doi:10.1152/physiolgenomics.00119.2001 10:113-126, 2002. First published Jun 18, 2002;Physiol Genomics

Adamson Y. Q. Zhou, F. S. Foster, D. W. Qu, M. Zhang, K. A. Harasiewicz and S. L.

You might find this additional information useful...

32 articles, 11 of which you can access free at: This article cites http://physiolgenomics.physiology.org/cgi/content/full/10/2/113#BIBL

15 other HighWire hosted articles, the first 5 are: This article has been cited by

  [PDF]  [Full Text]  [Abstract]

, May 1, 2008; 294 (5): H2248-H2256. Am J Physiol Heart Circ PhysiolK. Tobita N. Momoi, J. P. Tinney, L. J. Liu, H. Elshershari, P. J. Hoffmann, J. C. Ralphe, B. B. Keller and

and growthModest maternal caffeine exposure affects developing embryonic cardiovascular function 

[PDF]  [Full Text]  [Abstract], July 3, 2008; 103 (1): 24-33. Circ. Res.

Steer, A. C. Schuh, S. L. Adamson, M. I. Cybulsky and P. A. Marsden A.-M. Teichert, J. A. Scott, G. B. Robb, Y.-Q. Zhou, S.-N. Zhu, M. Lem, A. Keightley, B. M.

Unidirectional Circulatory SystemCommencement of Expression in the Embryo Occurs With the Establishment of a Endothelial Nitric Oxide Synthase Gene Expression During Murine Embryogenesis: 

[PDF]  [Full Text]  [Abstract], November 1, 2008; 295 (5): H2046-H2053. Am J Physiol Heart Circ Physiol

M. Gan J. Gronros, J. Wikstrom, U. Brandt-Eliasson, G. B. Forsberg, M. Behrendt, G. I. Hansson and L.

mouse model of atherosclerosisEffects of rosuvastatin on cardiovascular morphology and function in an ApoE-knockout 

[PDF]  [Full Text]  [Abstract], December 5, 2008; 103 (12): 1483-1491. Circ. Res.

Cross and H. J. Duff G. Q. Teng, X. Zhao, J. P. Lees-Miller, F. R. Quinn, P. Li, D. E. Rancourt, B. London, J. C.

LethalityDevelopmental Defects in the Right Ventricle and Its Outflow Tract and Embryonic Homozygous Missense N629D hERG (KCNH2) Potassium Channel Mutation Causes 

[PDF]  [Full Text]  [Abstract], March 1, 2010; 55 (3): 729-737. Hypertension

B. A. Croy S. D. Burke, V. F. Barrette, J. Bianco, J. G. Thorne, A. T. Yamada, S. C. Pang, M. A. Adams and

Pregnant MiceSpiral Arterial Remodeling Is Not Essential for Normal Blood Pressure Regulation in

including high-resolution figures, can be found at: Updated information and services http://physiolgenomics.physiology.org/cgi/content/full/10/2/113

can be found at: Physiological Genomicsabout Additional material and information http://www.the-aps.org/publications/pg

This information is current as of August 14, 2010 .  

http://www.the-aps.org/.American Physiological Society. ISSN: 1094-8341, ESSN: 1531-2267. Visit our website at July, and October by the American Physiological Society, 9650 Rockville Pike, Bethesda MD 20814-3991. Copyright © 2002 by thetechniques linking genes and pathways to physiology, from prokaryotes to eukaryotes. It is published quarterly in January, April,

publishes results of a wide variety of studies from human and from informative model systems withPhysiological Genomics

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

toolboxApplications for multifrequency ultrasoundbiomicroscopy in mice from implantation to adulthood

Y. Q. ZHOU,1,4 F. S. FOSTER,2,5 D. W. QU,1 M. ZHANG,2,5

K. A. HARASIEWICZ,2,5 AND S. L. ADAMSON1,3,4

1Samuel Lunenfeld Research Institute at Mount Sinai Hospital, and 2Sunnybrook andWomen’s College Health Sciences Centre, Departments of 3Obstetrics/Gynecology, 4Physiology,and 5Medical Biophysics, University of Toronto, Toronto, Ontario, Canada, M5G 1X5Received 18 December 2001; accepted in final form 11 June 2002

Zhou, Y. Q., F. S. Foster, D. W. Qu, M. Zhang, K. A.Harasiewicz, and S. L. Adamson. Applications for multi-frequency ultrasound biomicroscopy in mice from implanta-tion to adulthood. Physiol Genomics 10: 113–126, 2002. Firstpublished June 18, 2002; 10.1152/physiolgenomics.00119.2001.—A new multifrequency (19–55 MHz) ultrasoundbiomicroscope with two-dimensional imaging and integratedDoppler ultrasound was evaluated using phantoms andisoflurane-anesthetized mice. Phantoms revealed the biomi-croscope’s lateral resolution was between 50 and 100 �m,whereas that of a conventional 13 MHz ultrasound systemwas 200–500 �m. This difference was apparent in the mark-edly higher resolution images achieved using the biomicro-scope in vivo. Transcutaneous images of embryos in pregnantmice from �2 days after implantation (7 days gestation) tonear term (17.5 days) were obtained using frequencies from25 to 40 MHz. The ectoplacental cone and early embryoniccavities were visible as were the placenta and embryonicorgans throughout development to term. We also evaluatedthe ability of the biomicroscope to detect important featuresof heart development by examining embryos from 8.5 to 17.5day gestation in exteriorized uteri using 55 MHz ultrasound.Cardiac looping, division of the outflow tract, and ventricularseptation were visible. In postnatal imaging, we observed theheart and kidney of neonatal mice at 55 MHz, the carotidartery in juveniles (�8 g body wt) and adults (�25 g body wt)at 40 MHz, and the adult heart, aorta, and kidney at 19 MHz.The coefficient of variation of carotid and aortic diametermeasurements was 1–3%. In addition, blisters in GRIP1 �/�embryos and aortic valvular stenosis in two adults werereadily visualized. Using image-guided Doppler function, lowblood velocities in vessels as small as 100 �m in diameterincluding the primitive heart tube at day 8.5 were measur-able, but high blood velocities (�37.5 cm/s) such as in theheart and large arteries in late gestation and postnatal lifewere off-scale. Accurate cardiac dimension measurements

were impeded by poor temporal resolution (4 frames/s). Insummary, the multifrequency ultrasound biomicroscope is aversatile tool well suited to detailed study of the morphologyof various organ systems throughout development in miceand for hemodynamic measurements in the low velocityrange.

Doppler; heart; placenta; embryo; newborn; development;aorta; umbilical cord; kidney; mouse; biomicroscope; carotidartery; juvenile; blood velocity; heart valves

PHENOTYPIC EVALUATION of genetically altered mice pre-sents an important rate-limiting step for physiologistsworking toward assigning function to each of the�30,000 genes in the genome. Methods suitable onlyfor adult mice are insufficient because genetic modifi-cations often cause embryonic or perinatal lethality,thus appropriate technology for rapid and accurateassessment of phenotype in mice at all stages of devel-opment is required to accelerate progress in this task.

The current paper evaluates the application of anewly developed multifrequency ultrasound biomicro-scope that was designed specifically for imaging andhemodynamic evaluation of mice. Like other ultra-sound systems, the biomicroscope provides noninva-sive and real-time images, as well as Doppler bloodvelocity measurements, and is suitable for serial invivo studies, for instance, to study phenotypic expres-sion during development or following gene induction orablation. In addition, ultrasound biomicroscopy is theonly small-animal imaging method suitable for image-guided injections because the images are available inreal time and the imaging apparatus does not interferewith access to the mouse. Although conventional ultra-sound systems that use frequencies from 7 to 15 MHzhave proven useful for imaging and Doppler studies ofcardiac function in adult mice (6) and for Dopplerstudies in mouse embryonic hearts (11), image resolu-tion and Doppler sample volume size are inadequatefor detailed study of morphology and hemodynamics in

Article published online before print. See web site for date ofpublication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: S. L. Ad-amson, Samuel Lunenfeld Research Institute, Mount Sinai Hospital,Rm. 138P, 600 Univ. Ave., Toronto, Ontario, Canada, M5G 1X5(E-mail: adamson@mshri.on.ca).

Physiol Genomics 10: 113–126, 2002.First published June 18, 2002; 10.1152/physiolgenomics.00119.2001.

