Xia 2018
Xia 2018
Xia 2018
Ultrasound in Med. & Biol., Vol. 00, No. 00, pp. 112, 2018
Copyright © 2018 World Federation for Ultrasound in Medicine & Biology. All rights reserved.
Printed in the USA. All rights reserved.
0301-5629/$ - see front matter
https://doi.org/10.1016/j.ultrasmedbio.2018.11.001
Original Contribution
TAGEDPJINGJING XIA,* YI YANG,* CHENWENBAO HU,* RUI MENG,* QIUJU JIANG,*,y RONG LIU,*,y YANYAN YU,*,y
ZONGHAI SHENG,*,y FEI YAN,* LIJUAN ZHANG,* ZHIFENG SHI,z HAIRONG ZHENG,*,y and WEIBAO QIU*,yTAGEDEN
* Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of
Sciences, Shenzhen, China; y Shenzhen Key laboratory of Ultrasound Imaging and Therapy, Shenzhen, China; and z Shanghai
Pituitary Tumor Center, Huashan Hospital, Shanghai Medical College, Fudan University, Shanghai, China
(Received 15 January 2018; revised 27 September 2018; in final from 3 November 2018)
Abstract—Precisely evaluating the characteristics of a glioma tumor in vivo is challenging when performing sur-
gical resection clinically. The infiltration characteristics of a tumor make precise resection difficult because of
uncertainties about the surrounding vasculature and the relationships with functional structures. Magnetic reso-
nance imaging is routinely used to distinguish the area of a glioma, but it cannot resolve details of the vascular
network around or inside the tumor. Ultrasound imaging is a real-time imaging modality that has been applied
clinically in intra-operative surgery, and the sensitivity of flow measurements in the brain is improved by ultra-
fast plane wave imaging. This study applies a plane wave-based power Doppler imaging method to visualize the
blood flow distribution in glioma models in vivo. This new imaging method makes it possible to delineate the flow
structure of a glioma tumor in the brain of a small animal. The tumor can be distinguished from normal brain tis-
sue, and different sections of the tumor contain different flow structures. The normalized blood flow intensities
(mean § standard deviation) within regions of interest were 0.33 § 0.13, 0.72 § 0.15, 0.36 § 0.23 and 0.06 § 0.07
for the type I normal rat, type I glioma rat, type II normal rat and type II glioma rat, respectively. Quantification
analysis verified the feasibility of using this plane wave-based Doppler imaging method to evaluate brain tumors
in small animals. (E-mail: [email protected]) © 2018 World Federation for Ultrasound in Medicine & Biology.
All rights reserved.
Key Words: Glioma, Plane wave ultrasound, Power Doppler imaging, Flow measurement, In vivo imaging.
1
ARTICLE IN PRESS
2 Ultrasound in Medicine & Biology Volume 00, Number 00, 2018
highly resolved neurovascular map of the glioma. The 11 angles per frame
an orthotopic brain tumor. The rats were injected with method, and can efficiently remove clutter signals and
C6 cells in the right brain 3 mm laterally and 1 mm preserve blood flow signals even at low frequencies by
axially from the bregma point (Valable et al. 2007). A removing the proper eigen-components (Osmanski et al.
burr hole with a diameter of 1 mm was drilled, and 10 2014).
