HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2261011848v1 226/1/272    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Green, J. D. Articles by Li, D. Search for Related Content PubMed PubMed Citation Articles by Green, J. D. Articles by Li, D.

Two- and Three-dimensional MR Coronary Angiography with Intraarterial Injections of Contrast Agent in Dogs: A Feasibility Study¹

¹ From the Departments of Radiology (J.D.G., R.A.O., J.P.F., R.T., Y.L., S.V., D.L.) and Biomedical Engineering (J.D.G., D.L.), Northwestern University Medical School, 448 E Ontario St, Suite 700, Chicago, IL 60611; and Siemens Medical Solutions, Chicago, Ill (Y.C.C.). Received November 19, 2001; revision requested December 20; final revision received April 25, 2002; accepted May 8. Supported in part by National Institutes of Health (NIH) grant HL 38698 and Siemens Medical Solutions, Erlangen, Germany. R.A.O. supported in part by NIH K08 DK60020. Address correspondence to D.L. (e-mail: d-li2@northwestern.edu).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Magnetic resonance (MR) images of coronary arteries were acquired with an inversion recovery–prepared technique after intraarterial injection of contrast material in five dogs. Real-time two-dimensional projection images were obtained with a temporal resolution of 3 frames per second. Three-dimensional electrocardiographically triggered high-spatial-resolution images were obtained with a fraction of the contrast agent required for intravenous injections. Background tissues were adequately suppressed in all images. On the basis of this experimental data, the optimal contrast agent concentration for two-dimensional real-time projection imaging was 6%. This preliminary work shows that contrast material–enhanced MR angiography with intraarterial injections is feasible with the proposed techniques.

© RSNA, 2002

Index terms: Animals • Coronary vessels, MR, 54.121413, 54.12142 • Interventional procedures, experimental studies • Magnetic resonance (MR), contrast enhancement, 54.12143 • Magnetic resonance (MR), guidance, 54.12143 • Magnetic resonance (MR), vascular studies, 54.12141

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Magnetic resonance (MR) imaging with catheter-directed intraarterial injections of a contrast agent is potentially useful in endovascular interventions (1,2). Current catheter-based endovascular procedures involve the use of x-ray fluoroscopy and iodinated contrast agents to provide the operator with detailed real-time angiographic information. Although x-ray fluoroscopic guidance provides high-resolution images of blood vessels, it does not provide satisfactory soft-tissue characterization and exposes the patient to ionizing radiation and potentially nephrotoxic iodinated contrast agents (3,4).

MR angiography is a potential noninvasive alternative to conventional angiography, without the risks of ionizing radiation or nephrotoxicity. Contrast agent– enhanced MR angiography with intraarterial injections is potentially useful during interventional procedures such as angioplasty (1,5). Although it is invasive, MR angiography may still have an advantage over conventional angiography as a diagnostic tool because it does not involve the use of ionizing radiation. MR imaging also confers the ability to obtain excellent soft-tissue contrast and enables evaluation of organ function (6,7) before and after an intervention. When a contrast agent is delivered locally, the dose may be minimized to allow for multiple injections in a single session (8).

For coronary endovascular procedures, real-time vascular "road maps" are used to guide catheters from the access site—typically the common femoral artery—into the coronary artery. Road maps are required for procedures performed with either radiographic or MR imaging guidance. For MR imaging–guided interventions, MR angiography may be used to assess the location and severity of coronary artery disease, as well as to provide the vascular road map for catheter tracking. High-temporal-resolution MR images are required for vascular road maps, and high-spatial-resolution MR images are required for diagnosis.

For coronary artery road map acquisitions, a large through-plane coverage is required to keep the relevant anatomy in the imaging plane. Previous studies have used a conventional gradient-echo sequence to obtain real-time two-dimensional (2D) thick-section projection images of coronary arteries after intraarterial injections of a contrast agent (9). A large flip angle (90°) has been used to suppress signals of the background tissues. To our knowledge, three-dimensional (3D) MR imaging of coronary arteries after intraarterial contrast agent injections has not been reported to date.