1094-8341/02 $5.00 Copyright © 2002 the American Physiological Society 113

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

the developing heart of embryos and juveniles (16). Thebiomicroscope takes advantage of the limited penetra-tion depths required for mouse imaging (�5–15 mm)by exploiting high frequencies so that higher resolutionimages and smaller Doppler sample volumes can beachieved. Image resolutions are increased �10-fold,which is nearly sufficient to compensate for the �20-fold difference in linear dimensions between adult miceand humans (e.g., aortic diameter �1–1.5 mm in mice,2–3 cm in humans).

The multifrequency biomicroscope is a new imagingsystem based on the technology of the 40 MHz proto-type instrument used extensively by Turnbull and col-leagues for imaging and Doppler studies of mouseembryos between 9.5 and 14.5 days of gestation andneonates between birth and 7 days of age (1, 8, 19, 32,33), and for image-guided microinjection of mouse em-bryos (10, 17, 22, 32). Unlike the prototype, the multi-frequency biomicroscope operates at a range of fre-quencies between 19 and 55 MHz and so can be usedfor imaging mice over a wider range of development. Inaddition, this system has integrated Doppler capabilityin the high-frequency range and a relatively smallsample volume. This study was conducted to explorethe potential applications and evaluate the limitationsof the multifrequency ultrasound biomicroscope for ex-amining morphology and hemodynamics during devel-opment of mice.

METHODS

Experiments were approved by the animal care committeeof Mount Sinai Hospital and were conducted in accord withguidelines established by the Canadian Council on AnimalCare.

The biomicroscope. A multifrequency ultrasound biomicro-scope (model VS40; VisualSonics, Toronto, Canada; http://

www.visualsonics.com) was used in the current study.The biomicroscope has a single transducer with a nominalcenter frequency of 40 MHz, a diameter of 3 mm, and a focallength of 6 mm. The transducer design has previously beendescribed and characterized (9). One of its features is that ithas a wide bandwidth of �120%. This allows the transducerto operate over the frequency range from 16 MHz to 64 MHz.The software of the multifrequency biomicroscope allows theuser to select 19, 25, 40, or 55 MHz as the driving frequencyfor the transducer from a menu on the monitor. Drivingfrequency can be changed during imaging sessions. In thisstudy, the transducer was mechanically scanned at �4frames/s to create an 8 � 8 mm two-dimensional (2D) image.The B-scan low-pass filter was set to 80 MHz, and thehigh-pass filter was “open” (Table 1). The transducer washeld stationary to obtain Doppler flow velocity spectra in realtime from a sample volume located within the 2D image. Thepulsed Doppler default settings [7 cycles per pulse, 40 MHzcenter frequency, 10 kHz pulse repetition frequency (PRF)]were used. The Doppler sample volume size was dependenton the number of cycles per pulse and the insonation fre-quency as shown in Table 2. The maximum measurablevelocity was 37.5 cm/s, and this would be achieved using aPRF of 20 kHz and an operating frequency of 20 MHz (Table1). Additional technical specifications and menu options areshown in Table 1.

Image resolution and comparative performance. The reso-lution in the lateral direction of the scanner operating at 19,25, 40, or 55 MHz was measured in vitro with a calibratedhydrophone. The resolution in the axial direction was calcu-lated using assumed filter settings for a pulse echo band-width of 50% (Table 3). In two phantom studies, the imageresolution of the biomicroscope was compared with that of aconventional ultrasound system, the Sequoia C256 with a

Table 1. Instrument specifications for themultifrequency biomicroscope

Feature Specification

B-scan frequency 19, 25, 40, 55 MHzFrame rate 2, 4, 8 Hz*Scan format bidirectionalLine spacing 15.6 �mB-scan high-pass filtering open, 5, 10, 30 MHzB-scan low-pass filtering 40, 50, 70, 80 MHz

Doppler mode pulsedDoppler frequency 20–55 MHzPulse repetition frequency 1–20 kHzMax. unaliased velocity at 0° 37.5 cm / s†Minimum velocity �1 mm/sRadio frequency cycles per pulse 2–16Doppler low-pass filtering 100 Hz to 25 kHz (steps

of 100 Hz)Doppler high-pass filtering 6, 60, 100, 320 Hz

Analog-to-digital conversion 125 MHz, 12 bitsMicropositioning x, y, z � 3 �m

*The maximum frame rate was 4 Hz at the time of study. †Themaximum unaliased velocity measurable at 0° insonation angle isobtained at a pulse repetition frequency of 20 kHz and a Doppleroperating frequency of 20 MHz.

Table 2. Doppler sensitivity pattern

n

fo � 20 MHz, Rlat � 104 �m fo � 40 MHz, Rlat � 68 �m

L, �m DSV, �10�3 �l L, �m DSV, �10�3 �l

4 128 1.09 64 0.2338 257 2.18 128 0.466

12 385 3.27 192 0.69916 513 4.36 257 0.932

Axial sample volume length (L) and Doppler sample volume (DSV)for different pulse lengths at 20 and 40 MHz. The DSV is modeled asa cylinder with diameter given by the measured lateral resolution(Rlat) and length defined by the number of cycles in the transmittedwaveform. The length of the pulse is given by L � (1⁄2) (cn/1.2 fo)where c is the speed of sound, fo is the operating frequency, and n isthe number of cycles in the Doppler radiofrequency (rf) transmitwaveform. The Doppler sample volume is then given by DSV � L(Rlat/2)2 where Rlat is the measured lateral resolution.

Table 3. Spatial resolution as a function of frequencyfor the multifrequency biomicroscope

Nominal Frequency,MHz

Lateral Resolution,�m

Axial Resolution,�m

19 104.2 81.125 88.2 61.640 68.2 38.555 62.5 28

Lateral resolution was measured with a calibrated hydrophone.Axial resolution assumes filter settings for a pulse echo bandwidth of50%.

114 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

model 15L8 transducer operating at 13 MHz (Acuson, Moun-tainview, CA; http://www.acuson.com), that is currentlyused for echocardiography in mice at the University of To-ronto and elsewhere (see Ref. 6). A phantom with parallelglass filaments separated by 1,000, 500, 200, 100, and 50 �mwas imaged in the short-axis view within the focal zone of thetransducer. The array of filaments was either perpendicularor parallel to the ultrasound beam to determine lateral oraxial resolution, respectively. The mean percent discrepancy[(�xm � xa�)/xa � 100, where xm is the measurement and xa isthe actual dimension] was determined from five caliper mea-surements per separation distance per frequency for bothultrasound systems. Image distortion was tested by imaginganother phantom with two parallel, fluid-filled cylindricalcavities 1,000 �m and 460 �m in diameter in an agaroseblock. Both short- and long-axis views were obtained.

Mouse embryonic imaging. Timed-pregnant ICR (CD-1)mice (Harlan Sprague Dawley, Indianapolis, IN) were stud-ied. Day 0.5 of pregnancy was defined as 12 noon on the daya vaginal plug was found after overnight mating. Implanta-tion in this strain occurs approximately day 5.0, and day 18.5is full term. During studies, all mice were lightly anesthe-tized with isoflurane in oxygen by face mask and werewarmed using a heated pad and heat lamp. Heart rate andrectal temperature of adult mice were monitored (modelTHM100; Indus Instruments, Houston, TX), and heatingwas adjusted to maintain rectal temperature between 36 and38°C.

For transcutaneous embryonic imaging, 14 pregnant micewere studied once between 6.5 and 17.5 days of gestation,and serial observations were conducted daily from 6.5 to 17.5days of gestation in another 2 pregnant mice. Once anesthe-tized, the mouse abdomen was shaved and further cleanedwith a chemical hair remover to minimize ultrasound atten-uation. After prewarming the ultrasound gel, an outer ring ofthick gel (Aquasonic 100; Parker Laboratories, Orange, NJ)was filled with a thinner gel (EcoGel 100; Eco-Med Pharma-ceutical, Mississauga, Ontario, Canada) over the region ofinterest, to provide a coupling medium for the transducer.Transcutaneous imaging and Doppler studies of embryoswere usually performed at 40 MHz. However, when imagingdeep embryos or large embryos in late gestation, a lowerfrequency (e.g., 25 MHz) was used to obtain greater depth ofpenetration. We usually imaged two to four embryos in ses-sions limited to �1-h duration. Sometimes a few embryoswere studied thoroughly, or a greater number were examinedmore selectively. Usually not all embryos in a litter wereimaged. Some might be too deep or in an inappropriateorientation to achieve high-quality images in the desiredplane of view. We limited time to minimize anesthetic expo-sure in this serial study, because anesthesia may adverselyaffect embryonic development (4).

