l phantoms ( IOOm l of wax)

Bone Heart lungs Musc le

Color co mbinat ion (mg %)

Black Wh ite Red

0. 1464 2 .342 0. 1464

0.1672 2 .342 0.0731 2.342 0.142

SRINIVASAN & SINGH: BIOLOGICAL TISSUE-EQUIVALENT PHANTOMS FOR OPTICAL IMAGING 535

Fi g. 7 - Photograph of the ti ssue eq uivalent phantoms of bone (a), heart (b), lungs (c), and muscle (d) .

compared to the other tissues . Bone has the least scat­tering coefficient because of its compact structure. Lungs and bone show similar reflectance profiles, which could be attributed to their color and composi­tion . Even though heart is a structured tissue it offers less scattering which could be attributed to higher absorption by the entrapped blood inside its cavities. The present observations on tissues optical parameters agree well with that of others1 2. The matching of the surface profiles of these phantoms with that of ti ssues show that the optical properties of these are similar. Further comparison, due to non-availability of data on such phantoms in the literature, could not be carried out.

Preparation of tissue-equivalent phantom materi als by mixing spec ific combination of wax colors in melted wax is a si mple but novel technique. Combina­tion of colors is required for development of the phan­toms but for heart only one color (red) has been used. The red color pigment is characteri zed by high scat­tering and less absorbing and black is hi ghly absorb­ing in nature, whereas, white is moderately scattering and less absorbing. The phantoms prepared by the above procedure are homogeneous, as the measured backscattered intensity at vari ous cut sections shows no significant variation. These are mechanically stable and can be cast into any ti ssue geometry if necessary .

[n contrast to the surface reflectance measurement6,

the present reflectometer determines the variation in intern al structure3

. Hence these phantoms represent overall variation in ti ssue composition. This technol­ogy could also be used to determine the influence of superficial sk in lesions like melanoma, haemangioma etc ., in the body. These phantoms could further be

used to test, calibrate and standardi ze optical imaging systems.

References I Kumar D, Chacko S & Singh M, Monte Carlo simulation of

photon scallering in biological ti ssue models, In dian J Bio­chelll Biophys, 36 (1999) 330.

2 Swarnamani S, William A & Singh M, Monitoring of tumor growth and it s drug-induced regression in rats by laser and ultrasonic reOec tometer, Curl' Science, 55 ( 1986) 770.

3 Wil son B C & Jacques S L, Optical reflectance and transmit­tance of ti ssues: Principles and appli cations, IEEE J Qualll £Ieclron, 26 ( 1990) 2 186

4 Chacko S & Singh M, Multi layer imaging of human organs by measurement of laser back scallering radiati on, Med Bioi Ellg COII/PUI, 37 ( 1999) 278.

5 Chance B, Leigh J S, Mi yake H, Smith D S, Nioka S, Greenfeld R, Finander M, Kau fm an K, Levy W, You ng M, Cohen P, Yoshioka H & Boretski R, Comparison of time­resolved and unresolved measu rement of deoxy -hemoglobin in brain, Proc Nat Acad Sci, 85 (1988) 497 1.

6 DeHalier E B & Depursinghe C, Simulati on of time resolved breast transilluminat ion, Med Bioi Eng COlllpul, 3 1 (1993) 165.

7 Shanthi S & Singh M, Laser refl ectance imaging of mammal­ian organs under ill vilro conditions, 11I110V Techll Bioi Med, 17( 1996) 443.

8 Chacko S & Singh M, 3D reconstruction of breast phan toms, IEEE Trails Biollled Eng, 47 (2000) 131.

9 Peters V G, Wyma n D R, Pallerson M S & Fi rba nk G L, Op­ti ca l properties of normal and di seased human breast ti ssues in the visibl e and in fra red, Phys Med Biol, 35 (1990) 13 17.

10 Anderson-Engles S, Berg R, Swanberg & Jariman D, Time resolved transilluminati on for medical diagnosti cs, Oplics Lellers, 15 ( 1990) 1179.

II Mitic G, Kobcr J, OllO J, Pil es E, Solkner G & Zinth W, Time gated transilluminat ion of biological ti ssues and ti ssue like phantoms, Appl Opl, 33 (1994) 6699.

12 Chcong W, Prahl S A & Welch A J, A review of the opti ca l properties of biologica l ti ss ues, IEEE J Quallllllll Eleclroll, 26 ( 1990) 2 166.