| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (A.N., C.C., F.F., F.P., F.V., R.P.) and Surgery "P. Stefanini" (G.S., P.F., N.B.), University of Rome "La Sapienza," Viale Regina Elena, 324, 00100 Rome, Italy. From the 2003 RSNA scientific assembly. Received March 22, 2003; revision requested June 13; final revision received January 14, 2004; accepted January 29. Address correspondence to A.N. (e-mail: alessandro.napoli@uniroma1.it).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty-two patients who were candidates for laparoscopic splenectomy underwent multiphasic multi–detector row CT. Two observers evaluated splenic volume with two hand-tracing editing modalities. Variability between the two observers was calculated with a reliability coefficient (Cronbach 
). A linear regression equation for each modality was generated to identify the correlation between the two observers. Multi–detector row CT angiography was evaluated for assessment of splenic vascular anatomy. Presence and number of both accessory spleens and parenchymal lesions were recorded.
RESULTS: Mean splenic volume was 1,050 and 1,046 mL, respectively, for observers A and B by using each-section editing (technique 1) and 1,067 and 1,068 mL for observers A and B by using distanced editing (technique 2). For each editing modality, reliability coefficient was higher than 0.99. Both techniques 1 and 2 were very highly predictive of specimen weight and had R2 values of greater than 0.99 (P < .001). CT angiograms correctly showed polar arteries in all cases and the presence of the arteria pancreatica magna in one case. Multi–detector row CT demonstrated the presence, number, and size of all accessory spleens and of focal parenchymal lesions.
CONCLUSION: Multi–detector row CT volumetric and anatomic evaluation provided accurate and reproducible information.
© RSNA, 2004
Index terms: Computed tomography (CT), maximum intensity projection, 775.12119 • Computed tomography (CT), multi–detector row, 775.12115 • Computed tomography (CT), volume rendering, 775.12119 • Spleen, CT, 775.12115, 775.12116, 775.12117, 775.12118, 775.12119 • Spleen, size, 775.92
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Splenic volumetric quantification with computed tomography (CT) was first reported in 1979 (1). Subsequently, other authors demonstrated the role of CT for assessment of splenic volume (2,3). With the advent of helical technology, continued application of this modality for volumetric evaluation of visceral parenchymal organs (4,5) has been discussed in the literature.
Offering CT as a preoperative imaging modality for laparoscopic splenectomy, however, requires obtaining not only reproducible volumetric data but also fine parenchymal and vascular detail. The examination also should provide the surgeon with a robust multiplanar organ display. Single–detector row CT was limited in meeting all of these goals (6), and therefore CT application to preoperative assessment for laparoscopic splenectomy was inadequate.
Multi–detector row spiral CT has allowed us to surpass most of the limitations of single–detector row spiral CT. In particular, the very thin section collimation, the high speed of acquisition, and the near isotropy of the voxels allow us to reformat images in any plane without substantial artifacts and with excellent anatomic detail (7–9). The high spatial resolution makes this technique particularly suitable for the assessment of parenchymal volume (10), abdominal vasculature (11,12), and most abdominal diseases (13–16).
Laparoscopy has markedly changed the treatment of various abdominal disorders, which include hematologic disorders that require splenectomy. Laparoscopic splenectomy is a widely accepted technique and has become the standard procedure for treatment of patients affected with benign diseases who have a spleen of normal size (17,18). Although the use of laparoscopic splenectomy in patients with "massive" splenomegaly (defined as interpole diameter > 20 cm or specimen weight > 1,000 g) is controversial, many authors reported this approach to be safe and effective, particularly if splenic anatomy is well known. They therefore have avoided potential complications (19). In fact, in order to perform laparoscopic splenectomy safely, surgeons need anatomic, vascular, and parenchymal information, as well as volume assessment, particularly in the case of massive splenomegaly (20).
The preoperative assessment for laparoscopic splenectomy has long been debated, mostly depending on individual confidence with the various imaging techniques, and still remains undefined (21). Multiphasic CT with a multi–detector row CT scanner now has the potential to afford complete and accurate preoperative imaging determination of splenic vascular anatomy, effective splenic volume and size, and the presence of accessory spleens and parenchymal pathologic conditions.
Thus, the purpose of our study was to prospectively evaluate multi–detector row CT for the determination of splenic volume, splenic vascular anatomy, and presence of accessory spleens and parenchymal lesions in patients who were undergoing laparoscopic splenectomy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Imaging Parameters
CT was performed with a multi–detector row CT scanner (Volume Zoom; Siemens, Erlangen, Germany) with 0.5-second gantry rotation time. All patients received an injection of a standard dose of 140 mL of iomeprol 350 (Iomeron 350; Bracco, Milan, Italy) through an 18-gauge peripheral venous access (generally an antecubital vein) at a flow rate of 4 mL/sec; a standard delay of 22 and 60 seconds, respectively, for the arterial and portal venous phases was applied in all patients. The bolus injection technique was used to administer contrast material with an automated injector (EnVision CT; Medrad, Indianola, Pa) and was carefully monitored by a physician (A.N.).