mL of cell suspension (106 cells/mL) was injected into We define the 300-IQ data packet as a 3-D matrix S
the brain within 10 min using a micro-injection driven (x, z, t), where (x, z) is the physical coordinates of the
by a Hamilton syringe (KD Scientific, Holliston, MA, pixel, and t is time. The mathematical expression of the
USA). The needle was slowly removed 5 min after the SVD filter (extracting the blood signal from S(x, z, t), con-
injection, and then the burr hole was sealed with bone verting the 3-D blood signal Sblood(x, z, t) to a 2-D matrix
wax and the scalp was sutured. MR images were Sblood(k, t) and applying the SVD decomposition) is
obtained before the experiment to ensure successful
X
N
tumor formation. If the tumor had been generated suc- Sblood ðx; z; t Þ ¼ ei Bi ðx; zÞ Si ðt Þ ð1Þ
cessfully, ultrasound imaging was performed 0.51 d i ¼ 1
after the MRI scan. There were 50 rats in total, and 28 of
them were selected for the study (16 rats failed without The matrix Bi(x, z) will give the spatial distribution
tumor, and 6 rats died during the skull surgery). of the signal Si(t) for the corresponding singular value of
For the ultrasound imaging experiment, the rat was ei ði ¼ 1; 2; . . . ; NÞ. The asterisk denotes the multipli-
shaved and then fixed on a 3-D stereotaxic apparatus cation. Based on the previous research, the tissue move-
(68028, RWD, Shenzhen, China) after anesthetization ments generate coherent signals while the vascular
with 10% chloral hydrate. However, although ultrasound signals coming from the randomly moving red blood cells
can propagate deep within tissues, cerebral US remains are uncorrelated. The larger singular values correspond to
limited by its poor penetration through the skull. An the tissue signal, and the intermediate singular values cor-
attenuation of 6.9 dB/cm at 15 MHz has been observed respond to the blood signal; the last few singular values
in previous studies (Fry and Barger 1978; Larrat et al. represent the noise signal (Song et al. 2016). Then, we
2010; Pinton et al. 2012), which renders small variations used the low-order and high-order singular value thresh-
in blood flow indistinguishable from noise. Therefore, to olds (90 and 270, respectively) to remove the markedly
avoid ultrasound attenuation by the skull, the skull was low and high singular values and extract the blood signal.
drilled over an area of approximately 0.8 £ 1.2 cm Additionally, a Butterworth high-pass filter with a
(bilateral window), with the dura remaining intact, and 25-Hz cutoff frequency was employed to reduce the
was sprayed with saline water for cleaning out and cool- noise further. Finally, we calculated the square of the fil-
ing during the drilling process. The whole brain could tered IQ data and averaged over the total 300 matrix to
then be imaged through the drilled window by an ultra- form one blood flow map. We thus acquired information
sound transducer with the aid of a coupling agent. on cerebral coronal blood flow, including the vascular
structure of the glioma tumor.
Ultrasound imaging processing To improve image quality, we also proposed the
All data were post-processed offline using MAT- dynamic receive aperture imaging technology for the
LAB (MathWorks, Natick, MA, USA). Delay-and-sum receive beamformer. We found that the sound field near
beamforming and coherence compounding yielded a the aperture fluctuates strongly, resulting in a interfer-
frame ensemble in the time domain, with 300 com- ence effect. Fortunately, these effects can be avoided by
pounded frames (the total acquisition time was 0.3 s) using smaller apertures in the near field (Harris 1978).
used as an ensemble to form a cerebral blood flow Compared with the beamformer using the boxcar win-
image, as illustrated in Figure 1. In detail, we first dows function, the use of dynamic aperture can maintain
obtained the beamformed RF data at each transmission a certain ratio between the effective aperture size and
angle with delay-and-sum beamforming and coherently the focusing depth. The equation, based on a boxcar win-
summed the RF data of all 11 titled angles to form one dow function, is
frame matrix. Second, we took 300 time-continuous 8
>
compounded frames as an ensemble and applied Hilbert < 1; n pitchx=z fnumber =2
>
WðnÞ ¼ n ¼ 0 » N1
transfer to each frame matrix to obtain 300-continuous- >
>
IQ data matrix. In the third step, a high-pass spatiotem- : 0; n pitchx=z > fnumber =2
poral filter based on singular value decomposition
ð2Þ
(SVD) was applied to the 300-IQ data packet in the time
domain to adaptively suppress slow-time clutters of
low-frequency tissue signals. This adaptive SVD filter is where pitch means the distance between two adjacent
a robust and reliable orthogonal matrix decomposition elements, and (x, z) are the physical coordinates of the
ARTICLE IN PRESS
4 Ultrasound in Medicine & Biology Volume 00, Number 00, 2018
a b
1mm
Fig. 2. Effect of dynamic receive aperture technology on Doppler imaging. (a) Beamforming without dynamic aperture.
