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) 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 Westra, S. J. Articles by Laks, H. Search for Related Content PubMed PubMed Citation Articles by Westra, S. J. Articles by Laks, H.

Cardiac Electron-Beam CT in Children Undergoing Surgical Repair for Pulmonary Atresia¹

¹ From the Department of Radiological Sciences (S.J.W., J.H., M.F.M.G., M.I.B.) and the Divisions of Pediatric Cardiology (A.G.) and Cardiothoracic Surgery (H.L.), UCLA School of Medicine, Los Angeles, Calif. From the 1997 RSNA scientific assembly. Received October 29, 1998; revision requested December 28; revision received February 2, 1999; accepted April 30. Address reprint requests to S.J.W., Department of Radiology, Children's Memorial Hospital, 2300 Children's Plaza #9, Chicago, IL 60614-3394 (e-mail: SWestra@childrensmemorial.org).

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To study whether electron-beam computed tomography (CT) is as accurate as conventional angiocardiography for the characterization of the true pulmonary arteries and the aortopulmonary collateral vessels in children undergoing surgical correction for pulmonary atresia.

MATERIALS AND METHODS: Twenty-three children with pulmonary atresia underwent 48 cardiac-triggered dynamic contrast material–enhanced electron-beam CT studies. Correlation was made with surgical findings in all patients and with 34 cineangiocardiograms. Data from reconstructed electron-beam CT images and cineangiocardiograms were reviewed for the presence, caliber, and origin of true pulmonary arteries and aortopulmonary collateral vessels; for stenosis; for thrombosis; and for the patency of vascular conduits and shunts.

RESULTS: Electron-beam CT was more sensitive than angiography in the identification of hypoplastic and/or nonconfluent branch pulmonary arteries, coronary anomalies, conduit and shunt thrombosis, and other postoperative complications, but it was less sensitive in the demonstration of stenoses at collateral vascular origins and anastomoses. Overall test parameters for electron-beam CT and angiography to characterize pulmonary vascularity were similar (sensitivity, 0.94 vs 0.90; specificity, 0.99 for both; accuracy, 0.97 vs 0.95). Three-dimensional reconstructions, although they were helpful in conveying electron-beam CT findings to referring cardiologists and surgeons, did not add diagnostic information to that displayed on images of the transverse sections.

CONCLUSION: Electron-beam CT complements conventional diagnostic angiocardiography in preoperative evaluation, especially in the detection of hypoplastic pulmonary arteries. It is well suited for postoperative shunt surveillance.

Index terms: Angiocardiography, 51.1242, 56.1242 • Computed tomography (CT), electron beam, 51.12113, 56.12113 • Computed tomography, in infants and children • Computed tomography, three-dimensional, 51.12117, 56.12117 • Heart, abnormalities, 515.142, 533.171 • Heart, CT, 51.12113, 51.12117, 56.12113, 56.12117 • Pulmonary arteries, abnormalities, 564.155

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Pulmonary atresia with ventricular septal defect and multiple aortopulmonary collateral vessels is a complex cyanotic heart lesion (1). The optimal surgical approach to this lesion remains controversial, but staged repairs are favored in certain patients who are considered to be suitable candidates for surgery. The technique for the repair varies widely between patients and even between institutions that have experience in the treatment of this complex anomaly. Most authorities recommend ligation of the aortopulmonary collateral vessels early in life (2), with the creation of an anastomosis from the collateral vessels to the true pulmonary arterial system (when present) and, thereby, with the conversion of these high-pressure systemic vessels to lower pressure pulmonary arteries. This must be performed before systemic pressures in these collateral vessels lead to the development of pulmonary vascular disease and refractory pulmonary hypertension (Eisenmenger physiology).

In our institution, a specific technique is favored that involves the construction of a tube, made of pericardium, to which both the true pulmonary arteries (when present) and the aortopulmonary collateral vessels are "unifocalized" (3–5). These pericardial tubes are extended anteriorly by means of Gore-Tex (expanded polytetrafluoroethylene [ePTFE] membrane; W.L. Gore & Associates, Flagstaff, Ariz) conduits, which are temporarily perfused via modified Blalock-Taussig shunts (from the subclavian arteries) and/or central shunts (from the aorta). The pressure in these unifocalization conduits can be down-regulated to levels within the physiologic range for pulmonary arteries. The initial unifocalization surgery is performed consecutively, by means of a lateral thoracotomy, on both sides (See Fig 1c for a diagrammatic representation of a left-sided pericardial tube unifocalization and a right-sided central shunt placement. An explanation of the letters in the figures appears in the Key Box). The final repair, performed by means of a median sternotomy, involves connection of both unifocalization conduits with a transverse ePTFE membrane graft, reconstruction of the pulmonary outflow tract with placement of a porcine valve in the pulmonary position, and closure of the ventricular septal defect with a patch (Fig 2). The end result of the final repair, therefore, is complete separation of the pulmonary and systemic circulations, which results in a marked improvement in cyanosis.