All patients underwent scanning in the supine position. The CT examination was performed with a triphasic helical CT protocol that included unenhanced scanning of the upper abdomen from the diaphragm to include the entire spleen with a low-spatial-resolution protocol (4 x 2.5-mm collimation, 5.00-mm section thickness, 5-mm reconstruction interval, 12 mm per rotation table feed, 130 mAs, and 120 kVp) and during the arterial and portal venous phases with a high-spatial-resolution protocol (4 x 1-mm collimation, 1.25-mm section thickness, 1-mm reconstruction interval, 5 mm per rotation table feed, 160 mAs, and 120 kVp). The described imaging technique represents a routine scanning protocol for patients who are undergoing thin-section CT of the abdomen at our institution. Postcontrast scan volumes were positioned to include the entire spleen.
Multi–detector row CT data obtained during both arterial and portal venous phases were retrospectively reconstructed with a standard soft algorithm at a 1-mm reconstruction interval. One-millimeter-thick CT images were analyzed with a real-time interaction approach, a dedicated workstation (Kayak XU 800; Hewlett-Packard, Palo Alto, Calif), and three-dimensional rendering software (Vitrea 2.6; Vital Images, Plymouth, Minn).
Image Analysis
CT images were independently evaluated by two readers from the same institution, both experienced in CT vascular and abdominal imaging, with different degrees of experience: reader A (C.C.) was a staff member in vascular radiology with extensive experience (8 years) in abdominal radiology, and reader B (A.N.) was a fellow with experience (4 years) in abdominal radiology. Interpretation of scans was performed with readers blinded to the results of the other readers.
Contrast-enhanced data sets were transferred to a dedicated workstation and were postprocessed by readers A and B who were free to use all available reconstruction algorithms including multiplanar reformatting, maximum intensity projection, and volume-rendering technique. Observers first assessed data by scrolling all images, and then they evaluated splenic volume, splenic vascular anatomy, and presence of accessory spleens. The total time required for image processing and interpretation was approximately 23 and 8 minutes, respectively, when using editing modalities 1 and 2.
Splenic volume.—Splenic volume measurement was obtained by means of the three-dimensional data set obtained during the portal venous phase. Splenic volume was independently measured by the observers with two hand-tracing editing modalities. In the first modality, observers (A1 and B1) measured the volume by tracing the spleen outline by hand on each 1.25-mm-thick transverse image obtained during the portal venous phase. In the second modality, observers (A2 and B2) did not perform tracing by hand on every transverse image. Instead, they followed every splenic contour change, which allowed automatic interpolation between the hand-traced images. To improve the accuracy of volumetric measurements, observers carefully included proximal hilar vascular structures, as these are resected together with the spleen at surgery.
Total splenic weight was assumed to be the gross specimen weight that was summed with the weight of the blood collected during removal of the spleen at the end of the surgical procedure (20). For volume measurements with CT, as the spleen has nearly the same density as that of water (1.003 g/mL ± 0.1 [SD]), we assumed that splenic volume equaled the calculated total splenic weight (1–3).
Vascular anatomy.—Multi–detector row CT angiography evaluation was performed at the workstation where readers were free to use all available reconstruction algorithms and to scroll transverse images.
Selective thin (10–50 mm) maximum intensity projection and volume-rendering techniques were performed to obtain CT angiograms of the hilar vascular anatomy of each spleen. Observers recorded the origin of the splenic artery from the celiac trunk, the presence of anatomic blood vessel variants, and the origin of small arteries that supply the spleen or surrounding organs, particularly the stomach and pancreas. Because of heterogeneous terminal artery branching patterns, arterial vascular images were sorted according to the classification of Michels (22). With this classification, splenic arterial geography is divided into two types, distributed and magistral. In the distributed type, the splenic trunk is short, and a varying number of branches, from six to 12, originate between 3 and 13 cm from the hilum. In the magistral type, there is a long main splenic artery that divides near the hilum into three or four large short terminal branches. Image quality in reconstructed transverse, coronal, or curved planes was diagnostic in all cases.
Accessory spleens and parenchymal lesions.—Observers assessed the presence of accessory spleens and splenic lesions first by analyzing transverse images; multiplanar reformatting was also performed with all data sets for an accurate evaluation of splenic anatomy and its relationships with surrounding organs. Oblique coronal and sagittal images provided additional information to facilitate the depiction of accessory spleens and parenchymal alterations. Number, dimension, and location of each accessory spleen and of splenic lesions were recorded by observers.
