Anthropomorphic breast phantomswith physiological water, lipid, andhemoglobin content for near-infraredspectral tomography

Kelly E. MichaelsenVenkataramanan KrishnaswamyAdele ShenoyEmily JordanBrian W. PogueKeith D. Paulsen

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Anthropomorphic breast phantoms with physiologicalwater, lipid, and hemoglobin content for near-infraredspectral tomography

Kelly E. Michaelsen,* Venkataramanan Krishnaswamy, Adele Shenoy, Emily Jordan, Brian W. Pogue, andKeith D. PaulsenDartmouth College, Thayer School of Engineering, 14 Engineering Drive, Hanover, New Hampshire, 03755

Abstract. Breast mimicking tissue optical phantoms with sufficient structural integrity to be deployed as stand-alone imaging targets are developed and successfully constructed with biologically relevant concentrations ofwater, lipid, and blood. The results show excellent material homogeneity and reproducibility with inter- and intra-phantom variability of 3.5 and 3.8%, respectively, for water and lipid concentrations ranging from 15 to 85%. Thephantoms were long-lasting and exhibited water and lipid fractions that were consistent to within 5% of theiroriginal content when measured 2 weeks after creation. A breast-shaped three-compartment model of adipose,fibroglandular, and malignant tissues was created with water content ranging from 30% for the adipose simulantto 80% for the tumor. Mean measured water content ranged from 30% in simulated adipose to 73% in simulatedtumor with the higher water localized to the tumor-like material. This novel heterogeneous phantom design iscomposed of physiologically relevant concentrations of the major optical absorbers in the breast in thenear-infrared wavelengths that should significantly improve imaging system characterization and optimizationbecause the materials have stand-alone structural integrity and can be readily molded into the sizes andshapes of tissues commensurate with clinical breast imaging. © 2014 Society of Photo-Optical Instrumentation Engineers

(SPIE) [DOI: 10.1117/1.JBO.19.2.026012]

Keywords: biomedical optics; medical imaging; tissues; spectroscopy.

Paper 130701R received Sep. 26, 2013; revisedmanuscript received Dec. 23, 2013; accepted for publication Jan. 13, 2014; publishedonline Feb. 18, 2014.

1 IntroductionPhantoms play a vital role in the development, validation, andquality control of imaging systems. Clinically, they are recom-mended for quality management or mandated for technicalsurveillance to avoid system malfunction and possible adverseeffects on patients undergoing examination. In research anddevelopment, phantoms are used for early-stage feasibility test-ing and performance evaluation; they assist in diagnosing errorsor underperforming instrumentation, and enable comparisons ofdata acquired on different imaging systems.

Phantom measurements have certainly been an importantpart of the development of near-infrared (NIR) spectral tomog-raphy (NIRST). NIR light (600 to 1000 nm) is preferentiallyabsorbed by hemoglobin, water, and lipids—tissue chromo-phores that are often altered in the presence of malignancy.1–5

Thus, breast imaging using NIRST has been studied extensively,and imaging systems deployed clinically often incorporatea homogeneous calibration phantom. These phantoms provideinformation on imaging system accuracy by offering an exper-imental environment where known chromophore concentrationsin the phantom can be compared to their recovered (imaged)counterparts, which in turn improves confidence during breastimaging when comparisons to known values are not possible.

Phantoms are typically used in validation studies of systemaccuracy and repeatability, and can be constructed from resins orplastics. They should be long lasting, homogeneous, durable,and possess optical properties similar to tissue. Extensivework has been reported on the development of these types of

phantoms, and several papers review the different materialsand scattering agents that are available.6,7 Measurements fromthese phantoms should be highly repeatable so that any changesin the data can be attributed to changes in the system. At present,commercial companies exist that produce customized homo-geneous phantoms with tissue-like optical properties in a widerange of sizes and material options.8

A second type of optical phantom, referred to as an anthropo-morphic phantom, is intended to mimic the breast more closelyin both physical shape and tissue composition. It is important forinvestigating an NIRST system’s ability to recover tissue chro-mophores in different concentrations, locations, and sizes withina heterogeneous volume of material with a scale similar to thebreast. These phantoms assist in the optimization of data collec-tion and image reconstruction for a given imaging system, aswell as in determining which patient populations are most likelyto benefit from the technique. Similar to system validation phan-toms, durability and repeatability are important. However, main-taining a spectral absorption profile and absorber concentrationssimilar to those in the tissue of interest is paramount to successin developing an anthropomorphic phantom. To mimic thebreast closely, they should be composed of hemoglobin, water,and lipids in varying physiological concentrations and havetheir central zones more similar to fibroglandular tissue andtheir outer areas more representative of adipose tissue in con-cordance with the typical breast parenchymal pattern.9