View larger version (119K): [in this window] [in a new window]   Figure 1a. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (149K): [in this window] [in a new window]   Figure 1b. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (156K): [in this window] [in a new window]   Figure 1c. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (122K): [in this window] [in a new window]   Figure 1d. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (128K): [in this window] [in a new window]   Figure 1e. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (166K): [in this window] [in a new window]   Figure 1f. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (117K): [in this window] [in a new window]   Figure 1g. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (125K): [in this window] [in a new window]   Figure 1h. Images depict staged surgical repair in a 9-year-old boy with pulmonary atresia and absent pulmonary arteries. (a) Transverse thick-slab maximum intensity projection electron-beam CT image, obtained after the right unifocalization procedure, shows a dilated ascending aorta (A), which is overriding a high ventricular septal defect (not shown), and a left-sided aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch. (b) Corresponding image obtained 5 days after the left unifocalization procedure shows a new pericardial tube (P), which has been placed on the left and to which the left-sided collateral vessels have been anastomosed after ligation from the aorta. The left unifocalization structure (U, P) is temporarily perfused via a modified Blalock-Taussig shunt (B). The origin of a large, previously ligated, right-sided collateral vessel (C) is also seen at this level. (c) Diagram depicts the postoperative anatomy in b. The ePTFE membrane portions (U, B) are depicted in light blue, and the pericardial tube of the left-sided unifocalization (P), which is anastomosed with the aortopulmonary collateral vessels, is depicted in yellow. The * indicates the catheter position for the angiograms in d and e. (d) Posteroanterior and (e) (on p 505) lateral angiograms obtained with selective injection of the left unifocalization structure via the modified left-sided Blalock-Taussig shunt shows patent anastomoses with aortopulmonary collateral vessels (C). (f, g) Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels after final repair surgery show that both unifocalization conduits (U in f) have been connected by means of a reinforced (corrugated) transverse ePTFE membrane graft, which is anastomosed (arrows in g) with the reconstructed pulmonary outflow tract, which contains a porcine neopulmonary valve (PV in g). The ventricular septal defect was also closed with a patch (not shown). Note the decrease in the caliber of the ascending aorta (A) relative to that of descending aorta (D), as compared with that depicted in a and b, which resulted from the separation of the cardiac output into the neopulmonary and systemic circulations. (h) Coronal thick-slab maximum intensity projection electron-beam CT image shows to better advantage the anastomoses between the original aortopulmonary collateral vessels and the pericardial tubes (P) and shows the large blind-ended stump (C) of a ligated right-sided aortopulmonary collateral vessel, which was also shown in b.

View larger version (178K): [in this window] [in a new window]   Figure 2. Diagram depicts the postoperative anatomy after complete repair for pulmonary atresia in a patient with confluent branch pulmonary arteries. Aortopulmonary collateral vessels have been ligated from the descending aorta (D) and have been anastomosed with bilateral pericardial unifocalization tubes (P). The pulmonary arterial confluence has been connected with the left unifocalization tube, whereas on the right, a smaller shunt was placed to unifocalize the distal portion of the right branch pulmonary artery (R) to the pericardial tube. The main pulmonary artery has been reconstructed with a ePTFE membrane tube, and a porcine neopulmonary valve (PV) has been placed at its inferior aspect to connect the reconstructed main pulmonary artery with the right ventricle. A reinforced (corrugated) ePTFE membrane tube connects the reconstructed main pulmonary artery with the right unifocalization tube. The ventricular septal defect has been closed with a patch (arrows), which resulted in complete separation of the neopulmonary circulation from the systemic circulation.

Conventional, incremental CT has been used since the early 1980s for the evaluation of the pulmonary arteries in congenital heart disease (6–8). The high contrast resolution at CT allows contrast material to be injected into a peripheral vein; this makes CT a much less invasive procedure than conventional angiocardiography. The evaluation of congenital heart disease and anomalies in the great vessels has been further improved with the development of electron-beam (fast) CT (9–11) and helical CT techniques (12–14). Therefore, CT now has additional advantages over angiography in terms of cost, speed of examination, radiation dose, and lack of the need for sedation in most patients.

We previously reported our experience with helical CT with three-dimensional image rendition in the evaluation of the pulmonary arteries in children with a variety of congenital heart lesions (15). We found CT to be superior to echocardiography and as accurate as angiocardiography. Although the cross-sectional imaging paradigm with CT theoretically should represent an advantage over conventional projectional angiocardiography, we were not able to prove that hypothesis because of our retrospective study method. Therefore, we performed this prospective study of the use of the optimal CT scanner for cardiac studies (the electron-beam CT scanner) and a well-defined imaging protocol in the pre- and postoperative assessment of patients with a single, well-defined clinical condition: pulmonary atresia. We tested the above hypothesis to answer the following questions: Is electron-beam CT equivalent or superior to angioardiography for the preoperative characterization of pulmonary vasculature and diagnosis of postoperative complications? Can electron-beam CT replace some portions of diagnostic angiocardiography?

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Patients
Over 3 years, 23 patients with pulmonary atresia and ventricular septal defect with multiple aortopulmonary collateral vessels who were potential candidates for unifocalization and/or complete repair operations and who underwent at least one electron-beam CT examination were enrolled in this study. There were 15 boys and eight girls (age range, 4 months to 13 years; median age, 6 years 4 months).

In our institution, electron-beam CT, in conjunction with angiocardiography, is part of the routine cardiac evaluation in patients undergoing surgical repair for severe cyanotic congenital heart lesions. All 48 electron-beam CT studies performed during this study were, therefore, performed on the basis of clinical indication. That is, they were performed for the initial assessment of the pulmonary vasculature prior to the first unifocalization operation (n = 7) or for postoperative assessment in patients who underwent at least one stage of the three-stage surgical repair procedure (n = 41) to assess suspected shunt and conduit stenosis or thrombosis, airway compression, and pulmonary parenchymal and/or pleural disease.

The majority of these 41 studies could be regarded as postoperative for one stage of the repair (eg, left unifocalization procedure) and, at the same time, as preoperative for another stage (eg, subsequent right unifocalization or final repair procedure). Of the 30 operations that were performed in the 23 patients during the study period, 13 were evaluated at electron-beam CT pre- and postoperatively.