(b) Beamforming with dynamic aperture. The signal-to-noise value of (b) is 15.1 dB higher than that of (a).
According to the corresponding power Doppler image in accurate reference. Figure 5(a, b) are MRI images of two
Figure 4c (display dynamic range = 45 dB), the tumor glioma rats in the coronal view scanned 0.5 d before ultra-
part in the right brain has a greater flow and a higher sound imaging, and Figure 5 (c, d) are the corresponding
intensity compared with the contralateral area of the plane wave Doppler images of Figure 5(a, b) (display
brain. Moreover, the flow pattern in the tumor area fea- dynamic range = 45 dB), respectively. The
tures more curved segments than normal tissue. glioma tumors in the rats are not easy to locate in Figure 5
Figure 4(b, d) are images of another rat model that (a, b) because of the low resolution of MRI; it is even
had been injected with C6 cells for 26 d. Similarly to more difficult to obtain the flow information. However, in
Figure 4a, the B-mode image in Figure 4d (dynamic the ultrasound power Doppler images in Figure 5(c, d), it
range = 50 dB) illustrates hyper-echoic signals over a is possible to identify the tumor from the blood flow dis-
large area in the right part of the brain. The tumor area is tributions within the tumor region, in which blood flow
larger than for the brain injected for 22 d, which may structures differ from those in normal regions of the brain.
indicate a larger tumor. The corresponding power Dopp-
ler image in Figure 4d (display dynamic range = 45 dB) Different sections of the glioma tumor
differs from that in Figure 4c. The flow pattern is also Figure 6 illustrates nine coronal-view plane wave
quite different from that of the contralateral area of the Doppler images of a glioma tumor in different sections.
brain, and the number and intensity of hyper-echoic sig- We obtained blood flow images of about 95 continuous
nals for blood flow is clearly decreased in a large area, sections along the direction from bregma to lambda. The
which may indicate some changes in the tumor. distance between two adjacent sections is 100 mm. We
Although this region still exhibits hyper-echogenicity, then chose several representative sections: one for every
tumor necrosis might already be present because of an five from section 15 to section 55. As for the blood flow
insufficient blood supply. structure in the tumor area in the right brain, there is a
tendency in this series of section images for the blood
Comparison with MRI flow to appear rich on both sides of the tumor, with
Magnetic resonance scanning was used on glioma necrosis arising gradually in the middle parts. It could be
rats for the first time at about 15 d after injection of C6 inferred that sections 15 to 25 and sections 45 to 55 are
cells to confirm tumor growth. Generally, a 2- to 3-mm near the edges of the tumor. The blood flow intensity is
diameter of tumor in the dorsal view is optimal, and then greater because new blood vessels are needed to support
0.5 d before the ultrafast ultrasound imaging experiment, tumor growth. On the other hand, sections 30 to 40 are
the second MRI scan was conducted to provide a more in the central part of the tumor, and very low blood flow
ARTICLE IN PRESS
6 Ultrasound in Medicine & Biology Volume 00, Number 00, 2018
Fig. 3. Comparison between different window functions. (a) In vivo rat cerebral blood flow image in the coronal view.
(b) Weighted curves of the window functions in the central element. The lateral normalized intensity profile of the three
vessels in the red-lined box were processed under different window functions. (c) Intensity was normalized respectively.
(d) Intensity was normalized uniformly.
was detected because of the insufficient blood supply. structure. The two ROIs in the same images are totally
These observations indicate the feasibility of using the symmetric and the same size.
proposed method to perform vascular imaging of a gli- The results plotted in Figure 7a are normalized
oma tumor. We used V3-D software (Peng et al. 2010) intensities calculated for 13 type I samples and 15 type
to reconstruct the series of sectional images into a 3-D II samples. The normalized intensities were 0.33 §
image in Supplementary Video S1 (online only). 0.13 (mean § SD), 0.72 § 0.15, 0.36 § 0.23 and 0.06
§ 0.07 for the type I normal ROI, type I glioma ROI,
Flow quantification type II normal ROI and type II glioma ROI, respec-
It has been found that the tumor region appears as a tively.
