Original Contribution

Optimal Technique for Performing Invasive Pulmonary Angiography for Chronic Thromboembolic Pulmonary Disease

Lawrence Ang, MD;  Anna McDivit Mizzell, MD;  Lori B. Daniels, MD, MAS;  Ori Ben-Yehuda, MD;  Ehtisham Mahmud, MD

Lawrence Ang, MD;  Anna McDivit Mizzell, MD;  Lori B. Daniels, MD, MAS;  Ori Ben-Yehuda, MD;  Ehtisham Mahmud, MD

Abstract: High-quality invasive pulmonary angiography is invaluable for the evaluation of chronic pulmonary thromboembolic disease. Optimization of multiple technical factors enables optimal angiography, crucial for identifying both high-grade pulmonary thromboembolic disease warranting surgical resection, and surgically inaccessible disease for interventional and/or targeted medical therapy. Appropriate strategies to address the pitfalls encountered during angiography are highlighted. This manuscript provides detailed guidance in performing hemodynamic assessment and invasive pulmonary angiography for the evaluation of chronic thromboembolic pulmonary disease.

J INVASIVE CARDIOL 2019;31(7):E211-E219.

Key words: chronic thromboembolic, CTEPH, pulmonary angiography, pulmonary hypertension

Chronic thromboembolic pulmonary hypertension (CTEPH) is an under-diagnosed form of pulmonary hypertension, occurring in 0.56%-3.2% of acute pulmonary embolism survivors, with additional patients presenting without previous known embolic events.1,2 Because CTEPH can be effectively treated by pulmonary thromboendarterectomy surgery or by pulmonary balloon angioplasty, the diagnosis of CTEPH and its differentiation from other forms of pulmonary hypertension (PH) is essential in the work-up of patients with PH.2-10 Pulmonary angiography remains an essential step in the diagnosis and evaluation of patients with suspected CTEPH, and can be achieved safely even in patients with severe PH. Optimization of diagnostic angiographic technique is critically important for identifying surgically accessible chronic clot and/or inaccessible distal vessel disease treatable by interventional therapy.11-13 We present herein the optimal, safest approach to produce the highest-quality pulmonary angiography, drawing from over 35 years of experience at the University of California, San Diego Medical Center.14

Patient Selection

Patients referred for invasive pulmonary angiography are those with a high likelihood of pulmonary thromboembolic disease based on medical history and non-invasive pulmonary perfusion imaging.2-5 Most commonly, these individuals carry a formal diagnosis of pulmonary embolism (with or without confirmed deep vein thrombosis), longstanding pacemaker lead or central venous infusion catheter, perfusion defects with ventilation/perfusion mismatch involving multiple lobar regions, or computed tomography pulmonary angiography demonstrating thromboembolic disease. Invasive testing in such individuals provides opportunities to perform the following: (1) digital subtraction angiography of the pulmonary vasculature; (2) hemodynamic measurements to confirm resting or exercise-induced pulmonary hypertension during right heart catheterization (RHC); and (3) preoperative coronary angiography. Each of these is important in the evaluation process.

Vascular Access

When performing pulmonary angiography, venous access that best facilitates selective canalization of each pulmonary artery (PA) should be used. In our experience, the right internal jugular vein (IJ) allows for RHC and bilateral PA angiography in the most expedient manner using standard balloon-tipped, flow-directed catheters. Patients with CTEPH commonly demonstrate enlargement of right heart chambers and depressed cardiac output, each of which poses additional challenges to successful RHC via the femoral veins.

Vascular ultrasonography is optimally used to evaluate the caliber, patency, and collapsibility of the IJ prior to venipuncture. A 7 Fr vascular introducer sheath is used in the majority of procedures, while 8 Fr sheaths are reserved for larger patients (ie, men >6.5 feet tall) for whom pulmonary angiography requires higher contrast injection flow rates. Occasionally, the right IJ is completely thrombosed and only small-caliber collateral veins are observed on ultrasound images. The jugular vein can also appear normal, while the superior vena cava is obstructed more distally and prevents safe passage of a guidewire into the right atrium. In such cases, access of either the left IJ or a femoral vein is required. The right femoral vein is preferred over the left femoral vein, given its relatively straight course to the right atrium. Alternative access using the subclavian veins may increase risk of pulmonary injury, while brachial vein access can be problematic due to overhead patient arm positioning during biplane angiography. Use of the left IJ or femoral veins may require the use of a 0.25˝ J-tipped wire to guide the balloon-tipped catheters. Advancing catheters between inferior vena cava filter struts is also feasible, but must be done carefully, with the balloon deflated and under fluoroscopic visualization.