Electron-Beam CT Protocol
Our protocol for the injection of contrast material at CT angiocardiography has been previously described in detail (15). In summary, 2 mL of a nonionic contrast material (Isovue 300 [iopamidol]; Squibb Diagnostics, New York, NY) per kilogram of body weight was injected by means of a power injector into the largest available vein in an extremity. All injection sites were tested by means of a rapid manual injection of a bolus of normal saline prior to the injection of contrast material. The injection rate was set between 0.5 and 2.5 mL/sec to deliver the total contrast dose over a time interval equal to the estimated duration of imaging. This injection was immediately followed with a bolus injection of 10 mL of normal saline.

Imaging was initiated 15–20 seconds after the beginning of the injection of contrast material, and this delay time was empirically determined on the basis of estimated times for circulation from the injection site to the aorta, without the use of a timing bolus. All images were obtained with the same electron-beam CT scanner (C-150XP; Imatron, South San Francisco, Calif). Incremental scanning of the heart and great vessels was performed in an inferosuperior direction from the inferior pulmonary veins to the thoracic inlet. In nine of the 48 electron-beam CT studies, a brief sedation was required, primarily in moving infants and in uncooperative young children.

Images in all infants and young children were acquired during quiet respiration, but in cooperative older children, image acquisition was generally performed during a single breath hold. We used the following technical imaging parameters: section collimation, 3 mm; scanning time per section, 100 msec; 130 kVp; and 630 mA. The table feed was determined from the need for longitudinal coverage (patient size) as follows: In infants and young children, a 2-mm table feed was used, which resulted in a 33% overlap of the sections. In older children, a contiguous nonoverlapping imaging profile was acquired with a 3-mm table feed per section. Section acquisition was electrocardiographically triggered at 80% of the R-R interval, with the acquisition of one section per heart beat at heart rates under 120 beats per minute or one section per every second heart beat at higher heart rates. The total number of sections depended on the need for longitudinal coverage and varied between 30 and 65. The total duration for imaging was dependent on the number of sections and on the heart rate; it varied between 25 and 55 seconds.

Postprocessing
All transverse sections were viewed interactively with soft-tissue and lung window settings on a commercially available computer workstation (Magic View; Siemens Medical Systems, Erlangen, Germany) by rapidly paging up and down through the transverse images to characterize pulmonary blood vessels by their central connections as pulmonary arteries, collateral vessels, or veins. The data from the transverse images were compressed into a number of thick-slab maximum intensity projections with variable thicknesses. These thick-slab maximum intensity projections were edited interactively by the user, who used cut-plane techniques to remove interposing ribs and portions of the spine. The resultant truncated images of the native sections were used to generate shaded-surface display images (with a threshold of 150 HU).

In addition, all image files were transferred to an O2 workstation (Silicon Graphics, Mountain View, Calif), and volume renditions were obtained by using the Vitrea 1.1 software (Vital Images, Minneapolis, Minn) for three-dimensional image rendition and analysis. We used the software application, with cut-plane and zoom capabilities, that was customized for viewing electron-beam CT image data and that was based on real-time, user-interactive, multiplanar thick-slab maximum intensity projection renditions. Display parameters, such as window width, level, opacity, and brightness, were chosen subjectively by the reviewing radiologist (S.J.W.) to best depict particular anatomic features of interest in the study. Selected frames were recorded on hard-copy film, and the entire set of images was saved on a magneto-optical disk for future reference. All postprocessing and imaging of the three-dimensional renditions was performed by the same radiologist. Postprocessing time for three-dimensional reconstruction, including printing, was approximately 30 minutes per study, and this cost was factored into the overall technical and professional billing fee for the procedure.

Angiographic Technique
Angiography was performed in all but one of the patients: a 4-month-old girl with absent pulmonary arteries and severely hypoplastic aortopulmonary collateral vessels. Her severe cyanosis did not improve after two central shunt procedures that were performed on the basis of only echocardiographic and electron-beam CT findings. She died of cardiac arrest at the age of 5 months.

For the purpose of correlating the findings at angiography and electron-beam CT, only studies that were performed within 6 months of the electron-beam CT without intercurrent surgery were considered. Thus, angiocardiographic correlation was possible for 34 of the 48 electron-beam CT studies. The majority (21 angiograms) of these 34 angiograms were obtained postoperatively (for shunt surveillance of suspected thrombosis or as guidance for endovascular interventions [coil embolization of collateral vessels, balloon angioplasty of stenoses]), whereas the remaining 13 angiograms were obtained in anticipation of surgery (for measurements of collateral vessel pressure and for coil embolization of collateral vessels that were considered too small for incorporation into unifocalization conduits and that were therefore deemed surgically unimportant).

Angiographic technique was tailored to the clinical indication for the examination. Therefore, it varied considerably throughout the study. It generally included aortography in the descending aorta, ascending aorta, or aortic arch, with inflation of an intraaortic balloon to demonstrate the origin and number of aortopulmonary collateral vessels and with selective injections of contrast material into the collateral vessels and pulmonary arteries. If necessary, this was supplemented with injections into the bilateral subclavian arteries and selective injections into Blalock-Taussig shunts. Unifocalization conduits were injected selectively in all postoperative studies; these were followed with more selective injections as needed. If these measures failed to fully demonstrate the pulmonary arterial anatomy, pulmonary venous wedge injections were performed to opacify the hilar arteries in retrograde fashion. All angiograms were obtained with biplanar fluoroscopy and with cine imaging in the posteroanterior and left lateral projections.