Previous breast-simulating phantoms have been constructedfrom hemoglobin, water, and intralipid.10 The latter is typicallyadded not to mimic breast lipid content, but to create optical

*Address all correspondence to: Kelly E. Michaelsen, E-mail: kelly.e.michaelsen.th@dartmouth.edu 0091-3286/2014/$25.00 © 2014 SPIE

Journal of Biomedical Optics 026012-1 February 2014 • Vol. 19(2)

Journal of Biomedical Optics 19(2), 026012 (February 2014)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

scattering. Thus, it is used in low concentrations by volume(∼1%). These phantoms effectively emulate the tissue opticalproperties of oxyhemoglobin, and in some cases, deoxyhemo-globin, but they do not represent physiologically relevant wateror lipid contents due to their low intralipid percentage.11–13

Development of physiologically relevant water and lipidphantoms is especially important for evaluating NIR imagingsystems that incorporate wavelengths >900 nm,14–16 whereabsorption by these chromophores is more significant than atlower wavelengths where the hemoglobin absorption dominates.Water and lipids are not only the main absorbers at longer opti-cal wavelengths; they also comprise the bulk of breast tissuevolume. Accordingly, several groups have described the devel-opment of water and lipid phantoms. For example, Merrittet al.17 correlated a series of water and lipid fractions withmagnetic resonance imaging and diffuse optical tomography.Nachabé et al.18 analyzed water and lipid contents at higherwavelengths. Most recently, Quarto et al.19 characterized severalrecipes for phantoms composed of water and lipids with threedifferent emulsification agents.

In this paper, a robust method is reported for creating semi-solid phantoms with physiologically relevant water and lipidvolume fractions that have sufficient structural integrity tostand alone. The free-standing character of these phantoms elim-inates the confounding effects of light channeling from a hous-ing container6 and allows anthropomorphic breast shapes andsizes to be created. The breast is typically composed of adiposetissue, ∼81% on average,20 and although adipose tissue is not100% lipids, it does have a lipid fraction up to 85%;21 hence,the most accurate breast phantoms should have high lipid con-tent. Here, we investigate the creation of phantoms with lipidcontents >70% in a free-form geometry. A major focus isthe systematic examination of emulsifiers to provide the physi-cal scaffolding necessary to create anthropomorphic free-stand-ing phantom structures. Combining water and lipid-basedphantoms with hemoglobin is also important. Thus, ease of cre-ation, durability, reproducibility, and cost and accessibility ofmaterials are additional factors that served as driving forcesfor the development of this new optical breast tissue phantom.

2 Materials and Methods

2.1 Phantom Creation

2.1.1 Material testing

The most effective phantom recipe was found by testing differ-ent combinations of lipid, emulsifier, and water. Water wasmixed with butter, margarine, olive oil, canola (rapeseed) oil,Crisco® (vegetable oil), and lard. For each of these combina-tions, a different emulsifier was used: guar gum, soy lecithin,and borax (sodium borate). These emulsification agents wereselected because they are ubiquitous, inexpensive, and nontoxic.The components were mixed using a common blender. Theliquid mixture was then poured into a small plastic containerand refrigerated overnight. Different ratios of fat to water phan-toms were created: 30∶70; 40∶60; 50∶50; 60∶40, and 70∶30,and tested with different concentrations of emulsifiers. Thesephantoms were then inspected for their malleability and homo-geneity. The purpose of these studies was to ensure that semi-solid models could be constructed from a wide range of waterand lipid combinations that simulate actual breast tissue.

To determine how the different emulsification agents alteredthe optical absorption characteristics of the water and lipid

combinations, they were mixed separately with liquid (heated)Crisco and water and imaged in a spectrophotometer from600 to 1000 nm.

2.1.2 Water and lipid only phantoms

After experimenting with a number of lipids and emulsifyingagents, lard and guar gum were selected and used in all ofthe following studies. Combinations of 15∶85, 25∶75, 30∶70,65∶35, 60∶40, and 50∶50 by volume of water:lipids (andvice versa) were measured. Initial work (not shown) involvedmixtures closer to 50∶50 in content. After satisfactory resultswere obtained in these cases, phantoms were created that hadmore extreme water and lipid fractions than are reportedhere. Additionally, several identical phantoms were constructedusing the same procedure to test the repeatability of the pro-cedure, and some phantoms were imaged longitudinally toassess their longevity at time points separated by 2 weekswhere the phantom was stored in a dark refrigerator betweenimaging sessions.