Right Heart Catheterization

RHC with pressure and cardiac output measurements is routinely performed prior to pulmonary angiography using a standard balloon-tipped Swan-Ganz thermodilution PA catheter (Edwards Lifesciences). Calculation of pulmonary vascular resistance requires accurate pulmonary capillary occlusion (wedge) pressure and cardiac output measurements. Accurate leveling of the transducer is essential for obtaining true zero. Computed tomography (CT) analysis of PH patients has demonstrated that placement of the transducer at 1/3 thoracic diameter below the anterior chest level corresponds most accurately to the mid-right atrium, which is the assumed zero.15 Pouch defects, or concaved proximal pulmonary vascular occlusions (common in the right descending PA), may preclude wedging the PA balloon catheter or produce a hybrid tracing.15 Typically, repositioning the catheter in the left descending PA with guidewire assistance allows for successful occlusion measurement. Deflating the balloon slightly may also allow the catheter to wedge in a smaller branch beyond a dilated segment. Extreme caution must then be exercised in inflating the balloon once the catheter is in the distal position to avoid PA rupture, a very rare but catastrophic complication. Confirmation of the wedge position can be verified by documenting the typical respiratory variations or, when in doubt, by gently obtaining a blood sample while in the wedge position and confirming arterial oxygenation.

Selective Pulmonary Angiography

Optimal pulmonary angiography to confirm the diagnosis of CTEPH is focused on identifying both obvious and subtle abnormalities (including PA branch occlusions, pouch defects, lining thrombus along the vascular wall, abrupt vessel tapering, intravascular bands, and web-like filling defects) with selective biplane angiography performed of each lung (Figure 1). Angiographic imaging is also prolonged to record levophase pulmonary vein drainage and exclude pulmonary vein stenosis or anomalies. Subselective invasive angiography of individual PA segments is feasible and provides the highest-quality images for identifying pulmonary thromboembolic disease (Figure 2); this is essential for balloon pulmonary angioplasty but is both highly impractical and unnecessary for diagnosing CTEPH and informing surgical candidacy. Rather, this indispensable technique is reserved for identifying target segments during balloon pulmonary angioplasty.

In our experience, power contrast injection for pulmonary angiography can be routinely and safely performed even in those with severe PH.14,16 While elevated PA pressure alone does not preclude the use of power contrast injection, those with severely decompensated congestive heart failure, significantly reduced cardiac output, or other evidence of end-organ failure (eg, acute renal failure) have pulmonary angiography postponed until organ dysfunction is medically optimized. Intravenous contrast bolus administration for pulmonary angiography can adversely contribute to volume overload by the effects of volume expansion and worsening of renal function. However, the majority of patients present for this procedure in a compensated low cardiac output state and can proceed immediately from RHC to selective pulmonary angiography. At our institution, the non-ionic, isosmolar contrast agent iodixanol (Visipaque; GE Healthcare) is routinely used to help mitigate the potential risk of contrast-induced acute kidney injury.17