Image Analysis and Reference Standard
All electron-beam CT images (transverse sections and three-dimensional renditions) were reviewed independently by two radiologists (S.J.W., J.H.) who had no knowledge of the angiographic findings at the time of review. Following this, all cineangiograms were retrospectively analyzed by one of the authors (S.J.W.), and findings were correlated with the angiographic reports that were rendered, in majority, by the same pediatric cardiologist (A.G.). All disagreements between the two reviewers of CT images and all discrepancies between the retrospective angiographic review and the official reports were resolved by consensus. Finally, the surgical reports in all patients were retrieved from the hospital information system, and the findings presented in these were used as the reference standard.

On all electron-beam CT images and angiograms, we visually rated the presence of the following: (a) main pulmonary artery, (b) left branch pulmonary artery, (c) right branch pulmonary artery, (d) pulmonary artery confluence, (e) branch pulmonary artery stenosis, and (f) stenosis of the origin of aortopulmonary collateral vessels. Stenosis was defined as a focal reduction of more than 50% of the luminal diameter, as measured with calipers in the transverse plane (at CT) or on the most appropriate (posteroanterior or lateral) angiographic image. In addition, we rated the number and point of origin of all visualized aortopulmonary collateral vessels, and we rated the patency of surgical conduits and shunts.

On preoperative images, the native vessels (pulmonary arteries and/or aortopulmonary collateral vessels) were assessed for the presence of stenosis, whereas on postoperative images, all visible anastomoses between these vessels and the surgical conduits were specifically examined for stenosis. Thrombosis was identified at electron-beam CT as a lack of opacification of, or as a partial filling defect within, the conduit or vessel lumen; in addition, thrombosis was identified at angiography by means of indirect signs, such as inability to catheterize and/or opacify a known shunt and the presence of a reactive collateral vascular network to bypass the thrombosed area.

Estimates of Radiation Dose
Because of the unique qualities of the beam and because of the receptor geometry of the electron-beam CT scanner, the radiation exposure at electron-beam CT is nonuniform and is asymmetrically distributed, with most of the skin exposure deposited at the posterior surface when the patient is in the supine position (16). To estimate the radiation dose that results from using a 3-mm section thickness (which is the most dose-effective, since the optimal combination of entry-beam and exit-beam collimation is being used) and the aforementioned exposure parameters, we performed measurements on a 16-cm Perspex phantom (CT head phantom; Radcal, Monrovia, Calif) with an ionization chamber to calculate the CT dose index. To provide a more conservative estimate, we applied a correction for the underestimation of calculation of the CT dose index in the ionization chamber, as suggested by McCollough et al (16), by multiplying the CT dose index estimates by 1.25. For the overlapping (3-mm thickness, 2-mm spacing) imaging profile that we used in younger children, we increased the estimates of radiation dose, which were calculated for the contiguous 3-mm scanning profile, by another 50%.

For diagnostic angiocardiography, we estimated the mean patient dose (ignoring the contributions for the performance of interventional procedures) as follows. From the procedural records that were kept, we calculated the mean total fluoroscopy time, the relative contributions of posteroanterior and left lateral fluoroscopy time to total fluoroscopy time, the mean number of angiographic cine runs per examination, and the mean run time per examination. On the basis of exposure measurements in a phantom of sufficient thickness to simulate the technical exposure factors expected in the average-sized child in our study population, we separately estimated the contributions of fluoroscopy and cineangiocardiography to the exposure to the posterior and right-lateral portions of the chest. Exposures were converted to skin dose by using the exposure-to-dose conversion factor f for muscle/water, 0.94 rad/R (36.4 Gy/C/kg).

Statistical Analysis
A qualitative assessment was performed for visualization of the pulmonary arteries, which was defined as the ability of the modality under study to depict caliber. For each patient, the scores for all six visualization criteria were tabulated for all electron-beam CT and angiographic studies, combined, that were performed in that patient. By using surgical findings as the reference standard, visualization scores were characterized as true-positive, false-positive, true-negative, or false-negative. By using these, test parameters (sensitivity, specificity, positive and negative predictive values, and accuracy) were calculated for each of the six rated criteria. Finally, all visualization scores were summed for all six criteria combined to generate overall summary test parameters for both imaging modalities. Differences between groups were tested at a confidence level of P less than .05.

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Apart from occasional nausea and vomiting after the injection of contrast material (which resolved spontaneously with simple measures and no specific treatment), there were no complications with our electron-beam CT series. There was no extravasation of contrast material at the injection site, and there were no marked systemic reactions to the contrast material. Angiographic complications were not specifically rated during this study. All transverse electron-beam CT images were of sufficient diagnostic quality to address the study questions. We obtained three-dimensional renditions for 37 of the 48 CT images, whereas on the remaining 11 images, patient motion during image acquisition precluded meaningful three-dimensional reconstructions. These reconstructions did not provide added diagnostic benefit over the transverse images.

Pulmonary Arteries
Both branch pulmonary arteries were absent at surgery in 10 of the 23 patients. In one patient, only a left branch pulmonary artery was identified in the hilum, and in another patient, only a right branch pulmonary artery was identified. In a third patient, both branch pulmonary arteries were present, but they were not confluent. One patient had a stenosis in the right branch pulmonary artery, which persisted after placement of a stent. The main pulmonary artery was absent in 13 of the 23 patients. The results of our comparative analysis of electron-beam CT and angiography versus surgery are summarized in the Table. Figures 3 and 4 illustrate the value of electron-beam CT compared with angiography in preoperative assessment, whereas Figures 1, 5, and 6 show postoperative findings.