The procedure for creating these phantoms is illustrated inFig. 1. First, lard was heated until melted (38°C� 2°C), andthen it was added to a mixture of water and 3% guar gumby weight. A handheld, automated mixer was immediatelyused to stir the ingredients at low and then medium speed set-tings. This mixture was then poured into containers that werecovered in plastic wrap. The phantoms were refrigerated over-night to solidify before being tested with a diffuse opticalspectroscopic imaging (DOSI) system.22 In many cases, thesephantoms were imaged the following day with no additionalmodifications; however, for thicker and larger phantoms,manual mixing using a handheld potato masher was required.Each measurement takes ∼2 s, and the phantoms were measuredmultiple times at different locations to assess heterogeneity.All scans were performed at room temperature.

Fig. 1 Schematic of the phantoms creation process.

Journal of Biomedical Optics 026012-2 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

2.1.3 Anthropomorphic phantoms

To create a more anthropomorphic phantom, different compo-nents were designed to represent fat, fibroglandular tissue,and tumor, where the first two layers were incorporated intoa breast shape mold. Porcine blood was added to phosphate-buf-fered saline to introduce hemoglobin content. Once the lard wasmelted, guar gum was added to the blood mixture, and immedi-ately afterward, lard was added while mixing with handheldbeaters. The tumor inclusion was formed in a 115-mL containerwith 80∶20 water:lipid ratio and 3% by weight guar gum with30 μM Hb. To construct the layer representing fibroglandulartissue, a 70∶30 water:lipid phantom with 3% by weight guargum and 20 μM Hb was made having a total volume of1400 mL. The thickness of this layer was ∼4.5 cm. To formthe layer that represented fat, a 30∶70 water:lipid phantom oftotal volume 360 mL was made with 3% by weight guargum and 10 μM Hb. It was 1.3 cm thick. All of the phantomswere refrigerated after mixing. In the case of an anthropomor-phic phantom, the remixing step was not necessary because thelayers of different compositions were thinner than in the waterand lipid only phantoms. Some mixing was performed in theprocess of adding the tumor inclusion to the fibroglandularlayer, as some of the fibroglandular material was removed sothat the tumor region could be added.

An anthropomorphic breast-shaped phantom with three dis-tinct tissue regions was imaged in a grid pattern similar to theone used for patient imaging with the DOSI system.22 The gridpattern consisted of 36 measurement points, each separated by1 cm in x and y directions and spanning 5 × 5 cm across thephantom. First, just the fibroglandular-like tissue layer wasimaged, then a section 2 cm in depth and diameter was removedfrom the phantom and replaced with the tumor-like material.The entire phantom was reimaged with measurements acquiredfrom the same positions. Last, a layer of adipose simulating tis-sue was placed on top of the fibroglandular/tumor layer and thegrid pattern of images was repeated.

2.2 Imaging System

Measurements were obtained on a DOSI system, currently onloan from the University of California at Irvine as part of a mul-ticenter clinical trial.23,24 This system possesses both frequencydomain and continuous imaging capabilities, which characterizereduced scattering coefficient and absorption across the NIRspectral bandwidth (650 to 1000 nm). The frequency domaincomponents use six distinct wavelengths from 650 to850 nm, sweeping through modulation frequencies from 50to 600 MHz with light detected via avalanche photodiodes.The continuous wave portion of the DOSI data samples thetissue at 1024 wavelengths between 580 and 1020 nm spaced∼0.5 nm apart using a broadband white light source anda CCD spectrophotometer. Measurements obtained with thissystem are reported to be accurate to 0.0006 mm−1 for theabsorption coefficient and 0.03 mm−1 for the reduced scatteringcoefficient.25 For the absorption and scattering properties of thephantoms studied here, this accuracy is equal to an error of ∼5%in these parameters.

The sample interface is a handheld probe with a fixed-sourcedetector separation of 28 mm. A series of phantom measure-ments were recorded for calibration prior to sample measure-ments.26 For initial experiments, the measurement probe wasplaced directly on the phantom surface and was cleaned after

each measurement. Later, phantoms were wrapped in a verythin layer of plastic to prevent direct contact with the handheldimaging probe. Comparisons of measurements in phantomswith direct contact versus plastic covering showed negligibledifferences.

2.3 Data Reconstruction

A power law fit of the data for the reduced scattering coefficientsmeasured at selected wavelengths defines the scattering proper-ties across the measured range. Chromophore concentrationswere calculated using the Beer–Lambert law for the measuredabsorption coefficients. Molar extinction values for hemoglobin,deoxyhemoglobin, water, and lipids27–29 were used in this proc-ess. Water and lipid fractions were constrained to 100%, andhemoglobin absorption was not considered for the phantomscomposed exclusively of water, lard, and emulsifier, but wasincluded for the anthropomorphic phantoms containing porcineblood.