We imaged one pregnant GRIP1 /� mouse at 12.5 and13.5 days of gestation to screen embryos for the skin blisterphenotype characteristic of GRIP1 �/� mutants (5), to illus-trate the use of the transcutaneous approach in phenotypedetection. In an additional experiment on one pregnant ICRmouse, an embryo at 10.5 days gestation was imaged trans-cutaneously using the multifrequency biomicroscope (40MHz), and the same embryo was then imaged with theconventional ultrasound system (13 MHz) to directly com-pare the imaging capabilities of the two systems.

Transuterine embryonic imaging was conducted in 10pregnant mice after surgical exteriorization of the uterus.One mouse was studied on each day from 8.5 to 17.5 days ofgestation. In each isoflurane-anesthetized pregnant mouse,one uterine horn (containing 5–7 embryos) was exposed

through an elliptical hole cut into the bottom of a petri dishwhich was filled with warm, circulating phosphate-bufferedsaline (PBS). To prevent leakage, the petri dish was sealed tothe cleanly shaven skin of the maternal abdomen usingdouble-sided tape. Four to six 6-0 silk sutures through themyometrium were tethered to a rubber ring at the edge of thedish to hold the uterus stationary. Care was taken to avoidany tension or pressure on the uterine vessels. Indomethacin(300 �M in PBS) was added to the circulating PBS in thepetri dish to inhibit spontaneous uterine contractions (36),which could move embryos through the plane of view. Ma-ternal body temperature was maintained at 36–38°C withthe use of a heating pad, a lamp, and a warmer for thecirculating PBS. The structure of the embryonic heart wasobserved using 55 MHz ultrasound, and Doppler flow wave-forms were recorded using 40 MHz ultrasound. Several sitesof interest within the heart were studied. Two to four em-bryos were observed in sessions limited to �1.5 h. At the endof the study, mice were killed while still anesthetized.

Postnatal mouse imaging. Mice were anesthetized withisoflurane as described above. The heart and kidney wereimaged at 55 MHz in six mouse neonates (ICR, wild type)at 1–3 days after birth and at 19 MHz in five adult mice(ICR, wild type). During all imaging, the position of themouse was adjusted to place the structure of interest �6 mmfrom the transducer, so it would be within the transducer’sfocal zone. Each experiment in postnatal mice lasted forabout 30–45 min.

In an additional experiment, the hearts of two adult ICRmice (7 wk, 27 and 34 g body wt) were imaged transcutane-ously using the multifrequency biomicroscope (19 MHz) andthen imaged with the conventional ultrasound system (13MHz) to directly compare the imaging capabilities of twosystems.

Two C57Bl6/C3H hybrid mice were imaged at 19 MHzto test the feasibility of using the system as a secondaryscreen to investigate the cause of the high peak aortic bloodvelocities in these mice. Peak blood velocities in the ascend-ing aorta were �200 cm/s, which is twice normal. The ab-normal peak velocities were detected in a primary high-throughput screen in our mouse mutagenesis program(http://www.cmhd.ca).

Another five adult mice [ICR, wild type, mean 28 � 0.8(SE) g] were imaged at 19 MHz daily over 4 consecutive daysto test the reproducibility of aortic inner diameter measure-ments using the biomicroscope’s on-screen calipers. Aorticdiameter during systole (i.e., when the aortic valve was open)was measured at the aortic annulus in the long-axis view andalso at the level of the proximal ascending aorta just distal tothe aortic sinuses in both long- and short-axis views. On eachday, six to eight successive diameter measurements wereobtained and were averaged for each site. The coefficients ofvariation within sessions and in the session means over 4days were calculated. Inner diameters of the common carotidartery in long-axis views were measured in additional fourjuvenile (8.1 � 0.6 g body wt) and six adult (25 � 2 g body wt)wild-type C57Bl6/J mice. Images were saved when vesselswere near their largest diameter (i.e., systole). The mean andcoefficient of variation were determined from six to eightsuccessive carotid diameter measurements.

For Doppler recordings, the sample volume was positionedover the structure of interest in the 2D image, and then thebiomicroscope was switched to Doppler mode. Doppler veloc-ity spectra appeared in real time on the biomicroscope’sscreen but were transferred to a second computer (DopplerSignal Processing Workstation; Indus Instruments, Houston,

115ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

Texas) for quantitative measurement of velocities and timeintervals.

RESULTS

In vitro validation studies. The lateral resolution ofthe multifrequency biomicroscope measured in vitrousing a hydrophone is shown in Table 3. Lateral reso-lution ranged from 62 �m at 55 MHz to 104 �m at 19MHz. These results are in good agreement with theimage resolution of the biomicroscope determined us-ing a glass filament phantom, which showed the lateralresolution of the biomicroscope was between 50 and100 �m at all frequencies (Fig. 1). The percent discrep-ancies in caliper measurements of filament separationdistances were 0% at 200-, 500-, and 1,000-�m separa-

tions and 10% at 100 �m for all frequencies. The lateralresolution of the conventional ultrasound system wasbetween 200 and 500 �m. The mean percent discrep-ancy in caliper measurements was 3% at 1,000 �m and4% at 500 �m. The 200-�m separation was not detect-able (Fig. 1).

The calculated axial resolution of the multifrequencytransducer is shown in Table 3. The axial resolutionranged from 28 �m at 55 MHz to 81 �m at 19 MHz. At40 and 55 MHz, the axial resolution of the biomicro-scope determined using a glass wire phantom in vitro(Fig. 1) was somewhat less than calculated (Table 3) inthat the 50-�m separation was not detectable at thesefrequencies. However, the 100-�m separation distancewas detectable at all frequencies as predicted. The

Fig. 1. Images of phantoms obtained using the conven-tional ultrasound system (left column; scale bars are 1mm) and the multifrequency biomicroscope operatingat 40 MHz (right column; smallest divisions on scalebars are 100 �m) are shown at approximately samescale. In all cases, the area of interest was placed in thefocal zone of the transducer. The same glass filamentphantom was imaged in the lateral direction in A and Band the axial direction in C and D. Crosshairs show thecalipers of the conventional ultrasound system (in Aand C), and plus signs show the calipers as aligned bythe operator on the monitor of the multifrequencybiomicroscope (B, D, F, and H). As shown in B, filamentseparation distances were 1,000 �m (caliper markers 1and 2), 500 �m (caliper markers 2 and 3), 200 �m(caliper markers 3 and 4), and 100 �m (caliper marker4 and the fifth unlabeled marker). The final filament is50 �m further to the right, but this separation intervalcannot be resolved by the multifrequency biomicroscopein either the lateral (B) or axial (D) direction. In theaxial direction, the most distant marker is not detectedby the multifrequency biomicroscope at the low gainused for this image (D), whereas it is detectable usingthe conventional ultrasound system (C). With the con-ventional ultrasound system, only the first 2 intervalscan be resolved (1,000 �m and 500 �m) in the lateraldirection (A), whereas intervals �100 �m can be re-solved in the axial direction (C). In F, the circular crosssections of the cylindrical cavities (1,000 �m and 460�m diameter) are easily detected and measured in thelateral and axial directions using the multifrequencybiomicroscope, whereas lateral echoes are difficult todetect with the conventional ultrasound system (E). Inlongitudinal section, the phantom appears as twostraight, parallel echoes when viewed by both ultra-sound systems (G and H).

116 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

percent discrepancy in caliper measurement in theaxial direction was 10% at 100 �m at all frequenciesand 4% or less for greater separation distances at allfrequencies, confirming the accuracy of caliper mea-surements. The axial resolution of the conventionalultrasound system, like that of the multifrequencybiomicroscope, was between 50 and 100 �m. The per-cent discrepancy in caliper measurement in the axialdirection on the Sequoia apparatus was 24% at 100 �mand 3% or less for greater separations.

Images of cylindrical cavities cast in agarose showedthat the biomicroscope introduced negligible image dis-tortion. The biomicroscope generated images of thecylindrical cavities at all frequencies (19–55 MHz)that, when viewed in cross section (i.e., short axis) were

circular and when viewed longitudinally (i.e., long axis)had straight, parallel sides as expected (Fig. 1). Likethe biomicroscope, the conventional ultrasound systemshowed little image distortion of the cylindrical cavitiesviewed longitudinally. Unlike the biomicroscope, theechoes generated by the lateral walls of the cylindricalcavities were not visible in cross-sectional views so thatcircular cross sections could not be confirmed (Fig. 1).