View larger version (155K): [in this window] [in a new window]   Figure 3a. Cross-sectional electron-beam CT image obtained at preoperative evaluation in a 2-year-7-month-old boy with pulmonary atresia demonstrates the advantage of electron-beam CT over projectional angiography. (a) Posteroanterior descending aortogram was interpreted as showing significant stenosis in the left proximal portion of the branch pulmonary artery (arrows). Both pulmonary arteries are confluent, and there are two tortuous right aortopulmonary collateral vessels (C) with stenotic origins that arise from the descending aorta (D). (b) Coronal and (c, d) transverse thick-slab maximum intensity projection electron-beam CT images show two tortuous right-sided aortopulmonary collateral vessels (C) in the posterior mediastinum and show a left aortopulmonary collateral vessel (C) that originates from the undersurface of transverse aortic arch (T in b). Aortopulmonary collateral vessels and branch pulmonary arteries (R, L in c) combine to form both pulmonary hila, with overlapping vascular territories. The branch pulmonary arteries are confluent, but no left branch pulmonary artery stenosis is present in c; absence of stenosis was confirmed at surgery. The appearance of the stenosis on the angiogram was caused by an overcrossing left superior pulmonary vein (V in d).

View larger version (151K): [in this window] [in a new window]   Figure 3b. Cross-sectional electron-beam CT image obtained at preoperative evaluation in a 2-year-7-month-old boy with pulmonary atresia demonstrates the advantage of electron-beam CT over projectional angiography. (a) Posteroanterior descending aortogram was interpreted as showing significant stenosis in the left proximal portion of the branch pulmonary artery (arrows). Both pulmonary arteries are confluent, and there are two tortuous right aortopulmonary collateral vessels (C) with stenotic origins that arise from the descending aorta (D). (b) Coronal and (c, d) transverse thick-slab maximum intensity projection electron-beam CT images show two tortuous right-sided aortopulmonary collateral vessels (C) in the posterior mediastinum and show a left aortopulmonary collateral vessel (C) that originates from the undersurface of transverse aortic arch (T in b). Aortopulmonary collateral vessels and branch pulmonary arteries (R, L in c) combine to form both pulmonary hila, with overlapping vascular territories. The branch pulmonary arteries are confluent, but no left branch pulmonary artery stenosis is present in c; absence of stenosis was confirmed at surgery. The appearance of the stenosis on the angiogram was caused by an overcrossing left superior pulmonary vein (V in d).

View larger version (156K): [in this window] [in a new window]   Figure 3c. Cross-sectional electron-beam CT image obtained at preoperative evaluation in a 2-year-7-month-old boy with pulmonary atresia demonstrates the advantage of electron-beam CT over projectional angiography. (a) Posteroanterior descending aortogram was interpreted as showing significant stenosis in the left proximal portion of the branch pulmonary artery (arrows). Both pulmonary arteries are confluent, and there are two tortuous right aortopulmonary collateral vessels (C) with stenotic origins that arise from the descending aorta (D). (b) Coronal and (c, d) transverse thick-slab maximum intensity projection electron-beam CT images show two tortuous right-sided aortopulmonary collateral vessels (C) in the posterior mediastinum and show a left aortopulmonary collateral vessel (C) that originates from the undersurface of transverse aortic arch (T in b). Aortopulmonary collateral vessels and branch pulmonary arteries (R, L in c) combine to form both pulmonary hila, with overlapping vascular territories. The branch pulmonary arteries are confluent, but no left branch pulmonary artery stenosis is present in c; absence of stenosis was confirmed at surgery. The appearance of the stenosis on the angiogram was caused by an overcrossing left superior pulmonary vein (V in d).

View larger version (148K): [in this window] [in a new window]   Figure 3d. Cross-sectional electron-beam CT image obtained at preoperative evaluation in a 2-year-7-month-old boy with pulmonary atresia demonstrates the advantage of electron-beam CT over projectional angiography. (a) Posteroanterior descending aortogram was interpreted as showing significant stenosis in the left proximal portion of the branch pulmonary artery (arrows). Both pulmonary arteries are confluent, and there are two tortuous right aortopulmonary collateral vessels (C) with stenotic origins that arise from the descending aorta (D). (b) Coronal and (c, d) transverse thick-slab maximum intensity projection electron-beam CT images show two tortuous right-sided aortopulmonary collateral vessels (C) in the posterior mediastinum and show a left aortopulmonary collateral vessel (C) that originates from the undersurface of transverse aortic arch (T in b). Aortopulmonary collateral vessels and branch pulmonary arteries (R, L in c) combine to form both pulmonary hila, with overlapping vascular territories. The branch pulmonary arteries are confluent, but no left branch pulmonary artery stenosis is present in c; absence of stenosis was confirmed at surgery. The appearance of the stenosis on the angiogram was caused by an overcrossing left superior pulmonary vein (V in d).

View larger version (121K): [in this window] [in a new window]   Figure 4a. Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels show a right-sided aortic arch and hypoplastic confluent branch pulmonary arteries in an 8-year-old boy who was evaluated for a possible unifocalization procedure. (a) Image depicts a single large nonstenotic aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch and that supplies the entire right lung. (b) In addition to several aortopulmonary collateral vessels (C) with stenotic origins that branch off the left anterior aspect of the right-sided descending aorta (D), there are hypoplastic confluent pulmonary arteries (arrowheads) at this level. Angiographic images (not shown) demonstrated only the aortopulmonary collateral vessels and failed to depict the pulmonary arteries. Images obtained at selective catheterization of the aortopulmonary collateral vessels demonstrated systemic pressures in the right lung and half-systemic pressures in the left. The surgical approach, which was based on these findings, consisted of a left unifocalization procedure (with incorporation of the hypoplastic left branch pulmonary artery and the left-sided aortopulmonary collateral vessels) and banding of the right-sided collateral vessel to down-regulate the pressures in the right pulmonary vascular bed; this was performed in anticipation of future right-sided surgical repair.