hyper-echoic area in B-mode images and that its blood Before analyzing this statistical result, we also
flow structures can be distinguished from those in the used 25 samples of the normal rat model to calculate
normal regions of the brain, with these differences vary- and normalize the mean flow intensities of the ROIs in
ing with tumor growth (tumor volume). Specifically, the left and right brain. In Figure 7b it is seen that the
compared with normal cerebral blood flow, the tumor intensities in the two ROIs are 0.44 § 0.06 and 0.49 §
flow structures could be divided into two types: (type I) 0.04, respectively. This indicates that the mean intensi-
regions with greater blood flow and (type II) regions ties in the left and right brain of normal rats are nearly
with poor blood flow. We calculated the mean flow equivalent. On the basis of this result, it can be proved
intensities by manually choosing two ROIs (indicated by that the intensities in the tumor area of the type I rat
the red-dashed squares in Fig. 4) within the tumor and model are much higher than those in the normal area,
the normal cerebral region for both types of flow and on the other hand, the intensities in the tumor area
ARTICLE IN PRESS
Brain tumor evaluation in small animals J. XIA et al. 7
a b
c d
1 mm
Fig. 4. Comparison between plane wave B-mode images and plane wave ultrasound Doppler images of brain tissue with
a glioma tumor. (a) B-Mode image from plane wave ultrasonography of a 22-d-old tumor in the right brain. (b) B-Mode
image from plane wave ultrasonography of a 26-d-old tumor in the right brain. (c) Blood flow image of the brain in (a).
(d) Blood flow image of the brain in (b).
of the type II rat model are much lower than those in the tissue in the stained region, consistent with the data
normal area. acquired with other imaging modalities.
Histology comparison
DISCUSSION
Hematoxylin and eosin staining was performed to
enable precise verification of glioma tumor. Figure 8 Precisely evaluating the characteristics of a glioma
compares the (a) MR image, (b) histology image, (c) tumor in vivo is a challenge when performing surgical
B-mode image and (d) plane wave power Doppler image resection clinically. Knowledge of the surrounding vascu-
of a glioma rat. As illustrated in the MR, B-mode and lature of the tumor is necessary to avoid both major brain
flow images, the glioma tumor grows in the right brain damage and recurrence. Ultrasound Doppler imaging
of this rat. Figure 8b is an image of the HE-stained slice method is a well-established tool for flow analysis and
about +1.92 mm from bregma scanned by electron quantification. A recently developed plane wave flow
microscopy panoramic. As we can see, the glioma tumor imaging method exhibits improved sensitivity (Mace et
area in the right brain is much darker than the normal al. 2011, 2013). Plane wave-based contrast-enhanced
ARTICLE IN PRESS
8 Ultrasound in Medicine & Biology Volume 00, Number 00, 2018
a b
c d
1 mm
Fig. 5. Comparison between traditional magnetic resonance images and plane wave ultrasound power Doppler images of
brain tissue with a glioma tumor. (a, b) Magnetic resonance images. (c, d) Blood-flow plane wave ultrasound images of
(a) and (b), respectively.
ultrasound CEUS presents several advantages over tradi- on a glioma tumor (Figs. 4 and 5), which makes it a valu-
tional ultrasound and can provide high-resolution images able tool for tumor characterization. A dynamic receive
of the vascular network of a rabbit kidney (Tremblay- aperture technology was developed to further improve
Darveau et al. 2014). Moreover, multi-frame-rate plane the imaging resolution of power Doppler measurement.
wave CEUS was developed for vasculature imaging in We quantitatively analyzed the effects of different win-
subcutaneous lymphoma and perfusion quantification dow functions on the receive beamformer in Figure 3,
(Leow et al. 2017). In addition, ultrasound localization and concluded that the Tukey window was an optimal
microscopy was utilized to visualize subcutaneous fibro- choice in suppressing the grating lobes and reducing
sarcoma angiogenesis (Lin et al. 2017). The present study their energy. The SNR value of the improved power
is the first to use plane wave ultrasound to obtain flow Doppler image was increased by 15.1 dB, as illustrated
information on a brain glioma tumor in a small animal in Figure 2.