The purpose of this study was to assess the feasibility of acquiring coronary artery MR angiograms with intraarterial injections of a contrast agent in dogs. These images would be collected with either high temporal resolution for 2D projection road maps or high spatial resolution for 3D MR angiography. We also aimed to demonstrate that a low dose of contrast agent could be used with a proper injection protocol. An inversion recovery (IR)-prepared technique (10,11) was used to suppress background signals.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Animal Preparations
In vivo imaging experiments were conducted in five hound dogs (Covance, Kalamazoo, Mich) that had a mean weight of 26.5 kg. All experimental procedures were approved by our institutional animal care and use committee. We developed a surgical model of catheter insertion into the canine left circumflex coronary artery (LCX). The surgical procedure involved a lateral thoracotomy to expose the heart. After opening the pericardium, we identified the LCX. We inserted a shortened 22-gauge catheter (Tygon; North Performance Plastics, Akron, Ohio) approximately 3 cm upstream from the bifurcation of the LCX and the left anterior descending coronary artery (Fig 1).

View larger version (141K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Photograph at surgery shows a 22-gauge catheter (arrowhead) inserted into the LCX (arrow) of a dog. The top of the image corresponds to the left side of the dog. The right side of the image corresponds to the caudal end of the dog.

In a similar manner, we also surgically placed two catheters into the left and right atria for delivery of saline. A chest tube was inserted into the left pleural space to reduce negative pressure and assist in withdrawal of pleural fluid. This chest tube was removed 4 days after the initial surgery.

MR imaging experiments were performed a minimum of 3 days after surgical preparation. On the day of the study, the dogs were tranquilized and transported from the animal facility to the MR imager. The dogs were then anesthetized by using methohexital (Brevital; Burns Veterinary Supply, Farmers Branch, Tex) administered through the right atrial catheter, intubated, and placed in the imager. Anesthesia was maintained throughout the experiment with inhalable isoflurane (IsoFlo; Webster, Sterling, Mass). Imaging was performed with a 1.5-T commercially available MR unit (Magnetom Sonata; Siemens Medical Solutions, Erlangen, Germany) with a maximum gradient strength of 40 mT/m and maximum slew rate of 200 mT/m/msec. The animals were sacrificed at the end of the study.

Acquisition of 2D Coronary Road Maps
Two-dimensional coronary projection road maps were acquired in three dogs by one of the investigators (J.D.G.). Projection imaging refers to a 2D MR acquisition in which thick sections are used for angiographic depiction (12). For coronary artery road map acquisitions, it is helpful to use projection imaging to capture the relevant anatomy and to increase image frame rate. A problem inherent in thick-section coronary artery imaging is that the amount of background tissue present in the imaging section is much larger than that of coronary artery blood for any given pixel. With a conventional imaging technique, coronary artery signal intensity could be obscured by background signal intensity because the sum of the signal intensity produced by the background could be substantially greater than the signal intensity of coronary blood. Background suppression is required to boost the relative conspicuity of the coronary artery.

In this study, we used a segmented, IR-prepared fast low-angle shot (FLASH) sequence to collect 2D real-time coronary artery projection images (Fig 2). The sequence divided k space into several segments in an interleaved fashion. Before data acquisition for each segment, a non–section-selective inversion pulse was applied (followed by an inversion time [TI] of 50 msec) to suppress background tissues. A FLASH sequence was used to acquire data. Scout images were first obtained to find the orientation of the LCX. The 2D real-time imaging protocol was applied, and, 1–2 seconds later, contrast agent (gadoteridol injection, ProHance; Bracco Imaging, Milan, Italy) diluted 3%–9% by volume with saline solution was injected into the subcutaneously tunneled LCX catheter by using an automated power injection system (Spectris; Medrad, Indianola, Pa).