3 Results

3.1 Creation of Optimal Phantoms

In creating semisolid phantoms consisting of water and lipids,lipid and emulsifying agents were tested across a range ofphysiologically relevant water to lipid ratios (30∶70 to 70∶30)as shown in Fig. 2(a). The amount of emulsifying agent was keptconstant for a given volume of phantom to ensure that its effecton optical properties would not vary across different phantomconcentrations. As expected, phantoms comprised mostly ofwater were more gelatinous than phantoms comprised mostlyof lipids.

When guar gum was used, all phantoms were semisolid andvisually homogeneous. Additionally, guar gum had the lowestsignal attenuation in spectrophotometry measurements in theNIR regime of any of the emulsifying agents presented here(data not shown). Thus, it appeared to be the most viable emul-sifier for water and lipid phantoms. Several lipid dominantmaterials underwent spectrophotometry analysis in the NIRrange, and Crisco (vegetable oil) and lard (porcine fat) matchedpublished absorption spectra.28,30 Ultimately, lard was selectedfor the phantoms as the types and percentages of fatty acidwere more similar in the animal fat than the vegetable oil.31

Despite the addition of an emulsification agent to these phan-toms, water and lipid peaks were clearly discernable as shownin Fig. 2(c). The lipid absorption peak is evident in the leftgraph as a sharp rise around 930 nm, while the water absorption,highlighted in the right graph, has a broad peak above 950 nmthat extends to ∼1000 nm. The average scattering amplitude of0.66 · 10−3λmb−1 and average scattering power of 0.42 of thelard-based phantom are on the low end of values found in humansubjects.32

3.2 Contrast Sensitivity

Linear contrast recovery of water and lipids was found whenmeasuring several water:lipid ratios from 15∶85 to 85∶15,roughly the physiologic limits for the macroscopic tissuesprobed by diffuse optical techniques.21 These results areshown in Fig. 3 with an R2 of 0.998 for the linear fit. Each phan-tom was measured at 10 distinct locations, and the mean stan-dard deviation of these measurements was 3.5%. However,as noted by other groups,17,19 lipid content is overestimated,

Journal of Biomedical Optics 026012-3 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

especially for high lipid ratios. Visual inspection of these phan-toms showed a color gradient as observed in the photograph inFig. 4, and measurements along the side of this phantom dem-onstrated a decrease in lipid content from the top to the bottom.These phantoms were subsequently remixed manually and thenremeasured. The recovered water and lipid content were againlinear with an R2 of 0.96, but with recovered values within 2.5%of their actual values on average as shown by the dashed line inFig. 3. They maintained homogeneity as well with an average

standard deviation of 4.1%. The phantom constituents do notseparate after manual remixing as long as the phantoms, oncefully formed, are maintained at room temperature or colder, soonly a single remixing step is required.

3.3 Reproducibility

In order to test the reproducibility of the phantom creation proc-ess, three phantoms composed of the same volume of water,

Fig. 2 (a) Several early phantom creations comprised 70% lipid and 30% water. The fats were butter,olive oil, Crisco, and canola oil (left to right). The first and third phantoms included guar gum as theemulsifier, while the second and fourth utilized soy lecithin. (b) The chemical formula for guar gum34.(c) Data from the diffuse optical spectroscopic imaging system for a mostly lipid (left) and mostly water(right) phantom. The top graph shows the scattering data and fits, while the bottom graph shows theabsorption results. Specific peaks for lipids and water can be discerned above 900 nm.

Fig. 3 Graphs of measured water (a) and lipid (b) fractions with (round) data points and standard devia-tions based on 10 measurements from the top of the phantoms. The dotted line is a linear fit for thephantomswith the higher water content before mixing, exhibiting good linearity. The x-shaped data pointsand corresponding dashed linear fit for the high lipid content phantoms were obtained after their materialswere manually mixed. The solid line depicts the actual water or lipid content.

Journal of Biomedical Optics 026012-4 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

lard, and guar gum were fabricated. Each phantom was createdby independently following the steps shown in Fig. 1, but allthree used ingredients from the same containers and weremade by the same individual. The results of this study areshown in Fig. 4(c). The recovered lipid contents were 60.7,59.7, and 65.0% for the three samples and had intraphantomstandard deviations of 1.97, 1.60, and 4.72%, respectively,and interphantom standard deviation of 3.81%.

3.4 Durability

Phantoms were evaluated at a variety of time points after initialcreation. Because of the need for cooling, no phantoms weretested without at least 4 h of refrigeration. Three phantomswith 15∶85, 25∶75, and 35∶65 water:lipid contents were testedafter 2 weeks of refrigeration. At the time of creation, the recov-ered lipid content was 87.9, 72.1, and 65.8%, respectively. Twoweeks later, the values were 87.3, 79.7, and 64.3%, respectively.After an additional week of refrigeration, some phantoms devel-oped mold, indicating that with proper refrigeration these phan-toms may last for several weeks.

