- Email: [email protected]
Comparison of MR and CT scanning in severe acute pancreatitis: Initial experiences
Clinical Radiology (1993) 48, 111 116
Comparison of MR and CT Scanning in Severe Acute Pancreatitis: Initial Experiences A. S A I F U D D I N , J. W A R D , J. R I D G W A Y and A. G. C H A L M E R S M R Imaging Unit, St James's University Hospital, Leeds, and C T Scanning Unit, The General Infirmary at Leeds
Magnetic resonance imaging was performed at 1.0 T in seven patients with severe acute pancreatitis. A T2-weighted spin echo sequence and a breath-hold multislice rapid gradient echo sequence (TurboFLASH) were used in each patient. TurboFLASH imaging was performed before and after intravenous gadopentetate-dimeglumine (Gd-DTPA). All MRI images were compared with a recent contrast-enhanced CT scan. Postgadolinium M R I was equivalent to contrast-enhanced CT in differentiating viable pancreatic parenchyma from areas of pancreatic necrosis. M R I identified the presence of gas in a case of pancreatic abscess but failed to identify small foci of pancreatic calcification demonstrated in one case by CT. M R I was also equivalent to CT in assessing the location and extent of peripancreatic inflammatory changes and fluid collections. However, MRI, particularly the T2-weighted spin echo, was superior to CT in characterizing the complex nature of such inflammatory changes in one case. Initial experience suggests that M R I is a valuable technique in assessing patients with severe acute pancreatitis. Saifuddin, A., Ward, J., Ridgway, J. & Chalmers, A.G. (1993). Clinical Radiology 48, 111-116. Comparison of M R and CT Scanning in Severe Acute Pancreatitis: Initial Experiences Accepted f o r Publication 15 February 1993
Computed tomography (CT) is now established as the imaging technique of~hoice in the assessment o f acute pancreatitis [1], and in the diagnosis of pancreatic parenchymal necrosis [2-5]. However, CT is associated with a relatively high radiation dose [6] and the need for administration of large volumes of intravenous contrast medium. These disadvantages are compounded by the fact that many patients with severe acute pancreatitis are relatively young and require multiple examinations, often at weekly intervals. For this reason, the value o f magnetic resonance imaging (MRI) in seven patients with severe acute pancreatitis has been assessed and a limited comparison with CT has been made.
P A T I E N T S AND M E T H O D S Seven patients with severe acute pancreatitis were investigated during the course of their illness with both CT and MRI. Patient details are presented in Table 1. Correspondence to: Dr A. G. Chalmers, Consultant Radiologist, CT Scanning Unit, Clarendon Wing, The General Infirmary at Leeds, Belmont Grove, Leeds LS2 9NS.
All CT scans were performed on a GE9800 CT scanner and consisted of: (i) a precontrast study, 10 mm thick slices at 20 mm intervals through the upper abdomen to cover the pancreas and any peripancreatic inflammatory changes or fluid collections, and (ii) a dynamic postcontrast study targetted to the pancreas, contiguous or alternate 5 mm thick slices through the pancreas during the intravenous injection of 150 ml Iopamidol 300. Follow-up scans without Contrast were performed to assess the progress of peripancreatic inflammatory changes or fluid collections, or to guide interventional procedures. Intravenous contrast medium was used to identify or further document the extent of pancreatic necrosis. M R I scans were performed on each patient early in the course of their illness. All scans were performed on a Siemens Magnetom 42 SP at a full field strength of 1.0 T. Each study consisted of a combination of: (i) an axial proton density (PDW) and T2-weighted (T2W) spin-echo sequence (TR 2000, TE 45/90 ms); and (ii) a breath-hold multislice, Tl-weighted rapid gradient echo sequence (TurboFLASH). For the breath-hold series, a TE of 4 ms and flip angle o f 80 ~ was used. In five cases, 11 slices were acquired using a T R of 100 ms. In two cases, 15 slices were
Table 1 - Details of the seven patients with severe acute pancreatitis Case no.