View larger version (18K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Sequence diagram of the imaging protocol used for 2D real-time projection imaging. There is no electrocardiographic (ECG) triggering. A non-section-selective -radian inversion pulse is followed by a TI and spoiling gradient. This is followed by the acquisition of N lines (flip angle, ) with use of a standard spoiled gradient-echo acquisition scheme. Immediately after N-line acquisition, the next inversion pulse is applied, and the cycle repeats. TR = repetition time.

To determine the appropriate TI value for background suppression, computer simulations with Matlab (Mathworks, Natick, Mass) were performed with the Bloch equation (13). The simulations showed that a TI of 50 msec resulted in uniform background suppression for a wide range of T1 values (fat, 250 msec; myocardium, 900 msec; nonenhanced blood, 1,300 msec) (Fig 3). Accounting for all parameters, the temporal resolution of 2D real-time imaging was 3 frames per second.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Plot of the TI required for tissue nulling versus tissue T1, as calculated with a simulation of the sequence outlined in Figure 2 with N = 43, repetition time = 2.3 msec, and = 15°. A TI of 50 msec was selected for the canine experiments because it enables signal nulling of a wide range of background tissues.

Three-dimensional Coronary MR Angiography
For the 3D coronary MR angiography experiments, two dogs were imaged by the same investigator as for 2D imaging (J.D.G.). The sequence was ECG triggered, and data were collected during mid-diastole to minimize cardiac motion effects. A 3D FLASH sequence was used for data acquisition. To suppress background tissues, a steady-state and IR magnetization preparation scheme was used (Fig 4). In each cardiac cycle, 100 non–section-selective radio-frequency preparation pulses were applied first, followed by a 180° non–section-selective inversion pulse, the TI, and data acquisition. By the end of the radio-frequency preparation pulse train, the magnetization of all background tissues was in a steady state. On the basis of our simulation results, after the application of the inversion pulse, a single TI could then effectively null the signals of the background tissues with T1 values ranging from 250 msec (fat) to 1,300 msec (nonenhanced blood). The TI and flip angle of the preparation pulses were determined empirically by using the simulation results illustrated in Figure 3 and the results of the 2D experiments as a guide.

View larger version (18K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Diagram of the ECG-triggered 3D MR angiographic sequence. To attain the same level of background suppression accomplished with the 2D protocol, 100 non-section-selective low-flip-angle preparation pulses (prep) are first initiated after the R wave. Preparation pulses are separated by repetition time (TR) preparation pulse (TRprep). At the end of this train of preparation pulses, a non-section-selective 180° pulse is applied, followed by a TI and the acquisition of N phase-encoding lines with FLASH. rf = radio-frequency.

Data Analysis
To calculate the coronary artery signal-to-noise ratio (SNR) in both 2D and 3D images, regions of interest were drawn in the LCX and in the background air outside the dog’s body by the same invetigator (J.D.G.) to measure LCX and background air signal intensity (the area of each region of interest was approximately 0.4–0.7 cm²). The SNR was defined as 1.25 times the signal intensity of the LCX divided by the signal intensity of background air (14). So that we could determine the optimal intraarterial contrast injection parameters for real-time road map acquisitions, the coronary artery SNRs from the nine different contrast injection schemes were compared by using a two-way analysis of variance and the Tukey multiple comparisons procedure (15). In this analysis, both the contrast agent concentration and contrast agent injection rate were independent variables. A P value of less than .05 was considered to represent a statistically significant difference.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Two-dimensional Coronary Road Maps
In all studies, application of the IR FLASH sequence resulted in excellent background suppression and vessel depiction. Figure 5 shows a typical sequence of images from the real-time 2D protocol. After the onset of injection, contrast agent entered the LCX through the catheter located in the proximal portion of the artery and was delivered into the distal portion, where it began to perfuse the myocardium. Owing to the low concentration of the injected contrast agent and the small injection volume, washout from the myocardium was fairly rapid.