Comparison of biomicroscopy and conventional ul-trasound imaging in mice. Transcutaneous images ofthe same mouse embryo, and same adult mouse heartwere obtained using the multifrequency biomicroscopeand the conventional ultrasound system (Fig. 2). The8 � 8 mm image generated by the multifrequencybiomicroscope was big enough to include the entire

Fig. 2. The same pregnant mouse at day 10.5 of gestation (A–C) and the same adult mouse heart (D–I) were viewedtranscutaneously using the conventional ultrasound system (13 MHz; 1st and 2nd columns) and using themultifrequency biomicroscope (3rd column). The 1st column (A, D, G) shows the image as viewed on the monitorof the conventional ultrasound system; the 2nd column (B, E, H) shows an enlargement of the region of interest sothat the scale is the same as that for images obtained using the multifrequency biomicroscope in the 3rd column(C, F, I). Shown are long-axis sections through embryos in the 1st row, through the adult left ventricle in the 2ndrow, and through the adult aortic root in the 3rd row. The multifrequency biomicroscope was operated at 40 MHzfor embryo imaging and 19 MHz for adult imaging. Scale bars in 2nd column � 1 mm; smallest divisions on scalebars in 3rd column � 100 �m. Am, amniotic cavity; Ao, aorta; AV, aortic valve; B, brain; CA, common atrium, CV,common ventricle; Em, embryo; En, endocardium; LV, left ventricle; OFT, outflow tract; Pl, placenta; S, skin.

117ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

conceptus at 10.5 days gestation (Fig. 2). The embryo,amniotic cavity, and placenta were visible, as were thecerebral ventricles, the common atrium, common ven-tricle, and outflow tract of the embryonic heart. Car-diac movement was detectable despite the low framerate (4 frames/s) relative to the high embryonic heartrate (3–4 beats/s). The scan distance and the depth ofpenetration of the conventional ultrasound system wasmuch greater, so that several embryos could be visual-ized in one image frame (�2.5 cm � 2 cm). Embryoswere visible within the amniotic cavity, and althoughdetails of the embryonic structure were not detectable,the beating of the heart was more readily detected thanwith the biomicroscope due to the much higher framerate of the conventional system.

The heart of the adult mouse completely filled theimage frame of the biomicroscope in long-axis view(Fig. 2). Penetration depth was just sufficient to visu-alize the posterior wall of the left ventricle of thisnormal mouse. The mitral, aortic, and pulmonaryvalves were visible, and the diameters of the pulmo-nary artery and aorta were readily measurable. How-ever, despite the high spatial resolution of the images,the relatively low frame rate of the biomicroscope leadto distortion artifacts, especially in phases of the car-diac cycle when the ventricular wall was moving rap-idly, and it was also not possible to capture images forcaliper measurements at precise time points duringthe cardiac cycle. The conventional ultrasound systemhad much better temporal resolution, so cardiac move-ments did not introduce distortion artifacts, but it wasnot possible to see the mitral or aortic valves, so thesestructures could not be used to detect phases of thecardiac cycle. However, spatial resolution was suffi-cient to visualize the left ventricular myocardium andendocardium and the ascending aorta, so ventricularwall thickness and aortic diameters were measurableas shown previously (6, 35). The pulmonary artery wasdifficult to visualize even in short-axis view in contrastto the biomicroscope, where it was readily imaged. Thescan length and depth of field of the conventionalsystem were more than sufficient to visualize the en-tire heart of the adult mouse (Fig. 2).

Embryonic imaging. The multifrequency biomicro-scope was used at 40 MHz to image early postimplan-tation development using a noninvasive, transcutane-ous approach. At 7 days of gestation, regions where thecross section of the uterus was enlarged were clearlyvisible within the maternal abdomen (Fig. 3A). Theenlarged regions were relatively dark. Near the centerof each enlargement was a small, echo-free region thatis likely the proamniotic cavity of the developing em-bryo (20). It was surrounded by a relatively brightregion known to be populated by embryonic tropho-blast giant cells that invade the maternal deciduaduring implantation in the mouse (20). By 7.5 days,three dark regions were visible within the conceptus(Fig. 3B). They likely correspond to the ectoplacental,amniotic, and exocoelomic cavities of the developingembryo (20). At �8.5 days, the embryo and amnioticmembrane were visible, and the allantois could be seen

emerging from the embryo and approaching the ecto-placental cone (where the chorio-allantoic placentalater develops; Ref. 20) (Fig. 3C). The implantation sitelooked similar whether viewed in vivo using ultra-sound (Fig. 3C) or in histological sections (Fig. 3D).Features were visible at even higher resolution whenthe uterus was exteriorized and 55 MHz ultrasoundwas used (Fig. 3, E and F). At this stage, a pulsatileblood velocity signal was first detectable in the hearttube, but no signal was detected within the allantois(Fig. 3, E and F).

The multifrequency biomicroscope proved useful forimaging many embryonic structures including theheart from the stage at which the embryo first dependson cardiovascular function for survival (�9.5–10.5days gestation) through to term using the transcuta-neous approach. Relatively superficial embryos wereimaged at 40 MHz, and deeper embryos (a more com-mon problem near term) were imaged at 19 or 25 MHz.On day 9.5, the amniotic membrane, amniotic and yolksac cavities, brain, cerebral ventricles, and heart werevisible using transcutaneous imaging (Fig. 4A). Laterduring embryonic development, many other structurescould also be visualized using transcutaneous imagingincluding the developing paw and forelimb, eyes, lung,liver, kidney, vertebrae, and veins (Fig. 4, B–F). Theskin blister phenotype observed in GRIP1 �/� em-bryos (5) was also readily detectable by the transcuta-neous approach (Fig. 5). The umbilical cord and pla-centa were visible and umbilical Doppler blood velocitywaveforms were detectable from day 9.5 of gestation.From day 10.5, Doppler blood velocity waveforms couldbe recorded separately from the umbilical artery andvein.

Higher resolution images of the developing heartwere obtained by exteriorizing the uterus and imagingat 55 MHz (Fig. 6). As observed previously (28), thespace within the embryonic heart and blood vesselsviewed at �14.5 days gestation appeared bright onultrasound images, whereas this space generally ap-pears dark in postnatal subjects when viewed at con-ventional frequencies (e.g., 7.5–15 MHz). This waslikely due to higher blood echogenicity caused by theshort wavelength of high-frequency ultrasound and therelatively large size of embryonic blood cells. Red cellnucleation may also be a factor, because blood ap-peared less echogenic in embryos near term (e.g., at16.5 days, Figs. 6H and 4E), when red cells are stilllarge but are no longer nucleated (25). On day 9.5, theU-shaped heart tube was clearly visible and Dopplerblood velocity waveforms could be recorded separatelyfrom the inflow and outflow regions of the heart tube(Fig. 6, A–D). On day 11.5, it was possible to detect theprocess of division of the outflow tract into the ascend-ing aorta and main pulmonary artery (not shown). Onday 12.5, the separation of the aorta and main pulmo-nary artery appeared complete (not shown), but theinterventricular septum was visibly incomplete, andflow streams from both ventricles could be seen enter-ing the aorta (Fig. 6F). By day 13.5, the embryonicventricles were fully septated, the atrioventicular

118 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

valves were visible, and the heart had a mature fetalform (Fig. 6G). After day 15.5, the heart chambersbegan to darken in the ultrasound image, and theventricular wall, endocardium, and septum becameeasier to discern (Fig. 6H). The improved contrast afterday 15.5 meant that ventricular chamber dimensionsand wall thickness measurements became feasible.

Postnatal imaging. After birth, the neonatal heartwas still sufficiently superficial to image at 55 MHz,whereas 19 MHz was required to get sufficient pene-tration depth to image the adult mouse heart (Fig. 7).In both cases, resolution was adequate for measuringdimensions of the ventricular wall and chambers, andthe mitral, aortic, and pulmonary orifices, and theascending aorta and main pulmonary artery and theirmain branches (Fig. 7). In adults, the diameter of theleft and right coronary arteries were measurable (Fig.7F), but coronary blood velocities were not, becausevessel movement during cardiac contractions was ex-cessive. The low scan rate of the transducer (4frames/s) relative to the high heart rate of mouseneonates (e.g., 240 min�1, which is 1 beat/frame) andadults (e.g., 480 min�1, which is 2 beats/frame) meant

that cardiac movement during a single scan often dis-torted the images. In addition, end-diastolic and end-systolic images were difficult to pinpoint. In postnatalmice, blood velocities in the aorta, pulmonary artery,and heart were too high to be recorded using the multi-frequency biomicroscope. On the other hand, mice have alarge thymus that provides a good acoustic window forclearly imaging the ascending aorta, aortic arch, andits branches. Moreover, with the exception of a shad-owed region near the thoracic inlet, the entire course ofleft and right carotid arteries could be readily followedfrom their origin past their bifurcation into the inter-nal and external carotid arteries (Fig. 7, C and D).