3.5 Anthropomorphic Test Case

The average water content of the fibroglandular simulating layerwas measured as 54.9%, 72.8% for the tumor layer prior toinsertion in the multilayer phantom, and 29.5% for the adiposelayer. As shown in Fig. 5, the tumor inclusion was recoveredwith 16% greater water content when compared to the fibro-glandular region. The actual tumor water content was 10%greater than its fibroglandular counterpart. The adipose regionwas recovered with a water content within 0.5% of the actualamount. After the tumor region was included in the phantom,measurements in that area showed higher water content andlower lipid content than the surrounding measurement points.Additionally, after the adipose layer was placed on top of theother two, measured water content decreased and lipid contentincreased as expected. Hemoglobin recovery was 33.2, 25.5,and 8.2 μm for the tumor, fibroglanduar, and adipose regions,respectively, relative to the actual amount of hemoglobin added,i.e., 30, 20, and 10 μm Hb in each of the three tissue types.

4 DiscussionAfter testing several types of emulsification agents and lipids,guar gum and lard were selected as the ideal constituents forcreating breast mimicking phantoms. Lard was preferable,as animal fat is likely more similar to the adipose contentin human tissue relative to vegetable-oil-based products.However, the percent of different types of fatty acids containedin the lard can change depending on the animal’s diet and thepart of the animal from which the fat was contributed.33

Regardless, changes in the near-infrared absorption due todifferent fatty acid composition are very small,31 indicatingthat both Crisco and lard could be used depending on theiravailability. To minimize the effects of impurities, all samplesproduced on a given day were constructed from the samebatch of melted lard.

Guar gum showed the lowest absorption in spectrophotom-etry studies and had the greatest thickening power and homo-geneity in phantom formation—characteristics that can beunderstood from its molecular structure. Guar gum is a highmolecular weight polysaccharide composed of highly branchedgalactose and mannose units.34 The chemical formula for guargum is shown in Fig. 2(b). Additionally, examinations of theabsorption spectra of phantoms composed of water, lipids, andemulsifier exhibited easily discernable peaks for their water andlipid constituents as shown in Fig. 2(c), and accurate chromo-phore recovery occurred as demonstrated in Fig. 3.

Water and lipids do not mix spontaneously, but vigorousmixing can create an emulsion, although the two will separateagain shortly thereafter. The large size and abundance ofbranches and hydroxyl groups on the guar gum creates bondsthat decrease molecular movement after mixing, thus creatinga stable emulsion that is thickened and more solid than the liquidcomponents. Last, cooling after mixing further inhibits separa-tion of water and lipids, and enhances the solidity of the phan-tom. The final product is moldable, and retains its shape afterapplication of minimal pressure.

This process is not infallible, as some separation of lipid andwater may occur as the phantom begins to solidify, causing thelinear concentration gradients shown in Fig. 4. The effect isa result of the time delay between the mixing and solidifica-tion of the phantom. Procedures exist to mitigate these effects.One method is postrefrigeration manual mixing, which was

Fig. 4 (a) Photograph showing visible differences between the top (left) and bottom (right) of the phan-tom. (b) Corresponding depth-dependent measurements confirming the greater presence of lipids atthe top of the phantom and higher water at the bottom of the phantom. (c) Repeatability of the phantomcreation process in three independently constructed phantoms with the same water and lipid ratio, eachmeasured five times.

Journal of Biomedical Optics 026012-5 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

very successful as shown in the data and dashed line in Fig. 3,and significantly improved the accuracy of the water and lipidquantification of the mixed material when compared to the samemeasurements on the unmixed material. Mixing techniques canhave an impact on the final phantom outcome. Manual mixinginvolves slow, gentle compression of the materials at roomtemperature by hand for a few minutes (the amount of timecan vary depending on the size of the phantom), not at highspeed, as blending the solid phantom may introduce air bubblesthat would alter the desired optical properties. When differentindividuals mixed the same phantom, average differences inmeasured water and lipid content were <5%.

Other steps that could be taken include minimizing the heightof the phantom, mixing the heated lipids with cold water todecrease the time for solidification, or cooling the mixturemore quickly. The first strategy was successful when makingthe thin adipose layer, whereas the other approaches were notinvestigated in this study, but would likely improve results infuture phantom experiments.

These phantoms possess intrinsic scattering properties result-ing from their water and lipid interfaces, which change with

concentration. Specifically, higher fat content leads to greaterscattering. This behavior is opposite to the scattering foundin breast tissue, possibly because tissue water is mostlyfound within cells that also possess a number of light scatteringorganelles, whereas these phantoms are composed of pure waterwith no additional scatterers. Differences in scattering, in addi-tion to differences in absorption, are accounted for in the fre-quency domain instrumentation.