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 5. A sample of frames from a 2D non-ECG-triggered real-time MR acquisition depicts the progression of contrast agent through the LCX after intraarterial injection of 3 mL of 9% diluted contrast agent. The temporal resolution was 3 frames per second. A, Before contrast agent enters the LCX, all signal in the field of view is suppressed. B, Contrast agent was injected into the LCX (arrow) through the catheter, which is located in the proximal portion of the artery. C, The LCX enhances, resulting in a coronary MR angiogram. A circumflex marginal artery (arrowhead) is visible in the distal portion of the LCX. D, Contrast agent perfuses into the myocardium (arrows) and, E, begins washing out of the myocardium.

View larger version (30K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Bar graph shows SNR versus contrast agent concentration and injection rate after the intraarterial injection of diluted contrast agent. Changing the injection rate or increasing the concentration more than 6% did not yield statistically significant improvements in SNR. [Gd] = gadolinium chelate concentration, a.u. = arbitrary units.

View larger version (57K): [in this window] [in a new window] [Download PPT slide]   Figure 7. Non-ECG-triggered 2D real-time projection MR images of the LCX of a dog after intraarterial injection of 3 mL of diluted contrast agent over 2 seconds with concentrations of A, 3%; B, 6%; and C, 9%. Depiction of the LCX is clearly better in B (arrow) and in C than in A.

View larger version (87K): [in this window] [in a new window] [Download PPT slide]   Figure 8. A, MR localization (scout) image shows the LCX (arrows) of a dog before injection of contrast agent. B, Maximum intensity projection image from 3D MR angiography of the LCX of the same dog after intraarterial injection of 12 mL of 6% contrast agent. Contrast agent is injected through the catheter in a retrograde fashion from the proximal portion of the artery (arrow). Two circumflex marginal arteries (arrowheads) are visible. The in-plane resolution of the image was 0.9 x 0.8 mm2. The position and orientation of this image are the same as those in A.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

For 2D real-time imaging, images were obtained by using a thick-section projection technique. Thus, they could be obtained rapidly, with little need for extensive orientation before injection of the contrast agent. No postprocessing was required to suppress the background. Temporal resolution was at real-time levels (3 frames per second). Among the tested contrast agent concentrations, 6% was optimal because it yielded statistically the same SNR as the 9% concentration, but with a reduced dose. A 3-mL injection of a 6% dilution of contrast agent is equivalent to the injection of only 0.18 mL of undiluted contrast agent. For a dog weighing 25 kg, 83 such LCX injections could be administered without exceeding the United States Food and Drug Administration–mandated gadolinium chelate dose limit of 0.3 mmol/kg/day.

For 3D imaging, higher spatial resolution is required to obtain diagnostic-quality images. Higher spatial resolution requires longer image acquisition times and, consequently, larger doses of injected contrast agent. However, these images were still obtained with a very small amount of contrast agent: 12 mL of a 6% dilution of contrast agent, equivalent to 0.72 mL of undiluted contrast agent. For a dog weighing 25 kg, 21 such injections for 3D MR angiography of the LCX could be administered without exceeding the Food and Drug Administration–mandated gadolinium chelate dose limit. In the future, the ability to perform this large number of separate contrast agent injections to obtain either 2D road maps or 3D diagnostic coronary MR angiograms will increase the likelihood that complex MR-guided coronary artery interventions can be completed.

The choice of the imaging sequence used in this study was motivated by the need to suppress background tissue across a large volume while retaining high signal intensity of contrast-enhanced blood. With use of a conventional steady-state gradient-echo sequence without magnetization preparation, a large flip angle (close to 90°) must be used to suppress the background signal, as has been reported previously (9). However, the high power deposition resulting from MR of a 90° flip angle may prevent the use of the shortest possible repetition time because of the Food and Drug Administration–mandated limits on the specific absorption rate.