in vivo. The image processing performed in this study
The proposed plane wave-based Doppler imaging revealed that glioma tumor regions exhibited blood flow
method can be used to obtain detailed flow information structures that differed from those in the normal regions
ARTICLE IN PRESS
Brain tumor evaluation in small animals J. XIA et al. 9
1mm
Fig. 6. Plane wave-based Doppler images of the glioma at different imaging locations in coronal views: several repre-
sentative sections of 95 continuous section images, and the step between two adjacent sections is 100 mm.
of the brain. In addition, different tumor growth periods signal intensities of the blood flow in the normal and
may be associated with quite different blood flow distri- tumor regions in the same imaging sections and used sta-
butions in tumor regions. The experimental results indi- tistical analyses to identify different positions of the
cated that blood flow was poor in the central part of a tumor.
tumor that had been growing for a long time. The greater In this article, we described a newly proposed
intensity of blood flow in the tumor periphery may method for providing information on vessel orientations
reflect the aggressiveness of the tumor in forming new and distributions. It is an important reference for exten-
blood vessels to support its growth. Moreover, as the sive resection to avoid both major brain damage and
tumor grows, its central part may be affected by necrosis recurrence. Moreover, the imaging method proposed
if blood flow decreases significantly. We compared the here has the potential to be a useful tool in determining
a b
Normalized Intensity
Normalized Intensity
Fig. 7. Normalized intensity of blood flow in the tumor and normal cerebral blood flow. (a) In the rat model of glioma,
blood flow intensity is greater in type I than in type II glioma. (b) Comparison of blood flow in the left and right ROIs of
the normal rat model. ROI = region of interest.
ARTICLE IN PRESS
10 Ultrasound in Medicine & Biology Volume 00, Number 00, 2018
Fig. 8. Mapping the stained brain slice of a glioma rat to the corresponding B-mode image and plane wave Doppler
image. The tumor area is much darker and is consistent with the plane wave Doppler image. (a) Magnetic resonance
image. (b) Histology image. (c) B-Mode image (50 dB). (d) Plane wave power Doppler image.
whether patients with tumor are suited for targeted ther- flow image. A field-programmable gate array and GPU
apy, which would provide the basis for the follow-up for (graphics processing unit) could be used to expedite the
targeting therapy of tumor blood vessels and reducing process by implementing parallel computing (Qiu et al.
the suffering of patients. Multimodality imaging 2013, 2015). Clinical testing during intra-operative sur-
(Kircher et al. 2012) could be implemented by combin- gery in humans could be performed once the system
ing gray-scale B-mode imaging, MRI and the proposed becomes sufficiently fast. Moreover, the glioma tumor
plane wave flow imaging. Registration of local flow images obtained in different sections in coronal and sagit-
images of the tumor region with B-mode images is illus- tal views in Figure 6 indicated the feasibility of establish-
trated in Figure 9(a, c). The details of the vascular net- ing 3-D vascular imaging of a glioma tumor.
work inside the tumor could be resolved. Registration of
ultrasound flow images with MR images is illustrated in
CONCLUSIONS
Figure 9(b, d). Such a combination of imaging modali-
ties would allow full evaluation of the characteristics of This study is the first to characterize the blood flow
the tumor. Combined with other imaging guidance meth- distributions of a brain glioma tumor in a small animal
ods, a more accurate and safe surgery could be con- using plane wave Doppler imaging. Vascular maps of
ducted. the glioma were obtained, and detailed flow information
The ultrasound images acquired in this study were for the whole brain—notably inside and around the
obtained by processing RF data using an offline computer, tumor—could be acquired. This new method of plane
which was a time-consuming process to obtain a single wave ultrasound can be used to evaluate malignant
ARTICLE IN PRESS
Brain tumor evaluation in small animals J. XIA et al. 11
a b
c d
1 mm
Fig. 9. Registration of different imaging modalities. (a, b) Registration of ultrasound flow image with the B-mode image
(a) and the magnetic resonance image (b) for a 22-d tumor model. (c, d) Registration the ultrasound flow image with the
B-mode image (c) and the magnetic resonance image (d) for a 26-d tumor model.