Sex~age
Complication
No. of CT scans*
Time betweenMR1 and most recent CT
I 2 3 4 5 6 7
F/21 F/35 M/37 M/71 F/30 M/57 M/52
Inflammatorymass in pancreatic head Necrosis, pseudocyst Pseudocyst Infected pancreatic necrosis Extensive extra-pancreatic fluid collections Necrosis, pseudocyst Necrosis, pseudocyst
2 (21.1.92 1.4.92) 6 (22.1.91 18.4.91) 6 (23.7.91-27.11.91) 6 (2.7.91-29.8.9I) 2 (1.8.91-13.8.91) 6 (27.6.91 7.8.91) 1I. (3.1.91-22.4.91)
16 days 3 days 2 days 7 days and same day Same day 30 days 7 days
* Dates indicate the time between first and last scans.
I 12
CLINICAL RADIOLOGY
acquired using a T R o f 156 ms and rectangular field of view. In each case the breath-hold period was 19 s. An axial acquisition was performed before, immediately after, and at 1 min following a bolus injection of 0.1 mmol/kg gadopentetate-dimeglumine (Gd-DTPA; MAGNEVIST, Schering). A coronal acquisition was performed at 2 min post-injection. The pancreatic appearances on each MRI sequence were established and pancreatic viability was assessed by the parenchymal response to the intravenous gadolinium. The location and extent of peripancreatic inflammatory changes were assessed. Each M R scan was compared with the most recent contrast-enhanced CT study.
(a)
RESULTS
Assessment of the Pancreas
(i) Spin-Echo Imaging Pancreatic tissue was easily identified in six out of seven cases (in one patient, case 4, there was gross parenchymal necrosis). On both T2W and P D W images the normal viable pancreatic parenchyma displayed intermediate signal intensity (Fig. la). In cases 1 and 3, a heterogeneous intensity inflammatory mass was demonstrated in the pancreatic head analogous to a mass of mixed fluid and solid density on the corresponding CT scans. The mass in case 3 was seen on MRI to contain a fluid-debris level (Fig. 2). This was best seen on T2W images and was more easily appreciated than on the corresponding CT section. CT in one patient (case 2) demonstrated several small foci of calcification in the pancreatic head, a finding not identified on any M R sequence (Fig. 3). In one patient (case 7), an area of presumed parenchymal necrosis appeared hyperintense and was indistinguishable from peripancreatic fluid. The two were also isodense on postcontrast CT (Fig. 4 a, b).
(b)
(ii) TurboFLASH Imaging On precontrast T u r b o F L A S H images, normal pancreatic tissue appeared of low signal intensity (Fig. lb) and was difficult to distinguish from surrounding bowel in two patients (cases 2 and 5), due to a paucity of peripancreatic fat. In all other cases, the pancreas was clearly distinguished from the hyperintense surrounding fat. The heterogeneity of signal intensity in the pancreatic head abnormalities seen on T2W images (cases 1 and 3) was far less evident on T u r b o F L A S H sequences. Following intravenous Gd-DTPA, clear enhancement of normal pancreatic parenchyma (Fig. 1c) was identified in all cases. The location and extent of enhancing tissue corresponded exactly to that seen on dynamic postcontrast CT studies (Fig. 5). Similarly, areas of non-enhancing parenchyma on MRI were identical to those defects documented by CT. In one patient (case 6) who had had a prior necrosectomy, it was difficult to identify the residual pancreas due to a paucity ofretroperitoneal fat. However, postgadolinium M R scans clearly differentiated viable pancreatic tissue from surrounding bowel. In case 7, the area of pancreatic necrosis that was isointense to fluid on T2W images was hypointense to fluid on T u r b o F L A S H images (Fig. 4c).