Additionally, flip angle drop-off at the section edges and radio-frequency excitation beyond the desired section thickness for a practical section profile can result in unsuppressed tissues from inside and outside the imaging section contributing data to the image, leading to artifacts and poor vessel depiction. These problems can be avoided by using non–section-selective radio-frequency pulses, but this could make background suppression more difficult because the background signal intensity present in the entire imaging volume may overwhelmingly exceed the vessel signal intensity. A section dephaser in the section-select direction has also been applied for background suppression in projection imaging (16), but this technique is limited by the tissue content within the imaging section.

With IR-prepared data acquisition, T1 weighting and background suppression are created with the inversion pulse, which is non–section-selective, followed by the TI. Data are acquired with a much smaller flip angle than that used in steady-state imaging, and the effects of section profile are greatly reduced.

For ECG-triggered contrast-enhanced MR angiography, IR preparation was found to be useful for improving the contrast-to-noise ratio between blood and background (10,11). However, with use of IR preparation in each cardiac cycle, only background tissues that have a narrow T1 range will be suppressed. If one uses combined steady-state and IR preparation techniques, it is possible to suppress the background tissues uniformly over a wide range of T1 values while retaining high blood signal intensity, as in non–ECG-triggered imaging (17).

This study had several limitations. First, our animal experiments were limited in number. We need to confirm these experiments in a larger sample. Although pigs are anatomically more similar to humans, dogs were used in this study because they tend to have better postsurgical survival rates, and our group has extensive experience with this type of canine experimental model. Second, although the Food and Drug Administration has approved intravenous administration of gadolinium chelates for MR imaging, intraarterial injections represent an unapproved route of administration. Finally, preparation of the animals was not completely analogous to preparation of a human patient for a clinical coronary intervention. Catheters were surgically inserted into the LCX rather than being placed percutaneously. A complete interventional MR procedure must not only be able to yield road maps and diagnostic images, but also the ability to guide catheter movements.

In conclusion, the results of this study demonstrate that MR angiography can depict coronary arteries with intraarterial injections of diluted contrast agent at concentrations as low as 6%. Background can be successfully suppressed by using steady-state and IR preparation. Both real-time 2D projection and high-spatial-resolution 3D MR angiographic images can be obtained. These techniques may prove useful for catheter-based imaging-guided interventional procedures in the coronary arteries.

	FOOTNOTES

 
Abbreviations: ECG = electrocardiographic, FLASH = fast low-angle shot, IR = inversion recovery, LCX = left circumflex coronary artery, SNR = signal-to-noise ratio, TI = inversion time, 3D = three-dimensional, 2D = two-dimensional