Laparoscopic splenectomy.—All patients underwent laparoscopic splenectomy. The spleen was extracted by using a retrieval device (Endo-bag; Ethicon Endosurgery, Cincinnati, Ohio). Morcellation of the organ was performed in all cases of benign disease. In cases of malignancy, the spleen was extracted in large pieces (3 x 3 cm) to save the splenic structural anatomy. There were no intraoperative complications. The volume of the specimen was evaluated immediately after removal of the spleen.
Comparisons.—In all patients, imaging findings were available to surgeons to assess the preoperative strategy and were retrospectively compared with operative findings and gross specimens by four authors (A.N., C.C., G.S., and P.F.).
Statistical Analysis
An reliability coefficient was calculated to determine the interobserver variability between observers A and B. Scatterplot graphic model analysis was used to identify the correlation between observers A and B in measuring total splenic volumes with the two editing modalities (techniques 1 and 2) in all 22 patients. To calculate whether editing modality 1 and 2 could significantly predict the standard of specimen weight, we used linear regression analysis for each modality, where the regression equation was forced through the origin.
To determine whether technique 1 or 2 better predicted specimen weight, we tested whether the R2 values from the two regression equations were markedly different.
A paired t test was used to identify whether the time for the measurement of the splenic volumes was different for the two editing modalities.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Vascular anatomy.—Vascular reconstructions were available to the surgeons to assess the best surgical strategy. Before vascular dissection, surgeons noted the hilar arterial anatomy: Sixteen distributed and six magistral splenic artery patterns were found.
In the distributed pattern, in which multiple branches arise from the main trunk 2 or 3 cm from the hilum, each terminal branch was divided between clips. In the magistral pattern, the pedicle formed by the artery and vein entered the hilum as a compact bundle and was transected en bloc with a single application of a vascular linear laparoscopic stapler. Three inferior and two superior polar segmental arteries were found. In a superior polar splenic artery, the terminal branches supplied the upper portion of the spleen and an accessory spleen, giving at last origin to the gastroepiploic artery. Inferior and superior polar vessels as well as short gastric vessels were divided with a harmonic scalpel (UltraCision; Ethicon).
Accessory spleens and parenchymal lesions.—A total of three accessory spleens were found. In the patient affected by idiopathic thrombocytopenic purpura, two accessory spleens were found. One of them that was 2 cm in diameter was in the inferior pole of the spleen, whereas the other was along the phrenosplenic ligament. The other accessory spleen was located in the hilar region in a patient who had non-Hodgkin lymphoma.
In two patients, focal splenic lesions were found. In a patient with non-Hodgkin lymphoma, eight lesions were found at pathologic examination. The lesions ranged in diameter between 7 and 30 mm (mean, 18 mm). In another patient with acute myeloid leukemia, six lesions were found at pathologic examination, and they ranged in diameter between 5 and 23 mm (mean, 13 mm). Finally, in a patient with massive splenomegaly, a superior polar infarct was demonstrated at pathologic examination.
Imaging Findings
Splenic volume.—Mean splenic volume was 1,050 and 1,046 mL for observers A1 and B1 and 1,067 and 1,068 mL for observers A2 and B2, respectively. For each editing modality, reliability coefficient was greater than 0.99, which indicated near-perfect agreement. For all other analyses, therefore, data were averaged across the two observers.
Both techniques 1 and 2 were very highly predictive of specimen weight and had R2 values of greater than 0.99 (P < .001) (Figs 1, 2). The adjusted R2 values for these regression equations were identical to the third decimal place, which indicated that both editing modalities were equally accurate predictors of specimen weight. When we compared the time for measurement of the splenic volumes with the two editing modalities, technique 2 was significantly faster than technique 1 (P < .001) without any cost in accuracy.
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Indications for laparoscopic splenectomy are similar to those for the open procedures in patients with disorders that include autoimmune disorders (idiopathic thrombocytopenic purpura, thrombotic thrombocytopenic purpura), hereditary hemolytic anemias, hematologic malignancies, and other miscellaneous disorders related to spleen dysfunction. Absolute contraindications are portal hypertension and severe coagulopathy (21). Massive enlargement of the spleen is not an absolute contraindication to laparoscopy. In this regard, the introduction of recent technical advances such as the hand-assisted device (28) helps maintain the advantages of the laparoscopic approach (29).
Laparoscopic splenectomy, however, entails technical difficulties because of splenic vascular anatomy and splenomegaly. Besides, this technique presents special problems, such as the necessity of dealing with a fragile and richly vascularized organ that is situated close to the stomach, the colon, and the tail of the pancreas and the difficulty of specimen extraction (30).