Despite this water-lipid separation issue, creating durable,homogeneous, repeatable semisolid phantoms across a broadrange of water-to-lipid ratios with easily accessible ingredientsand tools was demonstrated. Inter- and intraphantom variabilitywas <5%. The phantoms do not break and can be remolded intodifferent shapes as needed, unlike agar or gelatin materials thatoften crack. They can be compressed with manual force (moreeasily for higher water content but possible at all compositions),unlike hard resin phantoms. Phantoms measured 2 weeks aftercreation exhibited water and lipid contents within 5% of theiroriginally measured concentrations. Their semisolid consistencyeliminates the complexities of imaging phantoms within con-tainers where light channeling may occur. These phantoms

Fig. 5 (a) Schematic of a free-standing three-compartment phantom and the grid pattern used to assessits optical properties. A given phantom was evaluated at 36 locations on a 1-cm grid pattern as shown.Interpolated results across the 5 × 5 cm grid of individual measurements of optical properties areshown for water in (b) and lipids in (c) in the fibroglandular, tumor, and adipose regions. In the top row,the fibroglandular-simulating phantom was measured alone. The middle row depicts results whena tumor-like inclusion of ∼2 cm diameter was added near 20 to 40 mm in x and 60 mm in y directions.In the bottom row, measurement data are shown when a uniform adipose simulating layer was added ontop of the fibroglandular and tumor phantom. Photographs of the phantom regions are shown in (d).

Journal of Biomedical Optics 026012-6 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

were inexpensive to construct, costing <5.00 per 700 mL ofphantom material, and all of the ingredients and equipmentcould be purchased at a local grocery store.

An anthropomorphic breast-shaped phantom with three dis-tinct tissue regions was studied as a test case. It contained thethree main light absorbers in the NIR—hemoglobin, water, andlipid—and possessed intrinsic scattering properties, albeit onthe lower end of what is expected for human breast because itsmaterials do not contain human cells or organelles. Becausethe melting point of lard is well below the temperature of hemo-globin denaturation, blood can easily be incorporated into thisphantom.35 Adding hemoglobin to the water and lipid phantomsdid not substantially alter their water and lipid fractions. Fatcontent was overestimated in the fibroglandular and tumorphantom sections by ∼15 and 8%, respectively, likely due tothe thickness of these compartments (postrefrigeration mixingwas not performed in this case) but was <1% in error in theadipose layer. The tumor can be localized when added to thefibroglandular background as shown in Fig. 5. This exampledemonstrates the feasibility of producing physiologicallyrelevant NIR phantoms composed mainly of water, lipid, andblood that have sufficient structural integrity to be used withoutany other supporting containers or fixtures.

5 ConclusionsThe phantom process described here can be used to create accu-rate breast-like optical phantoms with matching biologicalcomposition and physical shape that have sufficient structuralintegrity to be used as stand-alone imaging targets. Linear recov-ery of water and lipid concentrations has been found for com-positions between 15 and 85% with errors of <5% of the actualamounts when postrefrigeration mixing was performed. Bloodcan readily be added to these phantoms, which are then com-posed of physiologically relevant percentages of the threemajor absorbers in the NIR regime. The phantoms are moldableand easily shaped by the user. They are also durable, long-last-ing and repeatable, easy to make, and inexpensive.

Given the positive characteristics of these phantoms, severalareas of research could benefit from their construction. Forexample, these phantoms can be used for evaluating imagingsystems that obtain limited spectral information above 900 nmin order to understand their sensitivity to water and lipidcontrast. Alternatively, a multicompartment phantom modelof different tissue types can be used to test region-based recon-structions for tomographic imaging systems as well as to assesssignal-to-noise characteristics in an optically heterogeneousenvironment. The semisolid character of these phantoms is use-ful for testing the effects of different breast shapes and sizes onpatient interfaces and at tissue boundaries. Use of these phan-toms, which closely mimic tissue optical properties, can provideinformation to help optimize imaging system development anddetermine which patients can be successfully imaged on thosesystems.

AcknowledgmentsThe authors would like to acknowledge collaborators at theUniversity of California at Irvine for permitting the use of a dif-fuse optical spectroscopic imaging system for this work as wellas John Winn, professor of chemistry at Dartmouth College,for his physical chemistry advice. This work was fundedby National Institutes of Health grants R01CA139449 andF30CA168079.

References1. Q. Fang et al., “Combined optical and x-ray tomosynthesis breast

imaging,” Radiology 258(1), 89–97 (2011).2. B. J. Tromberg et al., “Non-invasive in vivo characterization of breast

tumors using photon migration spectroscopy,” Neoplasia 2(1–2), 26–40(2000).