Author contributions: Guarantors of integrity of entire study, J.D.G., D.L.; study concepts, J.D.G., D.L., R.A.O., J.P.F.; study design, all authors; literature research, J.D.G., D.L., R.A.O.; experimental studies, all authors; data acquisition, all authors; data analysis/interpretation, J.D.G., D.L.; statistical analysis, J.D.G., D.L.; manuscript preparation, J.D.G., D.L.; manuscript definition of intellectual content, all authors; manuscript editing, D.L., R.A.O.; manuscript revision/review, J.D.G., R.A.O., D.L.; manuscript final version approval, all authors.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Yang X, Bolster BD, Jr, Kraitchman DL, Atalar E. Intravascular MR-monitored balloon angioplasty: an in vivo feasibility study. J Vasc Interv Radiol 1998; 9:953-959.[Medline]
2. Strother C, Unal O, Frayne R, et al. Endovascular treatment of experimental canine aneurysms: feasibility with MR imaging guidance. Radiology 2000; 215:516-519.[Abstract/Free Full Text]
3. Spring DB, Bettmann MA, Barkan HE. Nonfatal adverse reactions to iodinated contrast media: spontaneous reporting to the U.S. Food and Drug Administration, 1978– 1994. Radiology 1997; 204:325-332.
4. Spring DB, Bettmann MA, Barkan HE. Deaths related to iodinated contrast media reported spontaneously to the U.S. Food and Drug Administration, 1978–1994: effect of the availability of low-osmolality contrast media. Radiology 1997; 204:333-337.
5. Omary R, Frayne R, Unal O, et al. MR-guided angioplasty of renal artery stenosis in a pig model: a feasibility study. J Vasc Interv Radiol 2000; 11:373-381.[Medline]
6. Wilke NM, Jerosch-Herold M, Zenovich A, Stillman AE. Magnetic resonance first-pass myocardial perfusion imaging: clinical validation and future applications. J Magn Reson Imaging 1999; 10:676-685.[CrossRef][Medline]
7. Kim R, Wu E, Rafael A, et al. The use of contrast-enhanced magnetic resonance imaging to identify reversible myocardial dysfunction. N Engl J Med 2000; 343:1445-1453.[Abstract/Free Full Text]
8. Omary R, Frayne R, Unal O, Grist T, Strother C. Intraarterial gadolinium-enhanced 2D and 3D MR angiography: a preliminary study. J Vasc Interv Radiol 1999; 10:1315-1321.[Medline]
9. Serfaty J, Atalar E, Declerck J, et al. Real-time projection MR angiography: feasibility study. Radiology 2000; 217:290-295.[Abstract/Free Full Text]
10. Hofman MB, Henson RE, Kovacs SJ, et al. Blood pool agent strongly improves 3D magnetic resonance coronary angiography using an inversion pre-pulse. Magn Reson Med 1999; 41:360-367.[CrossRef][Medline]
11. Li D, Zheng J, Weinmann H. Contrast-enhanced MR imaging of coronary arteries: comparison of intra- and extravascular contrast agents in swine. Radiology 2001; 218:670-678.[Abstract/Free Full Text]
12. Dixon WT, Du LN, Faul DD, Gado M, Rossnick S. Projection angiograms of blood labeled by adiabatic fast passage. Magn Reson Med 1986; 3:454-462.[Medline]
13. Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic resonance imaging: physical principles and sequence design New York, NY: Wiley, 1999; 51-63.
14. Henkelman RM. Measurement of signal intensities in the presence of noise in MR images. Med Phys 1985; 12:232-233.[CrossRef][Medline]
15. Kleinbaum DG, Kupper LL, Muller KE, Nizam A. Applied regression analysis and other multivariable methods 3rd ed. Pacific Grove, Calif: Duxbury, 1998.
16. Unal O, Korosec F, Frayne R, Strother C, Mistretta C. A rapid 2D time-resolved variable-rate k-space sampling MR technique for passive catheter tracking during endovascular procedures. Magn Reson Med 1998; 40:356-362.[Medline]
17. Bos C, Bakker C, Viergever M. Background suppression using magnetization preparation for contrast-enhanced MR projection angiography. Magn Reson Med 2001; 46:78-87.[CrossRef][Medline]

This article has been cited by other articles:

				N. Zorger, M. Volk, O. W. Hamer, M. Lenhart, J. Seitz, B. Butz, and C. Paetzel Intraarterial Gadolinium-Enhanced MR Angiography in Humans for the Detection of Infrainguinal Arterial Stenoses Before and After Percutaneous Angioplasty Am. J. Roentgenol., October 1, 2005; 185(4): 867 - 872. [Abstract] [Full Text] [PDF]

				A.-C. Schulte, G. Bongartz, R. Huegli, M. Aschwanden, K. A. Jaeger, W. Ostheim-Dzerowycz, A. L. Jacob, and D. Bilecen Intraarterial Versus IV Gadolinium Injections for MR Angiography: Quantitative and Qualitative Assessment of the Infrainguinal Arteries Am. J. Roentgenol., September 1, 2005; 185(3): 735 - 740. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2261011848v1 226/1/272    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Green, J. D. Articles by Li, D. Search for Related Content PubMed PubMed Citation Articles by Green, J. D. Articles by Li, D.