View larger version (115K): [in this window] [in a new window]   Figure 4b. Transverse thick-slab maximum intensity projection electron-beam CT images obtained at consecutive levels show a right-sided aortic arch and hypoplastic confluent branch pulmonary arteries in an 8-year-old boy who was evaluated for a possible unifocalization procedure. (a) Image depicts a single large nonstenotic aortopulmonary collateral vessel (C) that originates from the undersurface of the aortic arch and that supplies the entire right lung. (b) In addition to several aortopulmonary collateral vessels (C) with stenotic origins that branch off the left anterior aspect of the right-sided descending aorta (D), there are hypoplastic confluent pulmonary arteries (arrowheads) at this level. Angiographic images (not shown) demonstrated only the aortopulmonary collateral vessels and failed to depict the pulmonary arteries. Images obtained at selective catheterization of the aortopulmonary collateral vessels demonstrated systemic pressures in the right lung and half-systemic pressures in the left. The surgical approach, which was based on these findings, consisted of a left unifocalization procedure (with incorporation of the hypoplastic left branch pulmonary artery and the left-sided aortopulmonary collateral vessels) and banding of the right-sided collateral vessel to down-regulate the pressures in the right pulmonary vascular bed; this was performed in anticipation of future right-sided surgical repair.

View larger version (149K): [in this window] [in a new window]   Figure 5a. Images depict thrombosis of the left unifocalization conduit in an 11-year-old girl who underwent bilateral unifocalization procedures and who presented with recent worsening in cyanosis after a short interruption in anticoagulation treatment. (a) Posteroanterior left subclavian arteriogram fails to depict opacification in the known modified Blalock-Taussig shunt and unifocalization conduit on the left. A few small intrapulmonary branches (arrows) fill via a fine network of collateral vessels. (b) Transverse thick-slab maximum intensity projection electron-beam CT image shows the thrombotic occlusion of the left unifocalization tube more directly, with patent right unifocalization structures. Clot formation (CL) has not propagated into the intrapulmonary branches, which enhance normally and which could be used at subsequent revision surgery of the left unifocalization.

View larger version (120K): [in this window] [in a new window]   Figure 5b. Images depict thrombosis of the left unifocalization conduit in an 11-year-old girl who underwent bilateral unifocalization procedures and who presented with recent worsening in cyanosis after a short interruption in anticoagulation treatment. (a) Posteroanterior left subclavian arteriogram fails to depict opacification in the known modified Blalock-Taussig shunt and unifocalization conduit on the left. A few small intrapulmonary branches (arrows) fill via a fine network of collateral vessels. (b) Transverse thick-slab maximum intensity projection electron-beam CT image shows the thrombotic occlusion of the left unifocalization tube more directly, with patent right unifocalization structures. Clot formation (CL) has not propagated into the intrapulmonary branches, which enhance normally and which could be used at subsequent revision surgery of the left unifocalization.

View larger version (155K): [in this window] [in a new window]   Figure 6a. Images depict a surgically important coronary anomaly, which was incidentally discovered at electron-beam CT prior to complete surgical repair, in 10-year-old boy. (a) Volume-rendition electron-beam CT image, which was rotated on the workstation to appear as if the patient is prone and which is viewed from above, shows that the right branch pulmonary artery (R), which is incorporated into a unifocalization conduit (U, P) together with several ligated aortopulmonary collateral vessels. This unifocalization complex was perfused via a central shunt (not shown) from the ascending aorta (A). The branch pulmonary arteries (R, L) are confluent and well developed. In addition, the left pulmonary artery is perfused via the modified Blalock-Taussig shunt on the left (not shown). Note anomalous left anterior descending coronary artery (arrowheads), which originates from the right coronary artery and which crosses the right ventricle. The left coronary artery (not shown) supplied only the circumflex artery. (b) Posteroanterior selective angiogram of the right unifocalization complex obtained via the central shunt shows that the confluent branch pulmonary arteries (R, L) are also opacified. The coronary anomaly was not demonstrated on an ascending aortogram (not shown) because of suboptimal technique. Final repair surgery was complicated by the need to construct a bypass for the anomalous left anterior descending coronary artery prior to constructing a right ventricular outflow tract conduit.

View larger version (135K): [in this window] [in a new window]   Figure 6b. Images depict a surgically important coronary anomaly, which was incidentally discovered at electron-beam CT prior to complete surgical repair, in 10-year-old boy. (a) Volume-rendition electron-beam CT image, which was rotated on the workstation to appear as if the patient is prone and which is viewed from above, shows that the right branch pulmonary artery (R), which is incorporated into a unifocalization conduit (U, P) together with several ligated aortopulmonary collateral vessels. This unifocalization complex was perfused via a central shunt (not shown) from the ascending aorta (A). The branch pulmonary arteries (R, L) are confluent and well developed. In addition, the left pulmonary artery is perfused via the modified Blalock-Taussig shunt on the left (not shown). Note anomalous left anterior descending coronary artery (arrowheads), which originates from the right coronary artery and which crosses the right ventricle. The left coronary artery (not shown) supplied only the circumflex artery. (b) Posteroanterior selective angiogram of the right unifocalization complex obtained via the central shunt shows that the confluent branch pulmonary arteries (R, L) are also opacified. The coronary anomaly was not demonstrated on an ascending aortogram (not shown) because of suboptimal technique. Final repair surgery was complicated by the need to construct a bypass for the anomalous left anterior descending coronary artery prior to constructing a right ventricular outflow tract conduit.