(c) Fig. 1 Case 3. (a) T2-weighted SE, (b) pregadolinium Tl-weighted TurboFLASH, and (c) postgadolinium Tl-weighted TurboFLASH images of the pancreas. Pancreatic parenchyma (arrow) has intermediate signalintensityon T2-weightedspin-echosequencesand low signal intensityon pregadoliniumTurboFLASHsequences.Followinggadolinium, uniform pancreaticenhancementis demonstrated. The apparent mass in the pancreatic body (curved arrow) is due to phase encoding artefact from the aorta.
Assessment of Pcripancreatic Inflammatory Changes and Fluid Collections
(i) Spin-Echo Imaging Peripancreatic inflammatory changes were identified as hypointense strands within the hyperintense retroperitoneal and mesenteric fat on PDW images (cases 1, 3 and 6). These corresponded to areas of increased density within
1 13
MR AND CT IN ACUTE PANCREATITIS
Fig. 2 - Case 3. T2-weighted SE image of the pancreas. A fluid-debris level is demonstrated in the pancreatic head cyst (arrow).
(a)
(b)
A (a)
(c) (b) Fig. 3 - Case 2. (a) Contrast-enhanced CT and (b) T2-weighted SE images in a patient with pseudocyst. Calcification in the pancreatic head is not identified on MRI. M R demonstrates the complex nature of the pseudocyst better than CT.
Fig. 4 - Case 7. (a) Contrast-enhanced CT, (b) T2-weighted SE, and (c) postgadolinium T l-weighted TurboFLASH images of the pancreas. An area of presumed necrosis in the pancreatic head (curved arrows) is distinguished from peripancreatic fluid (straight arrow) only on the TurboFLASH sequences. Gall-stones are identified in the gall-bladder (g).
114
CLINICAL RADIOLOGY
(a) (a)
(b)
(b) Fig. 5 - Case 4. (a) Contrast-enhanced CT and (b) postgadolinium T1weighted TurboFLASH images in a patient with gross pancreatic
necrosis. There is exact correlation between CT and MRI in the demonstration of viable pancreatic parenchyma (arrows). The fluid appears hypointense on the TurboFLASH MRI sequence. Splenicvein patency (s) is confirmed on MRI. the peripancreatic fat on CT. Fluid collections had moderately high signal intensity on P D W images becoming markedly hyperintense on T2W images (Figs 3b, 4b, 6b and 7b). Both sequences, particularly the T2-weighted images, identified a greater degree of heterogeneity within these collections than was evident on the corresponding CT study (Fig. 3b). In one patient (case 5) the signal intensity within a multiloculated fluid collection was identical to that of fluid-filled small bowel on T2W scans (Fig. 6). However, the two were easily differentiated on P D W images. In the patient with infected necrosis (case 4), a gas-fluid level was readily identified (Fig. 8).
(ii) TurboFLASH Imaging Inflammatory changes and fluid collections all appeared hypointense on precontrast T u r b o F L A S H sequences (Figs 4c, 5b and 6c). The heterogeneity of the collections was not as evident as that seen on T2W imaging. The gas-fluid level was difficult to identify on precontrast images of the pancreatic abscess. In two
(c) Fig. 6 - Case 5. (a) Contrast-enhanced CT, (b) T2-weighted S'E, and (c) postgadolinium Tl-weighted TurboFLASH images. Fluid-filled small bowel (small arrow) is isointense to fluid in a multiloculated paracolic collection (curved arrow) on the T2-weighted SE sequences. Enhancement of the smallbowelwallon postgadoliniumTurboFLASH images differentiatesthe two. patients (cases 5 and 6), fluid-filled small bowel was isointense on precontrast images to adjacent fluid collections. However, postcontrast images showed enhancement of the small bowel wall which allowed differentiation between the two (Fig. 6c). Other Findings
Splenic vein patency was confirmed in all cases by signal void on T2W and pregadolinium T u r b o F L A S H
MR AND CT IN ACUTE PANCREATITIS
! 15
(a)
(a)
(b) Fig. 7 - Case 4. (a) Contrast-enhanced CT and (b) T2-weighted SE images in a patient with infected pancreatic necrosis. A gas-fluid level is demonstrated on MRI.