Gerganov VM, Samii A, Giordano M, Samii M, Fahlbusch R. Two- Osmanski BF, Pezet S, Ricobaraza A, Lenkei Z, Tanter M. Func-
dimensional high-end ultrasound imaging compared to intraopera- tional ultrasound imaging of intrinsic connectivity in the living
tive MRI during resection of low-grade gliomas. J Clin Neurosci rat brain with high spatiotemporal resolution. Nat Commun
2011;18:669–673. 2014;5:5023.
Goodenberger ML, Jenkins RB. Genetics of adult glioma. Cancer Peng H, Ruan Z, Long F, Simpson JH, Myers EW. V3D enables real-
Genet 2012;205:613–621. time 3D visualization and quantitative analysis of large-scale
Harris FJ. On the use of windows for harmonic analysis with the dis- biological image data sets. Nat Biotechnol 2010;28:348–353.
crete fourier transform. Proc IEEE 1978;66:51–83. Pinton G, Aubry JF, Bossy E, Muller M, Pernot M, Tanter M. Attenua-
Hatiboglu MA, Weinberg JS, Suki D, Rao G, Prabhu SS, Shah K, Jack- tion, scattering, and absorption of ultrasound in the skull bone. Med
son E, Sawaya R. Impact of intraoperative highfield magnetic res- Phys 2012;39:299.
onance imaging guidance on glioma surgery: A prospective Pirzkall A, McKnight TR, Graves EE, Carol MP, Sneed PK, Wara
volumetric analysis. Neurosurgery 2009;64:1073–1081. WW, Nelson SJ, Verhey LJ, Larson DA. Mr-spectroscopy guided
Huse JT, Holland EC. Targeting brain cancer: Advances in the molecu- target delineation for high-grade gliomas. Int J Radiat Oncol Biol
lar pathology of malignant glioma and medulloblastoma. Nat Rev Phys 2001;50:915–928.
Cancer 2010;10:319–331. Qiu W, Yu Y, Chabok HR, Liu C, Tsang FK, Zhou Q, Shung KK,
Hygino da Cruz LC, Rodriguez I, Domingues RC, Gasparetto EL, Sor- Zheng H, Sun L. A flexible annular-array imaging platform for
ensen AG. Pseudoprogression and pseudoresponse: Imaging micro-ultrasound. IEEE Trans Ultrason Ferroelectr Freq Control
challenges in the assessment of posttreatment glioma. Am J Neuro- 2013;60:178–186.
radiol 2011;32:1978. Qiu W, Wang C, Li Y, Zhou J, Yang G, Xiao Y, Feng G, Jin Q, Mu P,
Kircher MF, De La Zerda A, Jokerst JV, Zavaleta CL, Kempen PJ, Mit- Qian M. A scanning-mode 2D shear wave imaging (s2D-SWI)
tra E, Pitter K, Huang R, Campos C, Habte F. A brain tumor system for ultrasound elastography. Ultrasonics 2015;62:89–96.
molecular imaging strategy using a new triple-modality mri-photo- Renner C, Lindner D, Schneider J, Meixensberger J. Evaluation of
acoustic-raman nanoparticle. Nat Med 2012;18:829–834. intra-operative ultrasound imaging in brain tumor resection: A pro-
Larrat B, Pernot M, Aubry JF, Dervishi E, Sinkus R, Seilhean D, Marie spective study. Neurol Res 2005;27:351–357.
Y, Boch AL, Fink M, Tanter M. MR-guided transcranial brain hifu Sanai N, Berger MS. Intraoperative stimulation techniques for func-
in small animal models. Phys Med Biol 2010;55:365. tional pathway preservation and glioma resection. Neurosurg Focus
Leow CH, Marta B, Stanziola A, Hernandez-Gil J, Long NJ, Aboagye 2010;28:E1.