Aortopulmonary Collateral Vessels
At surgery, a total of 74 aortopulmonary collateral vessels were identified and incorporated in unifocalization conduits in the 23 patients, with a mean of 3.2 collateral vessels per patient. Of these 74 aortopulmonary collateral vessels that were deemed surgically important, 39 supplied the right lung, 31 supplied the left lung, and four supplied both lungs. There was no significant difference (P < .05) in the number of collateral vessels per patient and between patients with and those without true branch pulmonary arteries.

At electron-beam CT, all of the 74 surgically important collateral vessels were identified, and an additional nine collateral vessels that were not identified during surgery were seen. Angiography revealed all but two of the surgically important collateral vessels and revealed an additional four collateral vessels that were not seen during surgery and that were not, therefore, incorporated into the unifocalization conduits. These four collateral vessels, which were also depicted at electron-beam CT, all required coil embolization at subsequent angiography. All other discrepancies between the number of aortopulmonary collateral vessels found at electron-beam CT, angiography, and surgery could be explained by the fact that the side on which the aortopulmonary collateral vessels in question were located was never specifically examined at angiography and surgery, whereas both sides were always examined at electron-beam CT.

Significant stenoses at the origins of the aortopulmonary collateral vessels that originated from the aorta were present at surgery in seven patients; these were detected at electron-beam CT in four patients and at angiography in six (Figs 3, 4). In six postoperative patients, a stenotic anastomosis between an aortopulmonary collateral vessel and a unifocalization conduit was identified at angiography; three of these anastomotic stenoses were depicted at electron-beam CT. Peripheral aortopulmonary collateral vascular stenoses were not specifically rated in this study, but our subjective impression was that these were more prevalent than were the stenoses at the origins, and that they were better depicted at angiography than at electron-beam CT.

The vast majority (70 collateral vessels) of the surgically important collateral vessels originated from the descending aorta or from the undersurface of the aortic arch, and only four originated directly from the subclavian arteries. Angiographically demonstrated collateral vessels that originated from the chest wall (supplied by internal mammary and intercostal arteries) were always small and tortuous and could not be used for incorporation into unifocalization structures; they were, therefore, considered surgically unimportant. However, such collateral vessels may be clinically important since they may bleed into the lung after surgery, and they frequently need to be embolized. Although electron-beam CT was not systematically studied for the imaging of chest wall collateral vessels, it was our general impression that electron-beam CT was insensitive to depict these.

Shunts and Thromboses
Modified Blalock-Taussig or central shunts were placed into 19 of the 23 patients, and these were depicted in all patients at electron-beam CT and in 14 patients at angiography (shunts were not seen in five patients because the angiographic study was focused to the side opposite the shunt). Shunt and unifocalization conduit thromboses were seen in four patients (Fig 5), and all four required repeat surgery. In all patients, electron-beam CT and angiography demonstrated that the thrombosis did not propagate beyond the collateral vascular anastomoses, so the involved collateral vessels could be used in the subsequent repair. Transient partial thrombosis with pulmonary embolus and infarction was seen at electron-beam CT in a unifocalization conduit immediately after percutaneous balloon angioplasty and placement of a stent for a collateral vascular stenosis; this was documented to resolve completely at a follow-up electron-beam CT study that was obtained 2 months later.

Thrombosis in a previously placed modified Blalock-Taussig shunt was seen at electron-beam CT in nine patients. Angiography demonstrated only two of these shunt thromboses with indirect signs (lack of opacification of the known shunt during the injection of contrast material into the appropriate subclavian artery) but did not demonstrate four of these thromboses, as contrast material was not injected into the appropriate subclavian artery. The remaining three patients with thrombosis in the modified Blalock-Taussig shunt, as identified at electron-beam CT, had previously undergone a complete surgical repair. This included ligation of the (now superfluous) shunts, which were left in place and in which, consequently, thrombosis developed. This finding was expected and was thought to have no clinical importance; no correlative angiograms were available in these patients.

Postoperative Complications at Electron-Beam CT
Postoperative seromas that compressed shunts and unifocalization structures were seen in two patients; both required drainage for decompression, which was surgical in one patient and percutaneous CT-guided in the other. In a patient with postoperative subacute bacterial endocarditis, electron-beam CT showed a mycotic pseudoaneurysm that was adjacent to one of the sinuses of Valsalva (aortic sinus). In another patient, a left-sided pleural hematoma that necessitated surgical evacuation was found; it resulted from bleeding chest wall collateral vessels. Compression of the right bronchial tree by a transverse vascular conduit made of the ePTFE membrane, which necessitated surgical repair with replacement of part of the conduit, was noted in one patient.

Incidental Findings at Electron-Beam CT That Influenced Surgical Management
In four patients, an aberrant left anterior descending coronary artery was found during our electron-beam CT analysis; it originated from the right main coronary artery and crossed the region of the right ventricular outflow tract (Fig 6). In all four patients, this finding was not prospectively reported at angiography either because no ascending aortogram had been obtained or because the quality of the ascending aortogram was judged, retrospectively, as insufficient to depict this anomaly. Surgery involving complete repair with reconstruction of the main pulmonary artery was substantially prolonged in two of these patients because of the unanticipated need to create a coronary bypass for this anomalous left anterior descending coronary artery prior to reconstructing the right ventricular outflow. Conduit calcification was seen at electron-beam CT but not at angiography in two patients; both required revision with partial replacement of their conduits.