images (Fig. la), and vessel enhancement on postgadolinium images (Fig. lc). This was of particular value in one patient (case 4) where CT had demonstrated a compressed splenic vein and occlusion could not be excluded (Fig. 5). In one patient with alcohol-related pancreatitis (case 5), CT evidence of fatty change in the liver was also identified on postgadolinium TurboFLASH images as a region of reduced parenchymal liver enhancement. In case 7, multiple gall-stones were identified in the gallbladder on T2W images (Fig. 4b). These were also seen on CT. Coronal postgadolinium scans were of value in identifying the relationships of inflammatory masses and fluid collections (Fig. 8), and in demonstrating the spread of inflammation into the mesentery. DISCUSSION At present, CT is the technique of choice for imaging severe acute pancreatitis [1]. Precontrast scans demon-
(b) Fig. 8 - ( a ) Case 5. Postgadolinium Tl-weighted TurboFLASH sequence demonstrating the relationship of a large right anterior pararenal collection (short arrow) to the pancreas (long arrow). (b) Case 3. Value of coronal imaging. Postgadolinium Tl-weighted TurboFLASH image, demonstrating the relationship of a pancreatic head cyst (curved arrow) to the transverse colon (long arrow) and the portal vein (short arrow).
strate the location and extent of pancreatic and peripancreatic inflammatory changes and fluid collections. Gallstones and biliary tract dilatation may be identified. Dynamic postcontrast CT has been shown to be accurate in the assessment of pancreatic necrosis [2-5] and vascular complications [7]. The need for repeated CT examinations during the course of the patient's illness is also well established [4]. Consequently, the accumulated radiation dose in relatively young patients can be significant. Early experience with MRI of the pancreas has been limited by artefacts from respiration, vascular pulsation and bowel peristalsis, related to the long scanning times
1 16
CLINICAL RADIOLOGY
with conventional spin-echo techniques. Also, the absence of a suitable small bowel contrast medium has been a limitation in assessing peripancreatic inflammatory changes. However, the advent o f breath-hold multislice T u r b o F L A S H sequences has largely overcome these problems. In a recent study comparing six different M R sequences [8], it was found that fat-suppressed fast low angle shot (FLASH) and fat-suppressed spin-echo (FSSE) sequences produced the best images of the pancreas, although non-suppressed F L A S H was considered an acceptable alternative if fat suppression was not available. Furthermore, normal postgadolinium appearances of the pancreas have been established, with the gland parenchyma showing uniform enhancement which is maximal in the first 2 min after injection [9,10]. Previously reported M R I findings in acute pancreatitis have not included patients with severe necrotizing disease. An early study at 0.08 T [11] demonstrated enlargement and loss of definition of the gland. Calculated Tl relaxation values were of no help in differentiating inflamed from normal pancreatic tissue. All fluid collections and pseudocysts were identified. A more recent study at 1.5 T [12] using combinations of FSSE and breath-hold ~multislice F L A S H and T u r b o F L A S H sequences before and after G d - D T P A , demonstrated enlargement and loss of definition of the pancreas in four patients with acute pancreatitis. Peripancreatic inflammation was manifest as low signal intensity strands within the hyperintense surrounding fat, best identified on F L A S H sequences. Delineation of the pancreas was best on FSSE sequences but the peripancreatic inflammatory changes were not appreciated. Uniform enhancement of the gland was seen in all cases. In the present study, a combination of axial dual echo T2W and breath-hold T u r b o F L A S H sequences before and after G d - D T P A was performed in seven patients with severe acute pancreatitis. The findings were compared with those demonstrated on recent dynamic contrastenhanced CT scans. In all cases, viable pancreatic tissue was identified on postgadolinium T u r b o F L A S H scans and the extent of enhancing tissue corresponded exactly to that seen on the respective postcontrast CT studies. M R I failed to detect the