Cardiac Imaging
Comparing Thin-Section and Thick-Section CT of Pericardial Sinuses and Recesses
OBJECTIVE. The aim of this study was to assess the prevalence and appearance of the pericardial sinuses and recesses on thin-section (2.5- or 3-mm) CT scans compared with thick-section (5- or 7-mm) CT scans.
MATERIALS AND METHODS. Nine hundred forty-one consecutive contrast-enhanced chest CT scans were retrospectively evaluated. Three hundred sixty-five patients underwent thin-section CT, and 576 patients underwent thick-section CT. The prevalence and appearance of every pericardial recess were determined.
RESULTS. Large recesses such as the superior aortic recess were depicted in 12.5–30.4% of patients using thick-section CT, whereas smaller recesses such as the postcaval recess were depicted in fewer than 5% of patients. With thin-section CT, the depiction rates increased significantly compared with thick-section CT (p < 0.01). Large recesses were depicted in 28.7–44.7% of patients, and smaller recesses were recognized in 10.8–19.8% of patients. Generally, most recesses were linear if they were small and became band-shaped as the fluid increased. However, the recesses were often visualized as crescent, triangle, spindle, ovoid, hemisphere, or irregular shapes.
CONCLUSION. Pericardial sinuses and recesses are more frequently and better depicted on thin-section CT scans. Knowledge of their locations and shapes is helpful for distinguishing pericardial fluid from abnormal findings such as lymphadenopathy and cystic lesions.
Within the pericardial cavity, small pockets are formed by the reflection of the pericardium, in which fluid can collect. These pockets are called pericardial sinuses or recesses, which include the transverse sinus (aortic recess and pulmonic recess), oblique sinus, postcaval recess, and pulmonary venous recess [1]. On CT, these sinuses and recesses can be recognized as areas of water attenuation around the great vessels. Some of the larger sinuses and recesses are relatively easy to recognize and unlikely to present a diagnostic problem. However, others, especially the less well-known ones, are difficult to differentiate from mediastinal or hilar abnormal lesions because they can simulate lymphadenopathy or cystic lesions. An understanding of their normal location and appearance, coupled with a knowledge of common pitfalls, prevents the misinterpretation of normal pericardial recesses as abnormal lesions. Previous reports described the CT appearance of some of these sinuses and recesses, and superior aortic recess and pulmonic recess have become well known [2–7]. However, many of these researchers used thicker collimation CT and described only one or a few recesses. Only three reports describe all pericardial sinuses and recesses in a small number of cases [8–10]. To our knowledge, ours is the first report that describes all recesses in a large number of cases using thinsection CT, including multidetector CT. We describe the prevalence and appearance of all pericardial sinuses and recesses and also determine whether there is a significant difference between the visualization of these sinuses and recesses on thin-section CT and on thick-section CT.
We retrospectively reviewed 365 consecutive thin-section CT scans and 576 thick-section CT scans of the chest obtained in our institution from January 2001 to July 2001 in 895 patients. Fortysix patients underwent both thin-section and thick-section CT. Each procedure was separated by an interval of 1–4 weeks. All images were acquired after the administration of IV contrast material. This study group consisted of 464 females and 431 males who were 17–97 years old (mean, 55.1 years). Thin-section CT scans were obtained to evaluate suspected pulmonary embolism, whereas thick-section CT scans were obtained for a routine check or for evaluation of metastatic disease, known or suspected lung cancer, pleural effusion, pulmonary infiltrates, or nodules. Patients who had pericardial effusion were excluded from this study.