Inner diameters of the aorta and common carotidarteries were measured with the multifrequency biomi-croscope. Aortic diameter was measured during systole(i.e., when aortic valves were open) using 19 MHzultrasound in adult mice (25–30 g body wt). Systolicdiameter is the most relevant when calculating volumeflow rate at this site, because it is during systole thatcardiac ejection (i.e., flow) occurs. Bright echoes corre-sponding to the near and far walls appeared only to beobserved when the aorta was viewed with the image

Fig. 3. Embryonic development frompostimplantation to the beating heart.Transcutaneous cross-sectional images ofthe pregnant uterus at 40 MHz are shownat day 7.0 (A), day 7.5 (B), and day 8.0 (C)of gestation. Inside the uterus, the deciduais relatively dark, and the eccentrically lo-cated conceptus is surrounded by a regionof brighter echoes. The fluid-filled cavitiesof the conceptus appear black. The implan-tation site at day 8 looks remarkably sim-ilar in C using ultrasound and D usinghistology. In E and F, transuterine cross-sectional images of the exteriorized uterusat 55 MHz are shown at day 8.5 of gesta-tion. In E, the embryo is viewed in crosssection at the level of the heart. Open em-bryonic head folds are visible to the right ofthe heart, as is the amniotic membranethat surrounds the embryo. The Dopplersample volume (indicated by the rectangleon the vertical line) has been placed overthe heart and the recorded Doppler signalis shown below. The signal is negative in-dicating that flow is away from the trans-ducer. Blood velocity is �5 cm/s assumingflow is parallel to the Doppler beam (indi-cated by the vertical line). In F, the Dopp-ler sample volume has been placed overthe allantois, but no blood velocity signalwas detectable (as shown in the Dopplerrecording below). The smallest division onthe scale bar to the right of all ultrasoundimages is 100 �m, and on the time scalebeneath Doppler recordings is 0.1 s. In Eand F, the y-axis is the Doppler shift fre-quency in Hz (0 to �2,500 Hz is indicatedby the vertical white bar to the left). Al,allantois; Am, amniotic cavity; AM, amni-otic membrane; D, decidua; EC, ectopla-cental cone; Em, embryo; EPC, ectoplacen-tal cavity; Ex, exocoelomic cavity; H, heart;PA, proamniotic cavity; S, maternal skin;Ut, uterus.

119ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

plane exactly though the midline of the aortic lumen(Fig. 7G). The easiest and most reproducible measure-ments were obtained from longitudinal views of theproximal ascending aorta just distal to the aortic si-nuses (Table 4, Fig. 7G). Measurements were alwayspossible at this site, but were sometimes difficult in theaortic annulus, probably due to higher attenuation andless optimal angulation. The coefficient of variation foraortic diameter measurement was between 1 and 3%(Table 4). The coefficient of variation in inner diametermeasurements of the common carotid artery was3–3.5% in juvenile and adult mice (Table 4).

In neonates, the whole kidney was visible within thedimensions of one image, and blood flow velocity was

measurable from the renal hilus region where therenal artery enters the kidney (Fig. 8, A and B). Inadults, only the cortex of the kidney was sufficientlysuperficial to be imaged in detail. However, flow veloc-ity waveforms could be obtained from the small termi-nal vessels in the cortical region (Fig. 8, C and D).

The multifrequency biomicroscope was used to im-age the ascending aorta and heart of two adult micethat had abnormally high aortic flow velocities. Bothmice had aortic valves that were irregular in shape andhyperechoic and which did not open normally duringsystole (Fig. 9, A and B). In addition, the ascendingaorta appeared to be abnormally dilated in both mice(diameters �2.5 and �2.0 mm). Histological examina-

Fig. 4. Embryonic structures visualized bytranscutaneous imaging at 40 MHz from day9.5 to 16.5 of gestation. A: a longitudinalcross-sectional view of an embryo at day 9.5showing the cephalic vesicle within the brainon the left, the bright heart region near thecenter, and the lower body to the right. B: theforelimb and paw, and placenta of a mouseembryo at day 11.5. C: embryonic eyes andbrain at day 13.5. D: cross section of theembryonic head at day 13.5 showing cerebralventricles in the brain. E: blood in the hepaticsegment of inferior vena cava of the embryoat day 16.5 of gestation appears dark,whereas blood-filled spaces earlier in gesta-tion appear bright as shown in A above. Theliver and lung are also visible in E. F: thelung, liver, kidney, and vertebrae of an em-bryo at day 16.5 viewed in longitudinal sec-tion. The smallest division on the scale bar tothe right of all ultrasound images is 100 �m.ASLV, anterior portion of superior horn oflateral cerebral ventricle; B, brain; CV, ce-phalic vesicle; E, eye; Em, embryo; FV, fourthcerebral ventricle; H, heart; IVC, inferiorvena cava; Li, liver; Lu, lung; P, paw; Pl,placenta; RK, right kidney; S, maternal skin;TV, third cerebral ventricle; Ut, uterus; V,vertebrae.

120 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

Fig. 5. Two-dimensional images of GRIP1 �/�mouse embryos. A: a skin blister (arrow) on theembryonic head at 12.5 days gestation is shown(imaged at 40 MHz). B: skin blisters (arrows) on thesurface of the lower body and tip of the limb bud areshown at 13.5 days gestation (imaged at 55 MHz).Smallest division on scale bar � 100 �m. E, eye; B,brain; L, limb; S, maternal skin.

Fig. 6. The embryonic heart from day 9.5 to 16.5of gestation viewed at 55 MHz through the exte-riorized uterus. A: the U-shape of the tubularembryonic heart at day 9.5 is visible. The Dopp-ler sample volume (rectangle on vertical line) hasbeen placed over the atrioventricular canal in A,and the corresponding Doppler blood velocitywaveform is shown in B. B: ventricular inflowvelocity in the atrioventricular canal is unidirec-tional and negative, indicating all flow is awayfrom the transducer. Peak velocity is �3 cm/s,assuming flow is parallel to the Doppler beam.The waveform has a double peak and therefore issimilar in shape to that of postnatal animals. InC, the Doppler sample volume has been placedover the outflow tract, and the correspondingDoppler waveform is shown in D. D: blood veloc-ity in the ventricular outflow tract is primarilyunidirectional (toward the transducer) with apeak velocity of �5 cm/s, assuming flow is par-allel to the beam. E: transverse view of the day10.5 embryonic heart. The common atrium, com-mon ventricle, and common atrioventricular ca-nal are visible. F: a transverse view of the heartat day 12.5. The ventricular septum is formingbut incomplete (the septum is the dark regionvisible between the left and right ventricles).Flow streamlines from the left and right ventri-cles (i.e., echoes streaks) can be seen leading intothe aorta especially in enlarged inset (arrowshighlight streamlines). G: four-chamber view ofthe heart at day 13.5 showing complete ventric-ular septation and visible mitral valve leaflets.H: long-axis view of the ventricles on day 16.5.Blood in ventricular chambers appears dark rel-ative to earlier in gestation, and internal cardiacmorphology is now visible. The smallest divisionon the scale bar ultrasound images is 100 �m,and on the time scale beneath Doppler record-ings is 0.1 s. In B and D, the y-axis is the Dopplershift frequency in Hz (0 to �2,500 Hz is indicatedby the vertical white bar to the left). Ao, aorta;AVC, atrioventricular canal; CA, common atri-um; CV, common ventricle; LA, left atrium; Lu,lung; LV, left ventricle; MV, mitral valve; OFT,outflow tract; RA, right atrium; RV, right ventri-cle; Ut, uterus.

121ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

tion revealed marked aortic valve leaflet dysmorphol-ogy (Fig. 9C) characterized by markedly increasedamounts of myxomatous connective tissue in thestroma of the leaflets. The valve dysmorphology likelyresulted in aortic valvular stenosis, leading to highascending aortic blood velocities, and a poststenoticaortic dilatation.