Catheter Selection

A number of catheters are used for contrast injection within the PAs. However, to optimize patient safety, angiographic quality, and operator ease-of-use, the standard Berman catheter (Teleflex), is ideally suited for this purpose (Figure 3A). The Berman is a balloon-tipped, end-capped, pressure-rated catheter that is safely and easily advanced through the right heart into the PA. The end cap protects the distal pulmonary vasculature from iatrogenic injury during high-pressure injections. Instead, multiple fenestrations along the catheter tip allow for pressure diffusion and rapid contrast egress (Figure 3B). The flow-directed, pre-shaped Berman usually tracks into the right PA with ease, so it can then be imaged first. Catheter redirection from the right PA to the left PA is easily performed by withdrawing the catheter to the main PA, inflating the balloon, straightening the catheter tip (using the stiff end of any 0.035˝ guidewire within the central lumen and abutting the distal cap), and advancing the catheter into the more vertical takeoff of the left PA. Once the catheter clearly enters the proximal left PA, the wire can be retracted from the tip and the catheter advanced more distally via flow direction alone. Advancement of the relatively stiff Berman catheter through the right heart is more difficult when performed from the femoral vein approach and should be avoided if possible. Other catheter options, such as standard Swan-Ganz, pigtail, Omni Flush (Angiodynamics), or other pre-shaped end-hole catheters, are also used for pulmonary angiography but have notable limitations. Standard Swan-Ganz catheters are not designed to withstand high-pressure injections and are limited to manual contrast injection through the end hole, which is insufficient for dense vessel opacification. High-pressure contrast injection through pre-shaped end-hole catheters is not recommended due to the risk of iatrogenic vessel injury. Use of relatively bulky pigtail and Omni Flush catheters,18,19 while compatible with high-pressure injections and able to be advanced over prepositioned guidewires, can pose challenges during position adjustment within distal branches or redirection to the contralateral lung (Figure 4). Suboptimal catheter shape may also increase risk of vessel injury during manipulation. The Omniflush catheter also has directed fenestrations that promote retrograde contrast flow toward the contralateral PA.

Catheter Positioning

Once the Berman catheter is advanced close to pulmonary capillary wedge position in the lung of interest, the distal balloon is deflated and the central lumen is flushed. During deep inspiration and breath hold by the patient, the catheter position is optimized to allow for small movement of the catheter tip with each heart beat (approximately 1-2 cm of migration during each beat). If the catheter is advanced too deeply, then the tip will appear immobile during the cardiac cycle and indicate positioning within surrounding thromboembolic material or a small-caliber PA segment (Figure 5). Manual or high-pressure contrast injection in this position risks iatrogenic vessel injury. However, if the catheter is not advanced sufficiently, the tip and proximal segment will swing widely during the cardiac cycle. High-pressure contrast injection in this position may cause the catheter to straighten and “kick” back to the main PA segment, resulting in unwanted contrast reflux into the contralateral lung and insufficient vessel filling in the lung being imaged (Figure 6). The optimal position for the Berman tip is within the descending PA distal to the lower-lobe superior segment (ie, within the basal trunk), which allows for simultaneous opacification of the middle and lower-lobe vessels, quickly followed by backfilling of upper-lobe vessels. The vasculature of these middle and lower lobes typically bears greater disease and is of higher clinical interest during angiography. In instances of lower-lobe occlusion, the catheter tip is advanced in the same manner, allowing for some catheter tip migration during a breath hold, but consequently in a more proximal and less deeply seated position (Figures 7 and 8).

Image Composition

Biplane pulmonary angiography requires attention to the following: (1) patient positioning; (2) flat detector angulation; (3) collimation; and (4) imaging system settings. Once the Berman catheter is positioned within the PA of interest, the patient raises and maintains both arms above shoulder level (usually with the patient’s hands clasped behind the head) to limit humeral bone interference and motion artifact. Arm boards and other overlying objects are removed from the fields of view if possible.

When composing right pulmonary angiograms, the frontal and lateral flat detectors are kept in straight anterior-posterior (AP) and left lateral (left anterior oblique [LAO] 90º) views, respectively. Frontal views are collimated around the right thoracic cavity. Superior collimation excludes overlap of clavicle and lung, while inferior collimation includes maximal lung field. Lateral collimation is set at the lateral lung border, while medial collimation includes medial lung fields and the right PA (after bifurcation from the main PA trunk).

Left PA angiography can be performed in straight AP and left lateral views. Rotation of views 20º leftward (to LAO 20º and LAO 110º, respectively) is performed to decrease overlap of the mediastinum and left lung. Left frontal collimation is similar to that of the contralateral lung, but the left PA has a more vertical (less horizontal) course than the right PA, and medial border collimation more closely approaches the lung border.

Lateral angiograms for both lungs are collimated superiorly and inferiorly around the lung borders, similar to frontal angiograms. Anterior and posterior collimation exclude the chest wall borders and include maximal lung field. The field of view can also be centered around the distal segment of an optimally positioned Berman catheter, and the resulting angiogram will demonstrate multiple vascular segments emanating from the center of the image (Figure 1A).

Use of digital subtraction is highly recommended and routinely performed at our institution. Frontal and lateral angiography is performed at 4 frames/second during contrast injection, then decreased to 1 frame/second during levophase. Slower frame rates can result in acquisition of fewer useful diagnostic images, which is particularly troublesome when artifacts degrade the available images.