CT scans were obtained using a multidetector CT scanner (LightSpeed QX/i, General Electric Medical Systems, Milwaukee, WI) and helical CT scanners (HiSpeed Advantage and HiSpeed CT/i, General Electric Medical System). Thin-section CT was performed on a LightSpeed QX/i scanner with 2.5-mm collimation and on HiSpeed Advantage and CT/i scanners with 3-mm collimation. Standard protocol involved 120–140 kV, 260–320 mA, and 29- to 42-cm field of view. With the QX/i scanner, we used 0.8 sec/rotation and a pitch of 6, and with a HiSpeed Advantage or CT/i scanner, we used 1.0 sec/rotation. Contiguous axial sections were obtained from the lung apices to the diaphragm after 150 mL of nonionic contrast material was injected into an antecubital vein at a rate of 3.5 mL/sec. The scanning delay range was 20–25 sec. Thick-section CT was performed using HiSpeed Advantage, and CT/i scanners with 5- or 7-mm collimation also scanned the whole lung after a bolus injection of 100–150 mL of contrast material at a rate of 2.0 mL/sec. Standard protocol involved 120 kV, 240–300 mA, 28- to 42-cm field of view, and 1.0 sec/rotation.
We defined the pericardial cavity according to previous researchers [1, 2] (Fig. 1). The transverse sinus was defined as the sinus behind the ascending aorta and pulmonary trunk, and it was divided into the following recesses: superior aortic recesses, inferior aortic recess, and left and right pulmonic recesses. The superior aortic recess can be divided into two parts: anterior and posterior portions. The anterior superior aortic recess lies in front of the ascending aorta, and the posterior superior aortic recess lies dorsal to the ascending aorta. There is a connection between them. The inferior aortic recess lies between the aortic root and the right atrium. It connects upward to the superior aortic recess. The left pulmonic recess lies beneath the left pulmonary artery extending posterolaterally to the proximal right pulmonary artery. The left pulmonic recess is continuous with the superior aortic recess anteriorly and with the transverse sinus laterally. The right pulmonic recess is the space surrounded mainly by the following three structures: the left atrium, the aortic root, and the proximal right pulmonary artery. The left atrium forms the floor and inferior boundary, and the aortic root and right pulmonary artery form the posterior and superior boundaries of this recess, respectively. The right pulmonic recess continues on the posterior superior aortic recess. The oblique sinus is the sinus behind the left atrium including posterior pericardial recess. Pericardial reflection between the right and left superior pulmonary veins transversely separates the oblique sinus and the transverse sinus. The oblique sinus is contiguous with the subcarinal region, which is called the posterior pericardial recess. The pericardial cavity proper includes the following three recesses: the left and right pulmonary venous recesses and the postcaval recess. The postcaval recess is posterior and right lateral to the superior vena cava and between the inferior vena cava and the coronary sinus. The left and right pulmonary venous recesses exist between the superior and inferior pulmonary veins on each side.
![]() View larger version (34K) | Fig. 1. —Drawing shows posterior aspect of heart. Transverse sinus, mainly behind ascending aorta and pulmonary artery, is divided into anterior (1) and posterior (2) portions of superior aortic recess, inferior aortic recess (3), and left (4) and right (5) pulmonic recesses. Oblique sinus (6) lies posterior to left atrium. Pericardial sinus proper includes postcaval recess (7) and left (8) and right (9) pulmonary venous recesses. Note connections (arrows) between sinuses and recesses. Ao = aorta, SVC = superior vena cava, lPA = left pulmonary artery, rPA = right pulmonary artery, lPVs = left pulmonary veins, rPVs = right pulmonary veins, LA = left atrium, IVC = inferior vena cava. |
CT scans were analyzed by two experienced radiologists who were unaware of the clinical information and CT technique. When there was a difference of opinion about CT findings, the final judgment was reached by consensus. To differentiate fluid in a pericardial sinus and recess from disease, we defined fluid in the pericardial cavity, using the criteria presented in prior reports; well-marginated homogenous near-water-attenuation structures (≈ –10 to 30 H) without a wall or rim of material with higher attenuation in an expected location of a sinus or recess. We measured attenuation values to confirm that they were consistent with those of fluid. We recorded the volume of every pericardial fluid pocket (small, moderate, or large). On thin-section CT, we also recorded the shape (point, line, band, crescent, spindle, hemisphere, ovoid, round, triangle, rhomboid, or irregular) (Fig. 2). The recesses that were difficult to distinguish from artifacts, occupied by mass lesions, or adjacent to pleural effusion and consolidation, were excluded from the analysis. The frequency of the fluid in every recess was statistically compared between each CT technique using the Mann-Whitney U test. The volume of the fluid was scored as follows: 0, no fluid; 1, small volume; 2, moderate volume; 3, large volume. Differences were considered statistically significant when p values were less than 0.01.