3. S. P. Poplack et al., “Electromagnetic breast imaging: results of a pilotstudy in women with abnormal mammograms,” Radiology 243(2),350–359 (2007).

4. M. G. Pakalniskis et al., “Tumor angiogenesis change estimated byusing diffuse optical spectroscopic tomography: demonstrated correla-tion in women undergoing neoadjuvant chemotherapy for invasivebreast cancer?,” Radiology 259(2), 365–374 (2011).

5. X. Intes, “Time-domain optical mammography SoftScan: initialresults,” Acad. Radiol. 12(8), 934–947 (2005).

6. B. W. Pogue and M. S. Patterson, “Review of tissue simulating phan-toms for optical spectroscopy, imaging and dosimetry,” J. Biomed. Opt.11(4), 041102 (2006).

7. J. Hwang, J. C. Ramella-Roman, and R. Nordstrom, “Introduction: fea-ture issue on phantoms for the performance evaluation and validation ofoptical medical imaging devices,” Biomed. Opt. Express 3(6), 1399–1403 (2012).

8. J. P. Bouchard et al., “Reference optical phantoms for diffuse opticalspectroscopy. Part 1—Error analysis of a time resolved transmittancecharacterization method,” Opt. Express 18(11), 11495–11507 (2010).

9. S. Thomsen and D. Tatman, “Physiological and pathological factors ofhuman breast disease that can influence optical diagnosis,” Ann. N. Y.Acad. Sci. 838, 171–193 (1998).

10. R. Michels, F. Foschum, and A. Kienle, “Optical properties of fat emul-sions,” Opt. Express 16(8), 5907–5925 (2008).

11. T. O. McBride et al., “Spectroscopic diffuse optical tomography forthe quantitative assessment of hemoglobin concentration and oxygensaturation in breast tissue,” Appl. Opt. 38(25), 5480–5490 (1999).

12. M. A. Mastanduno et al., “Automatic and robust calibration of opticaldetector arrays for biomedical diffuse optical spectroscopy,” Biomed.Opt. Express 3(10), 2339–2352 (2012).

13. K. Michaelsen et al., “Near-infrared spectral tomography integratedwith digital breast tomosynthesis: effects of tissue scattering on opticaldata acquisition design,” Med. Phys. 39(7), 4579–4587 (2012).

14. L. Spinelli et al., “Characterization of female breast lesions from multi-wavelength time-resolved optical mammography,” Phys. Med. Biol.50(11), 2489–2502 (2005).

15. J. Wang et al., “In vivo quantitative imaging of normal and cancerousbreast tissue using broadband diffuse optical tomography,” Med. Phys.37(7), 3715–3724 (2010).

16. V. Krishnaswamy et al., “A digital x-ray tomosynthesis coupled nearinfrared spectral tomography system for dual-modality breast imaging,”Opt. Express 20(17), 19125–19136 (2012).

17. S. Merritt et al., “Comparison of water and lipid content measurementsusing diffuse optical spectroscopy and MRI in emulsion phantoms,”Technol. Cancer Res. Treat. 2(6), 563–569 (2003).

18. R. Nachabé et al., “Estimation of biological chromophores using diffuseoptical spectroscopy: benefit of extending the UV-VIS wavelengthrange to include 1000 to 1600 nm,” Biomed. Opt. Express 1(5),1432–1442 (2010).

19. G. Quarto et al., “Comparison of organic phantom recipes and charac-terization by time-resolved diffuse optical spectroscopy,” Proc. SPIE8799, 879905 (2013).

20. T. R. Nelson et al., “Classification of breast computed tomographydata,” Med. Phys. 35(3), 1078–1086 (2008).

21. S. J. Graham et al., “Changes in fibroglandular volume and water con-tent of breast tissue during the menstrual cycle observed byMR imagingat 1.5 t,” J. Magn. Reson. Imaging 5(6), 695–701 (1995).

22. W. Tanamai et al., “Diffuse optical spectroscopy measurements ofhealing in breast tissue after core biopsy: case study,” J. Biomed. Opt.14(1), 014024 (2009).

23. A. E. Cerussi et al., “Diffuse optical spectroscopic imaging correlates withfinal pathological response in breast cancer neoadjuvant chemotherapy,”Philos. Trans. Math. Phys. Eng. Sci. 369(1955), 4512–4530 (2011).

24. A. Cerussi et al., “In vivo absorption, scattering, and physiologic proper-ties of 58 malignant breast tumors determined by broadband diffuseoptical spectroscopy,” J Biomed. Opt. 11(4), 044005 (2006).

Journal of Biomedical Optics 026012-7 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

25. K. S. No et al., “Design and testing of a miniature broadband frequencydomain photon migration instrument,” J. Biomed. Opt. 13(5), 050509(2008).