For successful performance of laparoscopic splenectomy, a detailed knowledge of splenic anatomy and volume and potential complications is essential. To our knowledge, this is the first study in which a systematic evaluation of multi–detector row CT for complete preoperative assessment of laparoscopic splenectomy was performed.
One of the questions we sought to answer was whether multi–detector row CT may provide accurate splenic volumetric determination. In this series of 22 patients who were evaluated for laparoscopic splenectomy, multi–detector row CT permitted accurate and reproducible splenic volumetric determination. There was a significant agreement between the two observers in measurements of total splenic volumes with two hand-tracing editing modalities. One of these editing modalities was significantly faster, with no loss in accuracy. In addition, there was significant (P < .001) agreement between the volume calculated by each observer and the splenic weight as measured at the time of surgery. Volume calculations with conventional and helical CT are reported to be relatively accurate (5). These calculations are performed by manually tracing around parenchymal margins on each CT image with an electronic cursor. The cross-sectional area (measured in square centimeters) within the region of interest is determined, and all individual areas are summed, yielding the total splenic volume (measured in cubic centimeters).
The use of multi–detector row technology has dramatically increased the speed of data acquisition, which has resulted in thin-section collimation and decreased motion artifacts, compared with that of conventional scanners. Its accuracy in volume determination has already been tested in other abdominal organs (10,15,31). Commercially available software allows interpolation between the hand-traced sections and automatic determination of the splenic volume, which results in faster image processing. The thin-section acquisition protocol allows accurate three-dimensional reconstructions of the upper abdomen and of the anatomic relationship with surrounding organs, which help surgeons in the determination of the surgical strategy. In fact, a preoperative decision can be made to use a standard laparoscopic approach for the enlarged spleen (defined as a spleen that weighs as much as 1,500 g) or a hand-assisted approach in the case of a super-massive splenomegaly (defined as a spleen that weighs > 1,600 g).
The described technique is simple and can easily be performed with high accuracy in a relatively short time. As demonstrated in this study, to achieve a correct volumetric assessment, it is not necessary to perform calculations in every 1-mm-thick image. Selection of sections in which organ contour changes are evident helps reduce the calculation time.
Multi–detector row CT with increasing applications such as three-dimensional and volumetric imaging is leading into a time in which use of highly detailed CT angiography may replace invasive studies for evaluation of most vascular diseases (13–32).
Complete vascular mapping of the spleen is now feasible because of the technical advances in multi–detector row technology with near isotropic data acquisition and high-resolution postprocessing imaging display. Preoperative multi–detector row CT angiograms were always available to surgeons and were used as a guide to dissection because the surgeons knew about the patterns in the present study.
In regard to vascular anatomy, findings on multi–detector row CT angiograms of the splenic artery may lead surgeons to use the hilar approach strategy, particularly during the learning curve time. They thus may use the vascular stapler when a magistral type of vascularization is present and clips when the distributed type is present. Additionally, before the area where the splenic trunk divides, there usually are a few slender branches to the tail of the pancreas. The most important of these is called the arteria pancreatica magna, a blood vessel familiar to vascular radiologists; occlusion of this branch with embolic material has been reported to result in pancreatitis (33). Next, the splenic artery divides into two to six first and second terminal branches, and these branches undergo two further levels of division into two to 12 penultimate and ultimate branches. Segmental and subsegmental division can occur either outside or inside the spleen. The number of arteries entering the spleen ranges from six to 36.
Another essential issue is the possibility of missing accessory spleens during laparoscopic surgery, and this possibility could be a key factor in the incomplete treatment of idiopathic thrombocytopenic purpura and spherocytosis (23). In regard to the presence and localization of accessory spleens, CT data correctly allowed the prediction of number and sites of aberrant spleen parenchyma. Intraoperative and imaging findings were identical.
A limitation of our study was that surgeons were aware of the imaging findings prior to surgery, but this awareness could not be avoided for obvious ethical reasons. Furthermore, the abdominal CT exploration was limited to the portion of the body that corresponded to the more caudal extension of the spleen, which prevented a complete assessment of remote accessory spleens.
Although this study was performed in a limited series of patients, results demonstrate that thin-section multi–detector row CT provides a detailed preoperative evaluation of parenchymal lesions and vascular anatomy of the spleen with accurate volumetric quantification, and findings of this evaluation can be useful for preoperative surgical planning.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. W. Farraher, H. Jara, K. J. Chang, A. Hou, and J. A. Soto Liver and Spleen Volumetry with Quantitative MR Imaging and Dual-Space Clustering Segmentation Radiology, October 1, 2005; 237(1): 322 - 328. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