![]() View larger version (5K) | Fig. 2. —Drawing shows shapes of pericardial sinuses and recesses: line (1), band (2), spindle (3), crescent (4), point (5), hemisphere (6), ovoid (7), round (8), triangle (9), rhomboid (10), and irregular (11). |
The frequency and the volume of every pericardial sinus and recess detected on thin- and thick-section CT are summarized in Table 1. Large recesses such as the anterior and posterior superior aortic recesses, left and right pulmonic recesses, and oblique sinus, which were easily identified using thick-section CT (7.7–30.4% prevalence), were depicted at 28.7–44.7% prevalence using thin-section CT. The small recesses such as the left and right pulmonary venous recesses, inferior aortic recess, and postcaval recess, which were hardly recognized using thick-section CT (2.5–4.7% prevalence), were depicted at 10.8–19.8% prevalence using thin-section CT. The prevalence significantly increased for every sinus and recess depicted on thin-section CT scans compared with those on thick-section CT scans (p < 0.01) (Fig. 3A, 3B).
![]() View larger version (171K) | Fig. 3A. —50-year-old woman with dyspnea. CT scan obtained with 2.5-mm collimation shows irregular-shaped right pulmonic recess (arrow) and spindle-shaped oblique sinus (arrowheads). |
![]() View larger version (157K) | Fig. 3B. —50-year-old woman with dyspnea. CT scan obtained with 5-mm collimation does not clearly show either of two sinuses seen on A. |
The appearances of pericardial sinuses and recesses are summarized in Table 2. A small effusion in the anterior and posterior superior aortic recesses was seen as triangular to crescentic and crescentic to hemispheric, respectively (Fig. 4). As the volume of these recesses increased, their shapes became crescentic to band and ovoid, respectively. The shape of the inferior aortic recess was observed as linear to band when small in volume and became a thick band as fluid increased (Fig. 5). The left pulmonic recess was linear to crescentic when small in volume (Fig. 6) and became a thick band to spindle shape as it increased. The shape of the right pulmonic recess varied. It was generally observed as rhomboid to a more complex irregular configuration regardless of its volume and sometimes had a band shape (Fig. 7). The shape of oblique sinus was observed as linear to band when small in volume and became a thick band as it increased (Fig. 3A). The postcaval recess appeared triangular when small in volume and became band-shaped when large in volume (Fig. 7). The right pulmonary venous recess often appeared as a small hemisphere surrounding the pulmonary vein (Fig. 8) and became round to band-shaped when large in volume. The left pulmonary venous recess was often recognized as linear to band-shaped when small in volume and became a thick band shape when large in volume (Fig. 9).