Contrast Injection Settings

Once the catheter, patient, image detectors, and collimators are optimally positioned, parameters for contrast injection are programmed for the power injector. Injection duration and flow rate vary between patients, while pressure rise and maximum pressure settings are left constant. A short pressure rise between 0.1-0.3 seconds (typically 0.2 seconds) is used to further reduce catheter migration during contrast injection. Maximum pressure is typically set at 600 psi and can be adjusted based on catheter pressure rating.

Injection flow rate is most related to the extent of contrast backfilling into the proximal PA segments and even inadvertent overflow into the contralateral lung, while injection duration allows for more sustained backfilling of vessel segments that are reached. For the majority of patients with impaired cardiac output and slower clearance of contrast from the pulmonary vasculature, shorter injections of 2.0 seconds are typically used. For those who demonstrate preserved cardiac output and more brisk contrast clearance, or to address instances when contrast briefly backfills the upper-lobe vessels but vessel opacification is suboptimal, injection duration is prolonged to 2.5 seconds.

A few general factors guide the injection flow rates and durations for angiograms in each lung: body size, cardiac index, and disease severity (Table 1). Patient height is an easy and reliable parameter to approximate the size of the thoracic cage, lungs, and pulmonary vasculature. Shorter individuals with smaller lungs require lower injection rates and volumes compared with taller individuals. Further reduction of injection flow rate can be applied to those with significantly depressed cardiac index due to right and/or left heart failure, or extremely severe disease. We routinely administer a 5-10 mL test injection of 1:1 saline to contrast mixture through the optimally positioned Berman catheter to observe flow through the PA tree and confirm our desired flow rate, duration, and total volume. PAs with brisk flow (preserved cardiac index, lower disease burden) are imaged with higher injection flow rates and volumes, while those with sluggish flow (highly impaired cardiac index and/or greater disease burden) are imaged with lower injection flow rates and volumes. In general, higher injection flow rate (resulting in brief contrast spillover to the contralateral lung and image degradation) is preferred over an insufficient injection rate (resulting in insufficient opacification of proximal and upper-lobe vessels) and repeat injections. Vessels with extremely high disease burden (eg, multiple lobar arteries occluded) can be imaged using significantly reduced contrast flow rates and volumes, or even a firm manual injection. Some patients have such extensive disease that only a minimal injection (eg, 10 mL) is sufficient to opacify the patent vessels (Figure 8).

Image Acquisition

Once contrast injection settings are programmed and various positionings are finalized, imaging technique can be established and angiography performed. To appropriately set automatic brightness control settings, lateral and frontal fluoroscopy may need to be triggered for a few seconds during a patient’s deep breath hold without contrast injection. Once the imaging technique is established in this manner, final angiography can be performed. One staff member clearly administers instructions and guides the patient into a deep, prolonged breath-hold maneuver. A second staff member carefully observes the patient initiate a breath hold and completely relax the abdomen, then activates the coupled biplane imaging system and contrast injector until levophase drainage of the pulmonary veins into the left atrium is observed. These final maneuvers are repeated for imaging of the contralateral lung (Table 2).

Image Artifact and Optimization

The most common culprit for poor pulmonary angiography quality is motion artifact of the lung and diaphragm during image acquisition. Some patients are unable to completely sustain a deeply inhaled breath due to their underlying lung disease. Breath release during image acquisition results in upward migration of the diaphragm and PA blurring (Figure 4). Images can also be acquired prematurely, while the patient’s abdomen is still relaxing during a breath hold, which results in blurred appearance of the diaphragm on masked images. Slightly delaying image acquisition until the patient’s chest and abdomen appear completely motionless can help avoid the latter issue. Programming a short delay between start of image acquisition and start of contrast injection (ie, 1.5 seconds) allows operators to further observe diaphragm motion and the option to abort contrast injection. Selection of a subtraction mask acquired immediately before contrast injection (and after initial patient motion has occurred) also reduces artifact severity.