DISCUSSION

Relative to existing ultrasound technology used formouse imaging, the main advance of the current biomi-croscope is that insonation frequency is adjustable over

a range of frequencies from 19 to 55 MHz. This fre-quency range enables the highest possible resolutiongiven the penetration depths required for imaging micethroughout development from implantation to adult-hood (�5 to �15 mm). The system would also besuitable for other applications with similar penetrationdepth and resolution requirements (e.g., other smallanimal species, superficial structures in larger ani-mals). The other main advance of the multifrequencybiomicroscope is the integration of pulsed Doppler ca-pability at these frequencies. Implementation of high-frequency Doppler permits low blood velocities such as

Fig. 7. Transcutaneous images of the heart andnearby vessels in postnatal mice. A: long-axis view ofheart at 3 days postnatal age imaged at 55 MHz. Theleft ventricular chamber and walls, mitral valve leaf-lets, aortic annulus, and left atrium are visible. B:short-axis view of heart at 2 days postnatal age in-sonated at 55 MHz, showing left and right ventricularchambers, ventricular walls, and interventricular sep-tum. C: long-axis view of left common carotid arteryfrom aortic origin to beyond the bifurcation into inter-nal and external carotid arteries; imaged with a trans-ducer frequency of 40 MHz in a 3-wk-old mouse. D:long-axis view of the left ventricular outflow tract,ascending aorta, and the innominate artery (firstbranch from the aortic arch); imaged at 19 MHz in anadult mouse. Length of ascending aorta from aorticannulus to the origin of the innominate is 3.72 mm. E:transverse view showing the aorta in cross section andthe pulmonary trunk and pulmonary arteries in lon-gitudinal section in an adult mouse insonated at 19MHz. Arrow points to a pulmonary valve leaflet. F:view of the left coronary artery branching off at theaortic root of an adult mouse viewed at 19 MHz.Coronary artery diameter is 0.32 mm at the site wherethe measurement calipers are shown. G: long-axisview of the left ventricular outflow tract of an adultmouse insonated at 19 MHz showing the ascendingaorta with caliper markers in place. The left ventricle,left atrium, and an aortic valve (arrow) leaflet are alsovisible. H: short-axis view of an adult mouse heartinsonated at 19 MHz showing the right ventricle, theleft ventricular chamber and wall, the interventricu-lar septum, and papillary muscles (arrow). The small-est division on the scale bars is 100 �m. AA, ascendingaorta; Ao, aorta; IA, innominate artery; LA, left atri-um; LCA, left coronary artery; LCCA, left commoncarotid artery; LV, left ventricle; LVOT, left ventricu-lar outflow tract; MPA, main pulmonary artery; RV,right ventricle; S, skin; Th, thymus.

122 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

in the early embryonic cardiovascular system and inthe microcirculation of mice at all ages to be studied.The same transducer is used for 2D imaging and pulsedDoppler, and the Doppler sample volume is small (e.g.,�70 �m lateral � 250 �m axial) relative to conventionalultrasound systems (e.g., �350 �m lateral � 1,000 �maxial). These capabilities allow the sample volume to belocated within small structures in 2D images.

Using phantoms, we showed that the lateral resolu-tion of a conventional ultrasound system operating at13 MHz was between 200 and 500 �m, and the axialresolution was between 50 and 100 �m. Given thesmall size of cardiac structures in adult mice and theeven smaller size of structures in mouse embryos, thisresolution is not ideal. Specialized clinical instruments

are available that use higher frequencies for imagingthe human eye (10 or 50 MHz; model P45; ParadigmMedical Industries, San Diego, CA; http://www.paradigm-medical.com), the human peripheral vascu-lature (20 MHz, model AU5; Biosound Esaote, Indianap-olis, IN; http://www.advanced-ultrasound.com), orthe human skin (20 MHz; EPISCAN, Longport Inter-national, Silchester, UK; http://www.longport-intl.com). However, these instruments do not supportDoppler ultrasound recordings, nor do they allow fre-quencies to be selected throughout the range bestsuited for mouse imaging (i.e., 19–55 MHz), so theseinstruments are less versatile than the multifrequencybiomicroscope.

It is not uncommon for mutant mice to die in theearly postnatal period or to first express cardiac phe-notypes as juveniles, but in vivo study in this age groupwas limited due to the dearth of technology. The mul-tifrequency biomicroscope enables the researchers toimage the mutant mice in vivo and assess functionaldefects in this age group.

The multifrequency biomicroscope can be used tovisualize intrauterine early postimplantation placen-tal and embryonic development (e.g., day 6.5 to 8.5).There was a close correspondence between histologyand ultrasound images at this stage, and histologicalartifacts (e.g., shrinkage during fixation and dehydra-tion), avoided by in vivo imaging approach, may ac-count for some of the differences. Very bright echoeswere located near the implantation site where large,polyploid trophoblast giant cells are known to be local-ized (24). These bright echoes strongly delineated the

Table 4. Intraluminal vessel diameter measurement

Vessel ViewDiameter, mm(mean � SD)

% Coefficient of Variation

WithinSession*

BetweenSessions†

Adult ICR mice, 25–30 g body wt, N � 5

Aorta long axis 1.411�0.029 1.3 2.1Aorta short axis 1.428�0.043 2.6 3.0Annulus long axis 1.485�0.042 3.0 2.9

Juvenile C57B16/J mice, 3 wk, 7–10 g body wt, N � 4

Carotid long axis 0.421�0.016 3.1 not done

Adult C57B16/J mice, 8 wk, 19–30 g body wt, N � 6

Carotid long axis 0.500�0.049 3.5 not done

*Variation for 6–8 measurements per session. †Variation in ses-sion means, one session per day over 4 days

Fig. 8. Kidney and renal blood velocity wave-forms in postnatal mice. A: long-axis view of theright kidney of a 2-day-old neonate imaged at 55MHz. Arrows show the border between the kid-ney and the liver. B: Doppler blood velocitywaveform recorded from the renal hilus regionof the kidney shown in A. Peak velocity is �4cm/s, assuming flow is parallel with the Dopplerbeam. C: short-axis view of the right kidney ofan adult mouse imaged at 19 MHz. Arrows showthe outer surface of the kidney. The hilus regionis too deep to visualize. The Doppler samplevolume (rectangle on vertical line) has beenplaced over a renal cortical artery to obtain theDoppler blood velocity waveform shown in D.Peak velocity is �6 cm/s, assuming flow is par-allel with the Doppler beam. The smallest divi-sion on the scale bar in ultrasound images is 100�m, and on the time scale beneath Dopplerrecordings is 0.1 s. In B and D, the y-axis is theDoppler shift frequency in Hz (0 to �2,500 Hz isindicated by the vertical white bar to the left).Li, liver; RK, right kidney, S, skin.

123ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

placental margin. Fluid-filled cavities within the devel-oping embryo were also visible. Thus the multifre-quency biomicroscope may be useful for studying mor-phological development of the embryo and placenta innormal pregnancies and in mutants that die in theearly postimplantation period.

The multifrequency biomicroscope was also able torecord morphology and hemodynamics in detail withinembryos between day 8.5 and 10.5 (�3–4 wk in thehuman; Ref. 27). At 40 MHz, images were similar tothose previously published using the 40 MHz prototypebiomicroscope (28). This period is important because itincludes critical developmental events. The embryoeverts so that the inner concave surface of the U-shaped embryo becomes the outer convex, dorsal sur-face, the open cephalic neural folds close to form thefluid-filled neural tube, and the heart is transformedfrom a straight tube to a looped heart with atrial andventricular septation underway. Furthermore, the em-bryo first becomes dependent on a functional placentaand cardiovascular system during this interval, and soit is an important stage to examine when searching forcauses of intrauterine lethality in mutant mice (7). Inthis regard, the larger image size (more embryos perview) and better time resolution (heart motion easier todetect) of the conventional ultrasound system makes itbetter than the biomicroscope for quickly evaluatingembryo number and viability and hence for determin-ing the gestational age of intrauterine lethality inmutant mice. The higher spatial resolution andsmaller image size of the multifrequency biomicroscopesuggests that it may be better suited to more detailedstudies such as investigating the cause of such deaths.

Conventional ultrasound systems are inadequate forimaging the internal anatomy of the embryonic mouseheart and the Doppler sample volume is too large toseparately record inflow and outflow waveforms inembryos throughout gestation (11, 16). In contrast,especially at 55 MHz, the resolution of the multifre-quency biomicroscope was sufficient to detect division

of the common outflow tract into the aorta and pulmo-nary artery, septation of the ventricle, as well as thechange from a common atrioventricular canal to twoseparate ventricular inflow tracts. The multifrequencybiomicroscope can also be used to assess cardiovascularfunction using Doppler velocity waveforms throughoutthis period of development. It is noteworthy that atgestational ages �15.5 days, the internal anatomy ofthe embryonic heart is still somewhat difficult to dis-cern because embryonic blood earlier than this stage isechogenic, so there is poor contrast between the myo-cardium and the cardiac chambers. On the positiveside, the echogenic blood enabled visualization of in-tracardiac flow streams, which was helpful in identify-ing flow channels and for choosing appropriate loca-tions for monitoring flow velocity waveforms.