Comparison of Estimates of Radiation Dose
From the measurements of the CT dose index in the phantom, we calculated absorbed doses in the posterior and lateral portions for the 3-mm contiguous imaging profile at 0.83 rad (8.3 mGy) and at 0.59 rad (5.9 mGy), respectively. After correcting for the underestimation of the CT dose index by using the chamber measurements, these rose to 1.03 rad (10.3 mGy) and 0.74 rad (7.4 mGy), respectively. The corrected doses for the 3-mm overlapping imaging profile were calculated to be 1.55 and 1.11 rad (15.5 and 11.1 mGy) at the posterior and lateral portions, respectively.

In our angiographic series, a mean fluoroscopy time (± SD) of 45 minutes ± 21 (80% in the posteroanterior projection and 20% in the lateral projection) was used, and a mean of 5 angiographic runs ± 2 (with a mean run time of 5 seconds at a rate of 30 frames/sec) were performed. Skin exposure at fluoroscopy was estimated from phantom measurements as 1.7 R/min (4.4 x 10^-4 C/kg/min) in the posteroanterior projection, 2.0 R/min (5.2 x 10^-4 C/kg/min) in the left lateral projection, and 12.0 mR (3.1 x 10^-6 C/kg) per frame in each direction for biplanar cineangiocardiography. This resulted in estimates of the mean skin entry exposure of 61 R ± 29 (15.7 x 10^-3 C/kg ± 7.5 x 10^-3) at the posterior portion and 18 R ± 8 (4.6 x 10^-3 C/kg ± 2.1 x 10^-3) at the right-lateral portion at fluoroscopy and estimates of 9 R ± 4 (2.3 x 10^-3 C/kg ± 1.0 x 10^-3) at the posterior and right-lateral portions at cineangiography. The combined skin radiation dose at angiography was, therefore, estimated to be 0.66 Gy ± 0.27 (66 rad ± 27) and 0.25 Gy ± 0.08 Gy (25 rad ± 8), at the posterior and right-lateral portions, respectively.

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Findings from our previous work with helical CT for congenital heart disease, in comparison with echocardiography and angiocardiography, showed that CT is a reliable exploratory test for the noninvasive assessment of global cardiovascular anatomy. CT compensates for the weaknesses of echocardiography, which is best suited for the assessment of the intracardiac anatomy (15). The presence of branch pulmonary arteries and peripheral stenoses and the patency of shunts are unreliably depicted at echocardiography, since acoustic window settings are often suboptimal to show these features and since the interposition of aerated lung tissue precludes the depiction of most aortopulmonary collateral vessels, unifocalization conduits, and surgical anastomoses.

In other anatomic regions, cross-sectional CT angiography aided with three-dimensional reconstruction has proved helpful in the evaluation of complex overlapping vascular structures that would superimpose on projectional radiographs (17–21). Angiography relies on the direct catheterization of the vascular structure to be opacified; this poses a substantial limitation on its use in the comprehensive evaluation of complex cardiovascular anomalies. At CT, all cardiac chambers and pulmonary vessels are opacified with one injection of contrast material. Difficult-to-catheterize vessels, such as hypoplastic branch pulmonary arteries (the angiographically "absent" pulmonary artery [6]) and pulmonary veins, which can be indirectly depicted only at angiography by using special invasive techniques (pulmonary venous wedge injections), are routinely depicted at CT. In our study, electron-beam CT frequently showed additional findings, such as coronary anomalies, postoperative seromas, pleural fluid collections, pulmonary parenchymal disease, and airway compression, that were important in the perioperative assessment and that were not routinely or well depicted at angiography.

Findings from the current study indicate that for the evaluation of hypoplastic central pulmonary arteries, the accuracy of electron-beam CT surpassed that of angiography. Electron-beam CT showed more collateral vessels than angiography, but these collateral vessels tended to be surgically unimportant. Angiography remains the most accurate test for the diagnosis of origin or peripheral stenosis in the aortopulmonary collateral vessels, aided by the feasibility to perform pressure measurements during selective catheterization of these collateral vessels.

A drawback of electron-beam CT compared with angiography is the fact that it is a relatively static imaging modality best suited to assess morphology. A limited functional cardiac assessment at electron-beam CT that is based on morphologic parameters is possible by using fast (50-msec) cine and flow multisection acquisitions. In this manner, reliable estimates of ventricular volumes at various points in the cardiac cycle can be obtained, which allow for the calculation of the ejection fraction, cardiac index, and myocardial mass and which allow for the study of regional motion in the ventricular wall (22,23). Real-time study of the timed flow of a bolus of contrast material through the heart and great vessels allows for the calculation of shunt fractions (24). In patients with pulmonary atresia, there is frequently a need for exact pressure measurements in all cardiac chambers, in the true pulmonary arterial system, in the aorta, and in each aortopulmonary collateral vessel; for selective injection studies to demonstrate communications and degree of overlap between various vascular territories; for oxygen saturation measurements in various locations; and for the calculation of pulmonary versus systemic blood flow by using thermodilution methods. For these hemodynamic and functional assessments (and for the performance of endovascular interventions), there currently is no substitute for invasive cardiac catheterization.

An advantage of CT over conventional angiography is the ability<