![]() View larger version (126K) | Fig. 4. —63-year-old woman with metastatic lung cancer. CT scan obtained with 3-mm collimation shows moderate volume of fluid (arrow) in triangular anterior portion of superior aortic recess. Note moderate volume of fluid (arrowhead) in hemispheric posterior portion of superior aortic recess. |
![]() View larger version (189K) | Fig. 5. —82-year-old woman with history of aortic aneurysm. CT scan obtained with 2.5-mm collimation shows small volume of fluid in linear inferior superior aortic recess (arrow) between ascending aorta and left atrium. |
![]() View larger version (162K) | Fig. 6. —48-year-old man with bilateral pleural effusions. CT scan obtained with 2.5-mm collimation shows linear left pulmonic recess (arrow) between right pulmonary artery and left superior pulmonary vein. |
![]() View larger version (180K) | Fig. 7. —70-year-old woman with chest pain. CT scan obtained with 2.5-mm collimation shows fluid (arrow) in band-shaped postcaval recess and moderate volume of fluid in irregular-shaped right pulmonic recess (arrowhead). Note small lymph node adjacent to oblique sinus (asterisk). |
![]() View larger version (161K) | Fig. 8. —48-year-old woman with malignant lymphoma. CT scan obtained with 3-mm collimation shows small volume of fluid (arrow) in hemispheric right pulmonary venous recess, which is projecting into lung parenchyma. |
![]() View larger version (209K) | Fig. 9. —65-year-old man with chronic heart failure. CT scan obtained with 2.5-mm collimation shows fluid in band-shaped left pulmonary venous recesses (arrow). Note lymph node (asterisk) adjacent to oblique sinus. E = esophagus. |
The pericardial cavity, which lies between the outer fibrous and inner serous pericardium, has several sinuses and recesses. Cadaveric studies have described the normal anatomy of these recesses [6, 8]. Vesely and Cahill [1] systematically described and named pericardial sinuses and recesses in their cross-sectional anatomic study. On CT, these sinuses and recesses are often observed as near-water-density structures even in patients without significant pericardial effusion. Initial CT studies using thick-section CT depicted some of these recesses, especially the larger ones such as the superior aortic recess and left pulmonic recess, which were seen in approximately 20% of patients. Protopapas and Westcott [4] reported that the left pulmonic recess was recognized in 14 (23%) of 61 patients without pericardial effusion. Although our results using thick-section CT had similar frequencies compared with those of previous studies, our results had a wider range depending on their size and location. Large recesses were identified at 7.7–30.4% prevalence, whereas smaller less-known ones were recognized in less than 5% of the cases.
Recently, multidetector CT and helical CT, which offer faster scanning times and better image quality, are often being used to evaluate various lung diseases. Modern CT scanners have contributed to the improvement of the depiction rate. Kubota et al. [2] reported that the transverse sinus was seen at 37–47% prevalence in 133 patients using helical CT with 3-mm collimation. Our study using thin-section CT also resulted in a remarkable increase in recognition of pericardial recesses. Large recesses, which are also easily identified on thick-section CT even when small in volume, were depicted at almost two to three times greater frequency. The small recesses, which are hardly recognized on thick-section CT because of their size (< 1 cm), were depicted at almost a five times greater frequency. The recesses easily affected by cardiac motion such as the inferior aortic recess were also recognized at approximately five times greater frequency. There is great advantage to using thin-section CT in detecting pericardial sinuses and recesses, especially smaller ones. Our results suggest that the risk of misdiagnosis of these recesses as an abnormal process may increase with better visualization of normal recesses.
In our study, the shapes of the pericardial sinuses and recesses varied. Their shapes are supposed to be related to their location and size. The anterior superior aortic recess and oblique sinus tended to be linear to band-shaped because they exist adjacent to the cardiac surface and extend parallel to the cardiac contour. On the other hand, the recesses surrounded by the vessels and cardiac structures such as the posterior superior aortic recess, the bilateral pulmonic recesses, and the inferior aortic recess tended to have various appearances depending on the gap formed by surrounding structures. The smaller recesses such as the right pulmonary venous recess and the postcaval recess tended to be triangular to round due to their small size.
There is a potential to misdiagnose these normal recesses as abnormal processes. A large recess is easily recognized and is unlikely to present a diagnostic problem, but the crescentic fluid in the anterior superior and inferior aortic recesses can mimic aortic dissection [11, 12] (Fig. 10). Chiles et al. [11] reported that the characteristic shape and location of the recess may prevent an erroneous diagnosis of type A aortic dissection. The band- or irregular-shaped fluid in the anterior superior aortic recess sometimes could be confused with a mediastinal mass, pericardial cysts, thymic cysts, or the thymus [13, 14]. Winer-Muram and Gold [14] reported that valuable signs are the lack of enhancement after contrast administration. Moreover, various mediastinal and hilar lymph nodes lie adjacent to every pericardial recess [15]. For example, paraaortic lymph nodes lie near the anterior superior aortic recess; lower paratracheal lymph nodes, near the posterior superior aortic recess; subaortic lymph nodes, near the left pulmonic recess; hilar and inferior mediastinal lymph nodes, near the right pulmonic recess; paraesophageal and inferior mediastinal lymph nodes, near the oblique sinus; azygous and hilar lymph nodes, near the postcaval recess; and lower lobar lymph nodes and pulmonary ligament nodes, near bilateral pulmonary venous recesses. Although most lymph nodes appear denser than most fluid collections on contrast-enhanced CT, it may still be difficult to distinguish lymph nodes from fluid collections in ovoid or round recesses. Above all, the posterior superior aortic recess is often separated from the ascending aorta by a fat plane and becomes ovoid to round [5]. The right pulmonary venous recess also tends to become round, projecting into the lung parenchyma (Fig. 11). Other recesses may even become ovoid or round when collecting a moderate to large volume of fluid. Furthermore, lymph nodes can appear to have near-water-density necrotic centers. To distinguish some lymph nodes from pericardial fluid, we believe the finding of an enhanced rim may be helpful in establishing the diagnosis of necrotic isodense lymph nodes [2].