Although embryonic development can be studied us-ing serial observations in intact mice using the trans-cutaneous approach, this approach does have impor-tant limitations. With the exception of the first one ortwo embryos near the cervix, it is difficult to positivelyidentify embryos in subsequent exams or for latertissue collection, and some embryos may be too deep orin an inappropriate orientation for detailed examina-tion. This may be especially problematic when embryoswithin a litter differ in phenotype and/or genotype.Exteriorization of the uterus and the use of either a twotime-point longitudinal design [as in survival microin-jection studies (17, 22)] or a cross-sectional study de-sign avoids these problems. Exteriorization also allowssome adjustment of embryo orientation to optimize theimaging view and improves image resolution by elim-inating ultrasound attenuation caused by interveningmaternal skin, muscle, and viscera as shown previ-ously (32). Furthermore, the reduction in required pen-etration allowed the multifrequency biomicroscope tobe used at a higher insonation frequency, thereby fur-ther enhancing image resolution. Exteriorization ne-cessitates careful temperature maintenance for stableembryonic cardiovascular function (21). It also tended

Fig. 9. The aortic annulus region of an adult mouse with a peak ascending aortic blood velocity �200 cm/s (morethan twice normal) imaged at 19 MHz. A: longitudinal view of the left ventricle and aorta showing bright, coarseechoes (between arrows) in the aortic orifice (contrast with normal mouse shown in Fig. 7G). Diameter at the aorticannulus was 1.3 mm but more distally the ascending aorta was dilated (diameter 2.0 mm). B: short-axis view of theaortic annulus showing bright echoes in the valve region (contrast with normal mouse shown in Fig. 7E). C:histological section through the aortic valve of the same mouse showing abnormally thickened aortic valve leaflets.The smallest division on the scale bar in A and B is 100 �m. AA, aortic annulus; Ao, ascending aorta; LA, leftatrium; LV, left ventricle; LCA, left coronary artery; MPA, main pulmonary artery; AV, aortic valve.

124 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

to exacerbate embryonic movement caused by uterinecontractile activity; however, indomethacin added tothe bath appeared to minimize this effect. Movementsdue to maternal breathing were usually very smallwhether using transcutaneous or transuterine imag-ing. Even in the presence of some maternal move-ments, images with adequate quality could be found inthe cineloop of eight images that are automaticallystored with each “save” command. In future work, thehigher image resolution and improved embryo stabilitypossible with this method should enhance the accuracyof image-guided embryonic injections (10, 17, 22, 32)and the alignment of sequential ultrasound imagesused for three-dimensional reconstructions (32, 34).Further studies are required to determine the embry-onic effects of uterine exteriorization and indomethacinexposure; however, this approach is much less invasivethan studying mouse embryos directly using intravitalmicroscopy and pulsed Doppler probes (14, 18, 30).

The multifrequency biomicroscope has lower tempo-ral resolution and a smaller image size than that ofconventional ultrasound systems. The frame rate was4 frames/s during this study, although 8 frames/s isnow supported. Frame rates are relatively slow be-cause the image is created using a single, mechanicallydriven ultrasound transducer. By comparison, conven-tional ultrasound instruments use multiple transducerelements arranged in an array to achieve frame rates�120 frames/s (6). Low temporal resolution is a limi-tation when imaging the adult mouse heart, whichtypically beats at �500 min�1, as well as in embryoswhere, at a typical heart rate of �250 min�1, there isonly �1 image per cardiac cycle at 4 frames/s (23). Theimage size of the multifrequency biomicroscope (8 � 8mm) was adequate for the normal adult mouse heartbut may be inadequate when imaging very large miceor mice with cardiac hypertrophy. Thus, for applica-tions requiring a high frame rate, and/or large imagesize, conventional ultrasound systems may be pre-ferred despite the decrement in image resolution.

The biomicroscope’s Doppler system is ideal for mea-suring low blood velocities in veins, small arteries, orarterioles or in the embryonic vasculature. For in-stance, blood flow velocities �1 cm/s were detectable inthe umbilical vein within 1 day of umbilical cord for-mation, as well as within the early primitive hearttube. The ability to detect low blood velocities in smallvessels is one strength of the system, but the fact thathigh blood velocities are off-scale is a limitation. Themaximum measurable velocity was calculated to be37.5 cm/s, assuming 0° between flow and beam direc-tions, 20 kHz PRF, and a Doppler operating frequencyof 20 MHz. However, blood flow velocities in the heartand major arteries often exceed this value even inneonatal mice.

Blood flow calculated from noninvasive Dopplerblood velocity and vessel diameter measurements is anespecially attractive method in mice, where their smallsize, small blood volume, and large surface area meanthat standard invasive methods for blood flow determi-nation are much more difficult and error-prone than in

larger species [e.g., flow probes (13), labeled micro-spheres (26), thermodilution (15)]. Conventional ultra-sound systems have been used to calculate cardiacoutput in adult mice from ascending aortic blood veloc-ity and diameter measurements (35). However, accu-racy was limited by the relatively large Doppler samplevolume and low spatial resolution of conventional ul-trasound. This limitation is even more critical in juve-niles and neonates. In our experience, when the mul-tifrequency biomicroscope is used for diametermeasurements in conjunction with a separate high-frequency pulsed Doppler system, blood flow can bemeasured in vessels such as the carotid artery andaorta even in mouse neonates and juveniles. We use aseparate 20 MHz transcutaneous Doppler system (In-dus Instruments, Houston, TX) to record blood veloci-ties up to 200 cm/s in mice from birth to adulthood fromvarious sites (e.g., ascending aorta, main pulmonaryartery, mitral and tricuspid orifices) as described pre-viously for adult mice (12, 29). The high spatial reso-lution of the multifrequency biomicroscope enables ac-curate, noninvasive vessel diameter measurements.Accuracy of diameter measurement is extremely im-portant when calculating volume blood flow, becausediameter is squared when blood flow is calculated. Themultifrequency biomicroscope can also be used to im-age the region of interest to establish vessel angle anddepth (e.g., useful for angle correction and settingpulsed Doppler range).

There is limited information on bioeffects of high-frequency ultrasound. Diagnostic ultrasound at con-ventional frequencies during pregnancy is regarded assafe, although, under certain conditions, some effectson fetal development have been observed in animalmodels (e.g., on growth and hematopoiesis) (31). Themultifrequency biomicroscope’s mechanical index is�1 (peak pressure is �6 MPa at 40 MHz) and thuswould generally be considered safe and unlikely toelicit mechanical tissue damage (3). There are also twostudies using the prototype ultrasound biomicroscopethat report that transcutaneous scanning of anesthe-tized pregnant mice at 40 MHz does not adverselyaffect embryonic development (28, 34). In the currentstudy, we scanned two pregnant mice using 40 MHzultrasound daily from 6.5 to 17.5 days of gestation withno apparent embryonic loss and no obvious abnormal-ity in the neonates. Nevertheless, the major risk ofultrasound exposure, heating (2), needs to be directlyassessed over the relevant frequency range (19–55MHz). In addition, stress due to handling and/or anes-thetic exposure during pregnancy may also affect em-bryonic development (4) and therefore also needs to befurther explored.

In summary, the multifrequency biomicroscope op-erates over a frequency range appropriate for imagingmice with high spatial resolution throughout develop-ment from implantation to adulthood and for recordinglow blood flow velocities in the order of a few millime-ters per second in vessels as small as 100 �m. Conven-tional ultrasound systems may nevertheless be pre-ferred in applications requiring a larger image size,

125ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from

greater depth of penetration, and/or greater temporalresolution, and for applications where blood velocityexceeds 37.5 cm/s.

We thank Shathiyah Kulandavelu, Carol Kolb, Donna Johnston,and Kathie Whiteley for assistance during experiments, Yong Lu forsharing his anatomic expertise and archive of histological sections ofmouse embryos with us during this work, and Friedholm Bladt forthe opportunity to image GRIP1 �/� embryos. We also thank Dr.Colin McKerlie and the Pathology Core of the Centre for ModelingHuman Disease (http://www.cmhd.ca) for pathology and histologyservices for the adult mouse with abnormal aortic valves. We thankVisualSonics for technical support and the Heart and Stroke/RichardLewar Centre of Excellence for lending us their Acuson Sequoia forthis work.