single case of pancreatic head calcification. In one patient (case 2), M R I was superior to CT in characterizing the complex nature of associated fluid collections. This feature is o f importance, since it has been shown that complex fluid collections are less likely to respond to percutaneous catheter drainage than simple fluid collections [13]. The potential for T u r b o F L A S H M R sequences in distinguishing inflammatory changes from presumed necrosis was well illustrated by case 7, where both postcontrast CT and T2W M R I failed to differentiate the two. In a single case of pancreatic abscess, M R I was able to identify gas within the collection, which was subsequently confirmed on CT performed for interventional purposes. The multiplanar imaging capability of M R I was also found to be of value in selected cases, allowing better definitic~n of the anatomical relationships of pancreatic masses and fluid collec-
tions. M R I was equivalent to, and in one case, better than CT in assessing vascular patency. Although there were no cases of splenic or portal vein thrombosis in this study, current experience indicates this complication would be identified by M R I [14]. The major limitation of M R I at present is availability. The small number of cases in this series is partly due to the difficulty in transferring severely ill and often ventilated patients from one hospital to another for M R imaging. However, once such limitations are overcome, it would appear from the results from this study, that M R I has a valuable place in the investigation of patients with severe acute pancreatitis. Acknowledgements. We are grateful to: Mr M. J. McMahon for referring his patients; Medical Illustration at Leeds General Infirmary; and Mrs Susan Lawrence for typing the manuscript.
REFERENCES
1 Balthazar EJ. CT diagnosis and staging of acute pancreatitis. Radiologic Clinics of North America 1989;27:19-37, 2 Kivisaari L, Somer K, Standertskjold-NardenstamGG, Schroder T, Kivilaasko E, Lempinen M. A new method for diagnosis of acute haemorrhagic-necrotizingpancreatitis using contrast enhanced CT. Gastrointestinal Radiology 1984;9:27 30, 3 Larvin M, Chalmers AG, McMahon MJ. Dynamic contrast enhanced computed tomography: a precisetechnique for identifying and localising pancreatic necrosis. British Medical Journal 1990;300:1425-1428. 4 Balthazar EJ, Robinson DL, Megibow AJ, Ranson JHC. Acute pancreatitis; value of CT in establishing prognosis. Radiology 1990;174:331-336. 5 Johnson CD, Stephens DH, Sarr MG. CT of acute pancreatitis; correlation between lack of contrast enhancement and pancreatic necrosis. American Journal of Roentgenology 1991;156:93-95. 6 Documents of the NRPB. Patient Dose Reduction in Diagnostic Radiology 1990;1:15. 7 VujicI. Vascularcomplicationsofpancreatitis. Radiologic Clinicsof North America 1989;27:81 91. 8 SemelkaRC, SimmFC, Recht MP, DeimlingM, Lenz G, Laub GA. MR imagingof the pancreas at high fieldstrength; comparison of six sequences. Journal of Computer Assisted Tomography 1991;15:966971. 9 Chezmar JL, Nelson RC, Small WC, Bernadino ME. Magnetic resonance imagingof the pancreas with gadolinium-DTPA: Gastrointestinal Radiology 1991;16:139-142. I0 Hamed MM, Harem B, Ibrahim ME, Taupitz M, Mahfouz AE. Dynamic MR imaging of the abdomen with gadopentate dimeglumine: normal enhancement patterns of liver, spleen, stomach and pancreas. American Journal of Roentgenology 1992;158:303 307. 11 Smith FW, BaylissAP, HusseyJK, Robertson EM, Weir J, Crosher GA. Low-field(0.08 T) magneticresonanceimagingof the pancreas; comparison with computed tomography and ultrasound. British Journal of Radiology 1987;62:796 802. 12 Semelka RC, Kroeker MA, Shoenut JP, Kroeker R, Yaffe CS, Micflikier AB. Pancreatic disease: prospective comparison of CT, ERCP and 1.5 T MR imaging with dynamic gadolinium enhancement and fat suppression. Radiology 1991;181:785 791. 13 Lee MJ, Rattner DW, LegemateDA, Saini S, Dawson SL, Hahn PF, Warshaw AL, Mueller PR. Acute complicatedpancreatitis: redefining the role of interventional radiology. Radiology 1992;183:171 174. 14 Ward J, Martinez D, Chalmers AG, Ridgway JP, Robinson PJ, Smith MA. Rapid dynamic contrast enhanced magnetic resonance imaging of the liver and portal vein. British Journal of Radiology 1993;66:214-222.