26. A. E. Cerussi et al., “Tissue phantoms in multicenter clinical trials fordiffuse optical technologies,” Biomed. Opt. Express 3(5), 966–971(2012).

27. L. Kou, D. Labrie, and P. Chylek, “Refractive indices of water and ice inthe 0.65- to 2.5-μm spectral range,” Appl. Opt. 32, 3531–3540 (1993).

28. C. Eker, “Optical characterization of tissue for medical diagnostics,”Doctoral Thesis, Department of Physics, Lund Institute of Technology(1999).

29. W. G. Zijlstra, A. Buursma, and O. W. van Assendelft, Visible and NearInfrared Absorption Spectra of Human and Animal Haemoglobin:Determination and Application VSP (2000).

30. R. L. P. Van Veen et al., “Determination of visible near-IR absorptioncoefficients of mammalian fat using time- and spatially resolved diffusereflectance and transmission spectroscopy,” J. Biomed. Opt. 10(5),054004 (2005).

31. C.-L. Tsai, J.-C. Chen, and W.-J. Wang, “Near-infrared absorptionproperty of biological soft tissue constituents,” J. Med. Biol. Eng.21(1), 7–14 (2001).

32. B. Brooksby et al., “Imaging breast adipose and fibroglandular tissuemolecular signatures by using hybrid MRI-guided near-infrared spectraltomography,” Proc. Natl. Acad. Sci. U S A 103, 8828–8833 (2006).

33. D. E. Koch et al., “Effect of diet on the fatty acid composition of porkfat,” J. Anim. Sci. 27(2), 360–365 (1968).

34. Y. Kawamura, Guar Gum: Chemical and Technical Assessment, http://www.fao.org/fileadmin/templates/agns/pdf/jecfa/cta/69/Guar_gum.pdf(2008).

35. R. F. Rieder, “Hemoglobin stability: observations on the denaturation ofnormal and abnormal hemoglobins by oxidant dyes, heat, and alkali,”J. Clin. Invest. 49, 2369–2376 (1970).

Kelly E. Michaelsen is an MD/PhD candidate currently workingtoward a PhD in biomedical engineering at the Thayer School ofEngineering. She graduated from Dartmouth College with honors,majoring in physics andminoring in chemistry while pursuing researchin particle physics. She is interested in developing imaging methodsthat translate into more effective screening, diagnosis, and treatmentof disease. Her current research focuses on combining x-ray andoptical imaging modalities for breast cancer surveillance.

Venkataramanan Krishnaswamy received his PhD in opticalscience and engineering from the University of Alabama inHuntsville with specialization in optical systems design and engineer-ing. He is currently a research faculty member at the ThayerSchool of Engineering, Dartmouth College. His current areas ofresearch include imaging localized tissue scatter responseusing structured light and dark-field confocal spectroscopyapproaches and multimodal imaging systems development combin-ing near-infrared spectral tomography with mainstream clinical imag-ing modalities.

Adele Shenoy is an undergraduate student at Dartmouth Collegepursuing a degree in economics while fulfilling the premedical require-ments. As a Women in Science Scholar, Presidential Scholar,and Undergraduate Research Scholar, she has been deeply involvedin the development, testing, and optimization of optical phantoms for acombined near-infrared and breast tomosynthesis imaging system.

Emily Jordan is an undergraduate student at Dartmouth Collegemajoring in music and minoring in Italian studies and completingthe premedical curriculum. As a research associate, her work focuseson optical phantom development, hardware design, human subjectinterfaces, and imaging.

Brian W. Pogue is a professor of engineering, physics, andastronomy at Dartmouth College, and adjunct professor of surgeryat the Geisel School of Medicine. His research is in optical imagingsystems, with a focus on molecular and structural imaging ofcancer for surgical guidance and imaging radiation therapy. Hehas published 240 peer-reviewed papers. His research is fundedby the National Cancer Institute, and he is a fellow of the OpticalSociety of America.

Keith D. Paulsen is currently the Robert A. Pritzker professor ofbiomedical engineering at the Thayer School of Engineering atDartmouth, professor of radiology at the Geisel School of Medicine,and director of the Advanced Imaging Center at Dartmouth HitchcockMedical Center. His research has focused on the development andtranslation of advanced imaging technology, primarily for cancerdetection, diagnosis, therapy monitoring, and surgical guidance. Hehas authored more than 350 archival publications with an activeresearch program, continuously funded by the National Institutes ofHealth for 25 years.

Journal of Biomedical Optics 026012-8 February 2014 • Vol. 19(2)

Michaelsen et al.: Anthropomorphic breast phantoms with physiological water, lipid, and hemoglobin content. . .

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 23 Nov 2021Terms of Use: https://www.spiedigitallibrary.org/terms-of-use