Excessive backfilling of the proximal PA and spillover to the contralateral lung will impact the quality of lateral angiograms. Anteriorly directed PA segments in the lung of interest can be masked by opacification of the main PA trunk, while details of the remaining target segments can be obscured by background filling of contralateral vessels (Figure 6). Optimizing Berman catheter position and injection flow rate to avoid backfilling of the contralateral lung helps avoid this issue. Areas of excessive image brightness, or “burnout,” are due to suboptimal automatic brightness control settings and overexposure. Vascular structures appear absent in overexposed areas and are difficult to differentiate from vessels occluded due to thromboembolic disease. Optimization of automatic brightness control technique can avoid overexposure and angiogram degradation. Digital subtraction can also exacerbate this artifact, while viewing unsubtracted angiograms may be helpful.

Suboptimal digital processing of raw angiograms can result in poor balance of image brightness (dark vs bright) and/or image contrast (washed-out vs grainy). High-quality angiograms can be further improved by using digital postprocessing techniques to optimize these image elements (Figure 9) and ensure preservation of optimized images for future review. Collaboration between imaging system engineers, lab technologists, and physicians is crucial to perfect these techniques.

Basic Angiogram Interpretation

Each PA typically branches into ten different segments (Figure 10), with subtle differences of nomenclature between the right and left lungs (Table 3). Within the right lung, the superior lobar artery supplies the upper lobe and branches into apical (A1), posterior (A2), and anterior (A3) segments. Proximal branches emanating from the descending interlobar artery course anteriorly to the lateral (A4) and medial (A5) regions of the middle lobe, and posteriorly to the superior segment of the lower lobe (A6). The ongoing basal trunk branches into the medial basal (A7), anterior basal (A8), lateral basal (A9), and then posterior basal (A10) segments. By convention, the left lung may feature combined apicoposterior (A1+A2) and anteromedial (A7+A8) segments that each share common trunks. The left upper-lobe lingula also divides into superior (A4) and inferior (A5) segments, in contrast to lateral (A4) and medial (A5) segments of the right middle lobe.

Abnormalities on PA angiography indicating surgically resectable thromboembolic disease have been previously described. Classic findings include proximal PA occlusions (Figures 7 and 8), pouch defects, webs, bands, and abrupt vascular narrowing (Figure 9).20 Major vessel counting can help identify proximal occlusion of PA segments, but anatomical variation can make vessel identification imprecise. Notably, the posterior segment (A2) can branch off the apical segment (A1), while the medial basal segment (A7) can share an ostium with the anterior basal segment (A8) or be absent altogether. Angiograms are systematically interpreted from the proximal PA with characterization of the lumen wall, to the upper-lobar artery and its segments, then middle-lobar artery or lingular segments, then lower-lobe superior segments, and finally lower-lobe basal trunk and its segments. The most important aspect of interpreting pulmonary angiograms is identification of proximal vessel disease of the main PA, lobar arteries, and proximal segments that can generally be resected by surgical endarterectomy. In the absence of proximal vessel disease, angiographic findings of distal segmental or subsegmental artery disease rarely inform the decision to proceed to surgery in isolation from severity of PH, right ventricular dysfunction, patient symptoms, and comorbidities.21 High-quality angiograms with such findings are also useful as roadmaps to guide percutaneous revascularization by balloon pulmonary angioplasty.


High-quality invasive pulmonary angiography is invaluable for the evaluation, confirmation, and treatment planning of chronic pulmonary thromboembolic disease. Simultaneous optimization of multiple technical factors allows high-quality angiography, which is crucial for identifying both high-grade proximal pulmonary thromboembolic disease warranting surgical resection, and surgically inaccessible disease for interventional and/or targeted medical therapy.

Acknowledgments. The authors would like to recognize and thank Aaron Farmer, RT, for his invaluable experience performing and teaching high-quality angiography at UC San Diego.


1. Ende-Verhaar YM, Cannegieter SC, Vonk Noordegraaf A, et al. Incidence of chronic thromboembolic pulmonary hypertension after acute pulmonary embolism: a contemporary view of the published literature. Eur Respir J. 2017;49:1601792.

2. Mahmud E, Madani MM, Kim NH, et al. Chronic thromboembolic pulmonary hypertension: evolving therapeutic approaches for operative and inoperative disease. J Am Coll Cardiol. 2018;71:2468-2486.