![]() View larger version (133K) | Fig. 10. —67-year-old woman with pleural effusion. CT scan obtained with 2.5-mm collimation shows crescentic anterior superior aortic recesses (asterisk) that mimics aortic dissection. Fluid attenuation was 21.1 H. |
![]() View larger version (162K) | Fig. 11. —50-year-old woman with bilateral lower lobe pulmonary embolism. CT scan obtained with 3-mm collimation shows fluid in round right pulmonary venous recess projecting into lung parenchyma, which mimics lower lobar lymph node (arrow). Fluid attenuation was 26.7 H. |
Although thin-section CT improved the depiction rate of every pericardial sinus and recess, it has some limiting factors. Our protocol for thin-section CT was based on the clinical suspicion of pulmonary embolism. Therefore, bolus injection at a high rate resulted in a severe streak artifact from contrast material in the superior vena cava. The postcaval recess is small, and it is the recess most affected by this artifact, which was difficult to evaluate in many cases. We could assess only 63.8% of cases when evaluating the postcaval recess using thin-section CT, whereas thick-section CT at a slower injection rate allowed assessment of 82.3% of cases. Longer scanning delays or slower injection rates may help to prevent this artifact.
In addition, because pericardial effusion has been reported to influence the visualization of the pericardial sinuses and recesses, we excluded the cases with pericardial effusion [8]. However, patients with a suspected pulmonary embolism might have a greater prevalence of right cardiac failure with consequent increased fluid in the pericardial recesses. Levy-Ravetch et al. [8] reported that pericardial effusions were associated with more frequent visualization of pericardial recesses using 10-mm-thick CT slices. Although these researchers did not mention shape, every recess must no longer have appeared loculated but instead had become a thick band because every recess communicates with the others. When the fluid is distributed to the entire pericardial sac, the connection between every recess might be more easily identified.
Cardiac motion also influences the visualization of the pericardial sinuses and recesses and sometimes reduces image quality. The superior aortic recess and left pulmonic recess are not surrounded by mobile structures and are seldom affected by cardiac motion, whereas the inferior aortic recess, left pulmonary venous recess, and right pulmonic recess abutting the left atrium or ventricle are often blurred because of cardiac motion. Electron beam CT offered better image quality that was less affected by cardiac motion [9]. Groell et al. [10] reported that the depiction rate of recesses ranged from 23% to 81% of 100 patients without pericardial effusion, using ECG-triggered electron beam CT. However, electron beam CT has not become a commonly available technique. For better image quality with fewer artifacts, ECG-triggered thin-section CT is preferable.
In conclusion, thin-section CT provides improved visualization of pericardial sinuses and recesses. Fluid collections in these spaces, including less well-known spaces, can be observed more frequently on thin-section CT scans than on thick-section CT scans, even in cases without pericardial effusions. The results were not particularly unexpected; however, an understanding of the location and various appearances of fluid collections helps the radiologist avoid a misdiagnosis of lymphadenopathy or other mediastinal or hilar disease.
Address correspondence to F. Kodama.

Audio Available | 