Comparison of MR and CT Scanning in Severe Acute Pancreatitis: Initial Experiences A. S A I F U D D I N , J. W A R D , J. R I D G W A Y and A. G. C H A L M E R S M R Imaging Unit, St James's University Hospital, Leeds, and C T Scanning Unit, The General Infirmary at Leeds
Magnetic resonance imaging was performed at 1.0 T in seven patients with severe acute pancreatitis. A T2-weighted spin echo sequence and a breath-hold multislice rapid gradient echo sequence (TurboFLASH) were used in each patient. TurboFLASH imaging was performed before and after intravenous gadopentetate-dimeglumine (Gd-DTPA). All MRI images were compared with a recent contrast-enhanced CT scan. Postgadolinium M R I was equivalent to contrast-enhanced CT in differentiating viable pancreatic parenchyma from areas of pancreatic necrosis. M R I identified the presence of gas in a case of pancreatic abscess but failed to identify small foci of pancreatic calcification demonstrated in one case by CT. M R I was also equivalent to CT in assessing the location and extent of peripancreatic inflammatory changes and fluid collections. However, MRI, particularly the T2-weighted spin echo, was superior to CT in characterizing the complex nature of such inflammatory changes in one case. Initial experience suggests that M R I is a valuable technique in assessing patients with severe acute pancreatitis. Saifuddin, A., Ward, J., Ridgway, J. & Chalmers, A.G. (1993). Clinical Radiology 48, 111-116. Comparison of M R and CT Scanning in Severe Acute Pancreatitis: Initial Experiences Accepted f o r Publication 15 February 1993
Computed tomography (CT) is now established as the imaging technique of~hoice in the assessment o f acute pancreatitis [1], and in the diagnosis of pancreatic parenchymal necrosis [2-5]. However, CT is associated with a relatively high radiation dose [6] and the need for administration of large volumes of intravenous contrast medium. These disadvantages are compounded by the fact that many patients with severe acute pancreatitis are relatively young and require multiple examinations, often at weekly intervals. For this reason, the value o f magnetic resonance imaging (MRI) in seven patients with severe acute pancreatitis has been assessed and a limited comparison with CT has been made.
P A T I E N T S AND M E T H O D S Seven patients with severe acute pancreatitis were investigated during the course of their illness with both CT and MRI. Patient details are presented in Table 1. Correspondence to: Dr A. G. Chalmers, Consultant Radiologist, CT Scanning Unit, Clarendon Wing, The General Infirmary at Leeds, Belmont Grove, Leeds LS2 9NS.
All CT scans were performed on a GE9800 CT scanner and consisted of: (i) a precontrast study, 10 mm thick slices at 20 mm intervals through the upper abdomen to cover the pancreas and any peripancreatic inflammatory changes or fluid collections, and (ii) a dynamic postcontrast study targetted to the pancreas, contiguous or alternate 5 mm thick slices through the pancreas during the intravenous injection of 150 ml Iopamidol 300. Follow-up scans without Contrast were performed to assess the progress of peripancreatic inflammatory changes or fluid collections, or to guide interventional procedures. Intravenous contrast medium was used to identify or further document the extent of pancreatic necrosis. M R I scans were performed on each patient early in the course of their illness. All scans were performed on a Siemens Magnetom 42 SP at a full field strength of 1.0 T. Each study consisted of a combination of: (i) an axial proton density (PDW) and T2-weighted (T2W) spin-echo sequence (TR 2000, TE 45/90 ms); and (ii) a breath-hold multislice, Tl-weighted rapid gradient echo sequence (TurboFLASH). For the breath-hold series, a TE of 4 ms and flip angle o f 80 ~ was used. In five cases, 11 slices were acquired using a T R of 100 ms. In two cases, 15 slices were
Table 1 - Details of the seven patients with severe acute pancreatitis Case no.