3. Galiè N, Humbert M, Vachiery JL, et al; ESC Scientific Document Group. 2015 ESC/ERS guidelines for the diagnosis and treatment of pulmonary hypertension: the Joint Task Force for the Diagnosis and Treatment of Pulmonary Hypertension of the European Society of Cardiology (ESC) and the European Respiratory Society (ERS): Endorsed by: Association for European Paediatric and Congenital Cardiology (AEPC), International Society for Heart and Lung Transplantation (ISHLT). Eur Heart J. 2016;37:67-119.

4. Mehta S, Helmersen D, Provencher S, et al. Diagnostic evaluation and management of chronic thromboembolic pulmonary hypertension: a clinical practice guideline. Can Respir J. 2010;17:301-334.

5. Wilkens H, Lang I, Behr J, et al. Chronic thromboembolic pulmonary hypertension (CTEPH): updated recommendations of the Cologne Consensus Conference 2011. Int J Cardiol. 2011;154:S54-S60.

6. Fedullo PF, Auger WR, Kerr KM, Rubin LJ. Chronic thromboembolic pulmonary hypertension. N Engl J Med. 2001;345:1465-1472.

7. Piazza G, Goldhaber SZ. Chronic thromboembolic pulmonary hypertension. N Engl J Med. 2011;364:351-360.

8. Kim NH, Delcroix M, Jenkins DP, et al. Chronic thromboembolic pulmonary hypertension. J Am Coll Cardiol. 2013;62:D92-D99.

9. Kharat A, Hachulla AL, Noble S, Lador F. Modern diagnosis of chronic thromboembolic pulmonary hypertension. Thromb Res. 2018;163:260-265.

10. Coulden R. State-of-the-art imaging techniques in chronic thromboembolic pulmonary hypertension. Proc Am Thorac Soc. 2006;3:577-583.

11. Lang IM, Madani M. Update on chronic thromboembolic pulmonary hypertension. Circulation. 2014;130:8-18.

12. Madani M, Ogo T, Simonneau G. The changing landscape of chronic thromboembolic pulmonary hypertension management. Eur Respir Rev. 2017;26:170105. Print 2017 Dec 31.

13. Jamieson S, Pretorius GV. Chronic thromboembolic pulmonary hypertension. Semin Intervent Radiol. 2018;35:136-142.

14. Guillinta P, Peterson KL, Ben-Yehuda O. Cardiac catheterization techniques in pulmonary hypertension. Cardiol Clin. 2004;22:401-415.

15. Kovacs G, Avian A, Olschewski A, Olschewski H. Zero reference level for right heart catheterisation. Eur Respir J. 2013;42:1586-1594.

16. Nicod P, Peterson K, Levine M, et al. Pulmonary angiography in severe chronic pulmonary hypertension. Ann Intern Med. 1987;107:565-568.

17. Eng J, Wilson RF, Subramaniam RM, et al. Comparative effect of contrast media type on the incidence of contrast-induced mephropathy: a systematic review and meta-analysis. Ann Intern Med. 2016;164:417-424.

18. Nilsson T, Carlsson A, Mâre K. Pulmonary angiography: a safe procedure with modern contrast media and technique. Eur Radiol. 1998;8:86-89.

19. Velling TE, Brennan FJ, Hall LD. Pulmonary angiography with use of the 5-F Omniflush catheter: a safe and efficient procedure with a common catheter. J Vasc Interv Radiol. 2000;11:1005-1008.

20. Kawakami T, Ogawa A, Miyaji K, et al. Novel angiographic classification of each vascular lesion in chronic thromboembolic pulmonary hypertension based on selective angiogram and results of balloon pulmonary angioplasty. Circ Cardiovasc Interv. 2016;9:e003318.

21. Madani MM. Surgical treatment of chronic thromboembolic pulmonary hypertension: pulmonary thromboendarterectomy. Methodist Debakey Cardiovasc J. 2016;12:213-218.

From the University of California San Diego, Sulpizio Cardiovascular Center, La Jolla, California.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. The authors report no conflicts of interest regarding the content herein.

The authors report that patient consent was provided for publication of the images used herein.

Manuscript submitted February 4, 2019, provisional acceptance given February 12, 2019, final version accepted February 19, 2019.

Address for correspondence: Ehtisham Mahmud, MD, University of California San Diego, Sulpizio Cardiovascular Center, 9434 Medical Center Drive, La Jolla, CA 92037. Email: emahmud@ucsd.edu