EO, Tang MX. Multi-frame rate plane wave contrast-enhance ultra- Sanai N, Mirzadeh Z, Berger MS. Functional outcome after language
sound imaging for tumour vasculature imaging and perfusion quan- mapping for glioma resection. N Engl J Med 2008;358:18–27.
tification. Proc IEEE Int Ultrason Symp 2017;1–4. Sanai N, Polley MY, Berger MS. Insular glioma resection: Assessment
Lin F, Shelton SE, Espındola D, Rojas JD, Pinton G, Dayton PA. 3-D of patient morbidity, survival, and tumor progression. J Neurosurg
ultrasound localization microscopy for identifying microvascular 2010;112:1–9.
morphology features of tumor angiogenesis at a resolution beyond Selbekk T, Jakola AS, Solheim O, Johansen TF, Lindseth F, Reinertsen
the diffraction limit of conventional ultrasound. Theranostics I, Unsgard G. Ultrasound imaging in neurosurgery: Approaches to
2017;7:196–204. minimize surgically induced image artefacts for improved resection
Mace E, Montaldo G, Cohen I, Baulac M, Fink M, Tanter M. Functional control. Acta Neurochir (Wien) 2013;155:973–980.
ultrasound imaging of the brain. Nat Methods 2011;8:662–664. Song P, Manduca A, Trzasko J, Chen S. Ultrasound small vessel imag-
Mace E, Montaldo G, Osmanski BF, Cohen I, Fink M, Tanter M. Func- ing with block-wise adaptive local clutter filtering. IEEE Trans
tional ultrasound imaging of the brain: Theory and basic principles. Med Imaging 2016;36:251–262.
IEEE Trans Ultrason Ferroelectr Freq Control 2013;60:492–506. Tanter M, Fink M. Ultrafast imaging in biomedical ultrasound. IEEE
Montaldo G, Mace E, Cohen I, Berckoff J, Tanter M, Fink M. Ultrafast Trans Ultrason Ferroelectr Freq Control 2014;61:102–119.
compound Doppler imaging: A new approach of Doppler flow anal- Tremblay-Darveau C, Williams R, Milot L, Bruce M, Burns PN. Com-
ysis. In: Proceedings of the 2010 IEEE International Symposium on bined perfusion and Doppler imaging using plane-wave nonlinear
Biomedical Imaging: From Nano to Macro, Rotterdam, The Neth- detection and microbubble contrast agents. IEEE Trans Ultrason
erlands, April 1417, 2010. New York. : IEEE; 2010. p. 324–327. Ferroelectr Freq Control 2014;61:1988–2000.
Nakamura K, Ito Y, Kawano Y, Kurozumi K, Kobune M, Tsuda H, Urban A, Brunner C, Dussaux C, Chassoux F, Devaux B, Montaldo G.
Bizen A, Honmou O, Niitsu Y, Hamada H. Antitumor effect of Functional ultrasound imaging of cerebral capillaries in rodents and
genetically engineered mesenchymal stem cells in a rat glioma humans. Jacobs J Mol Transl Med 2015;1:007.
model. Gene Ther 2004;11:1155–1164. Valable S, Barbier EL, Bernaudin M, Roussel S, Segebarth C, Petit E,
Orringer DA, Chen T, Huang DL, Armstead WM, Hoff BA, Koo YE, Remy C. In vivo MRI tracking of exogenous monocytes/macro-
Keep RF, Philbert MA, Kopelman R, Sagher O. The brain tumor phages targeting brain tumors in a rat model of glioma. Neuro-
window model: A combined cranial window and implanted glioma Image 2007;37:S47–S58.
model for evaluating intraoperative contrast agents. Neurosurgery Zhou J, Tryggestad E, Wen Z, Lal B, Zhou T, Grossman R, Wang
2010;66:736–743. S, Yan K, Fu D-X, Ford E. Differentiation between glioma and
Osmanski BF, Montaldo G, Tanter M, Fink M. Aberration correction radiation necrosis using molecular magnetic resonance imaging
by time reversal of moving speckle noise. IEEE Trans Ultrason of endogenous proteins and peptides. Nat Med 2011;17:
Ferroelectr Freq Control 2012;59:1575–1583. 130–134.