We thank the Richard Ivey Foundation for funding the purchase ofthe multifrequency biomicroscope, the Canadian Institutes of HealthResearch for operating grant support, and the Ontario Research andDevelopment Challenge Fund for Fellowship support for Y. Q. Zhou.

F. S. Foster acknowledges a financial interest in the VisualSonicscompany.

REFERENCES

1. Aristizabal O, Christopher DA, Foster FS, and TurnbullDH. 40-MHz echocardiography scanner for cardiovascular as-sessment of mouse embryos. Ultrasound Med Biol 24: 1407–1417, 1998.

2. Barnett SB, Rott HD, Ter Haar GR, Ziskin MC, and MaedaK. The sensitivity of biological tissue to ultrasound. UltrasoundMed Biol 23: 805–812, 1997.

3. Barnett SB, Ter Haar GR, Ziskin MC, Rott HD, Duck FA,and Maeda K. International recommendations and guidelinesfor the safe use of diagnostic ultrasound in medicine. UltrasoundMed Biol 26: 355–366, 2000.

4. Barnett SB and Williams AR. Identification of mechanismsresponsible for fetal weight reduction in mice following ultra-sound exposure. Ultrasonics 28: 159–165, 1990.

5. Bladt F, Tafuri A, Gelkop S, Langille L, and Pawson T.Epidermolysis bullosa and embryonic lethality in mice lackingthe multi-PDZ domain protein GRIP1. Proc Natl Acad Sci USA99: 6816–6821, 2002.

6. Coatney RW. Ultrasound imaging: principles and applicationsin rodent research. ILAR J 42: 233–247, 2001.

7. Copp AJ. Death before birth: clues from gene knockouts andmutations. Trends Genet 11: 87–93, 1995.

8. Fatkin D, Christe ME, Aristizabal O, McConnell BK, Srini-vasan S, Schoen FJ, Seidman CE, Turnbull DH, and Seid-man JG. Neonatal cardiomyopathy in mice homozygous for theArg403Gln mutation in the � cardiac myosin heavy chain gene.J Clin Invest 103: 147–153, 1999.

9. Foster FS, Pavlin CJ, Harasiewicz KA, Christopher DA,and Turnbull DH. Advances in ultrasound biomicroscopy. Ul-trasound Med Biol 26: 1–27, 2000.

10. Gaiano N, Kohtz JD, Turnbull DH, and Fishell G. A methodfor rapid gain-of-function studies in the mouse embryonic ner-vous system. Nat Neurosci 2: 812–819, 1999.

11. Gui YH, Linask KK, Khowsathit P, and Huhta JC. Dopplerechocardiography of normal and abnormal embryonic mouseheart. Pediatr Res 40: 633–642, 1996.

12. Hartley CJ, Michael LH, and Entman ML. Noninvasivemeasurement of ascending aortic blood velocity in mice. Am JPhysiol Heart Circ Physiol 268: H499–H505, 1995.

13. Janssen B, Debets J, Leenders P, and Smits J. Chronicmeasurement of cardiac output in conscious mice. Am J PhysiolRegul Integr Comp Physiol 282: R928–R935, 2002.

14. Keller BB, MacLennan MJ, Tinney JP, and Yoshigi M. In vivoassessment of embryonic cardiovascular dimensions and functionin day-10.5 to -14.5 mouse embryos. Circ Res 79: 247–255, 1996.

15. Kissling G, Ross C, and Brandle M. Validity of thermaldilution technique for measurement of cardiac output in rats.Am J Physiol Heart Circ Physiol 265: H1007–H1013, 1993.

16. Linask KK, Huhta JC. Use of Doppler echocardiography tomonitor embryonic mouse heart function. In: Developmental

Biology Protocols, edited by Tuan RS and Lo CW. Totowa, NJ:Humana, 2000, p. 245–252.

17. Liu A, Joyner AL, and Turnbull DH. Alteration of limb andbrain patterning in early mouse embryos by ultrasound-guidedinjection of Shh-expressing cells. Mech Dev 75: 107–115, 1998.

18. MacLennan MJ and Keller BB. Umbilical arterial blood flowin the mouse embryo during development and following acutelyincreased heart rate. Ultrasound Med Biol 25: 361–370, 1999.

19. McConnell BK, Jones KA, Fatkin D, Arroyo LH, Lee RT,Aristizabal O, Turnbull DH, Georgakopoulos D, Kass D,Bond M, Niimura H, Schoen FJ, Conner D, Fischman DA,Seidman CE, and Seidman JG. Dilated cardiomyopathy inhomozygous myosin-binding protein-C mutant mice. J Clin In-vest 104: 1235–1244, 1999.

20. Muntener M and Hsu YC. Development of trophoblast andplacenta of the mouse. A reinvestigation with regard to the invitro culture of mouse trophoblast and placenta. Acta Anat(Basel) 98: 241–252, 1977.

21. Nakazawa M, Miyagawa ST, Morishima M, Kajio F, andTakao A. Effects of environmental hyperthermia on cardiovas-cular function in the rat embryo. Pediatr Res 30: 505–508, 1991.

22. Olsson M, Campbell K, and Turnbull DH. Specification ofmouse telencephalic and mid-hindbrain progenitors followingheterotopic ultrasound-guided embryonic transplantation. Neu-ron 19: 761–772, 1997.

23. Phoon CKL, Aristizabal O, and Turnbull DH. 40 MHz Dopp-ler characterization of umbilical and dorsal aortic blood flow inthe early mouse embryo. Ultrasound Med Biol 26: 1275–1283,2000.

24. Rinkenberger JL, Cross JC, and Werb Z. Molecular geneticsof implantation in the mouse. Dev Genet 21: 6–20, 1997.

25. Russell ES and Bernstein SE. Blood and blood formation. In:Biology of the Laboratory Mouse, edited by Green EL. New York:McGraw-Hill, 1966, p. 351–372.

26. Sarin SK, Sabba C, and Groszmann RJ. Splanchnic andsystemic hemodynamics in mice using a radioactive microspheretechnique. Am J Physiol Gastrointest Liver Physiol 258: G365–G369, 1990.

27. Sissman NJ. Developmental landmarks in cardiac morphogen-esis: comparative chronology. Am J Cardiol 25: 141–148, 1970.

28. Srinivasan S, Baldwin HS, Aristizabal O, Kwee L, LabowM, Artman M, and Turnbull DH. Noninvasive, in utero imag-ing of mouse embryonic heart development with 40-MHz echo-cardiography. Circulation 98: 912–918, 1998.

29. Taffet GE, Hartley CJ, Wen X, Pham T, Michael LH, andEntman ML. Noninvasive indexes of cardiac systolic and dia-stolic function in hyperthyroid and senescent mouse. Am JPhysiol Heart Circ Physiol 270: H2204–H2209, 1996.

30. Tanaka N, Mao L, DeLano FA, Sentianin EM, Chien KR,Schmid-Schonbein GW, and Ross J Jr. Left ventricular vol-umes and function in the embryonic mouse heart. Am J PhysiolHeart Circ Physiol 273: H1368–H1376, 1997.

31. Tarantal AF. Effects of ultrasound exposure on fetal develop-ment in animal models. In: Safety of Diagnostic Ultrasound,edited by Barnett SB and Kossoff G. New York: Parthenon, 1998,p. 39–51.

32. Turnbull DH. In utero ultrasound backscatter microscopy ofearly stage mouse embryos. Comput Med Imaging Graph 23:25–31, 1999.

33. Turnbull DH. Ultrasound backscatter microscopy of mouseembryos. In: Developmental Biology Protocols, edited by TuanRS and Lo CW. Totowa, NJ: Human, 2000, p. 235–243.

34. Turnbull DH, Bloomfield TS, Baldwin HS, Foster FS, andJoyner AL. Ultrasound backscatter microscope analysis ofearly mouse embryonic brain development. Proc Natl Acad SciUSA 92: 2239–2243, 1995.

35. Yang XP, Liu YH, Rhaleb NE, Kurihara N, Kim HE, andCarretero OA. Echocardiographic assessment of cardiac func-tion in conscious and anesthetized mice. Am J Physiol Heart CircPhysiol 277: H1967–H1974, 1999.

36. Yousif MH and Thulesius O. Tocolytic effect of the cyclooxy-genase-2 inhibitor, meloxicam: studies on uterine contractions inthe rat. J Pharm Pharmacol 50: 681–685, 1998.

126 ULTRASOUND BIOMICROSCOPY IN MICE

Physiol Genomics • VOL 10 • www.physiolgenomics.org

on August 14, 2010

physiolgenomics.physiology.org

Dow

nloaded from