Sex~age
Complication
No. of CT scans*
Time betweenMR1 and most recent CT
I 2 3 4 5 6 7
F/21 F/35 M/37 M/71 F/30 M/57 M/52
Inflammatorymass in pancreatic head Necrosis, pseudocyst Pseudocyst Infected pancreatic necrosis Extensive extra-pancreatic fluid collections Necrosis, pseudocyst Necrosis, pseudocyst
2 (21.1.92 1.4.92) 6 (22.1.91 18.4.91) 6 (23.7.91-27.11.91) 6 (2.7.91-29.8.9I) 2 (1.8.91-13.8.91) 6 (27.6.91 7.8.91) 1I. (3.1.91-22.4.91)
16 days 3 days 2 days 7 days and same day Same day 30 days 7 days
* Dates indicate the time between first and last scans.
I 12
CLINICAL RADIOLOGY
acquired using a T R o f 156 ms and rectangular field of view. In each case the breath-hold period was 19 s. An axial acquisition was performed before, immediately after, and at 1 min following a bolus injection of 0.1 mmol/kg gadopentetate-dimeglumine (Gd-DTPA; MAGNEVIST, Schering). A coronal acquisition was performed at 2 min post-injection. The pancreatic appearances on each MRI sequence were established and pancreatic viability was assessed by the parenchymal response to the intravenous gadolinium. The location and extent of peripancreatic inflammatory changes were assessed. Each M R scan was compared with the most recent contrast-enhanced CT study.
(a)
RESULTS
Assessment of the Pancreas
(i) Spin-Echo Imaging Pancreatic tissue was easily identified in six out of seven cases (in one patient, case 4, there was gross parenchymal necrosis). On both T2W and P D W images the normal viable pancreatic parenchyma displayed intermediate signal intensity (Fig. la). In cases 1 and 3, a heterogeneous intensity inflammatory mass was demonstrated in the pancreatic head analogous to a mass of mixed fluid and solid density on the corresponding CT scans. The mass in case 3 was seen on MRI to contain a fluid-debris level (Fig. 2). This was best seen on T2W images and was more easily appreciated than on the corresponding CT section. CT in one patient (case 2) demonstrated several small foci of calcification in the pancreatic head, a finding not identified on any M R sequence (Fig. 3). In one patient (case 7), an area of presumed parenchymal necrosis appeared hyperintense and was indistinguishable from peripancreatic fluid. The two were also isodense on postcontrast CT (Fig. 4 a, b).
(b)
(ii) TurboFLASH Imaging On precontrast T u r b o F L A S H images, normal pancreatic tissue appeared of low signal intensity (Fig. lb) and was difficult to distinguish from surrounding bowel in two patients (cases 2 and 5), due to a paucity of peripancreatic fat. In all other cases, the pancreas was clearly distinguished from the hyperintense surrounding fat. The heterogeneity of signal intensity in the pancreatic head abnormalities seen on T2W images (cases 1 and 3) was far less evident on T u r b o F L A S H sequences. Following intravenous Gd-DTPA, clear enhancement of normal pancreatic parenchyma (Fig. 1c) was identified in all cases. The location and extent of enhancing tissue corresponded exactly to that seen on dynamic postcontrast CT studies (Fig. 5). Similarly, areas of non-enhancing parenchyma on MRI were identical to those defects documented by CT. In one patient (case 6) who had had a prior necrosectomy, it was difficult to identify the residual pancreas due to a paucity ofretroperitoneal fat. However, postgadolinium M R scans clearly differentiated viable pancreatic tissue from surrounding bowel. In case 7, the area of pancreatic necrosis that was isointense to fluid on T2W images was hypointense to fluid on T u r b o F L A S H images (Fig. 4c).
(c) Fig. 1 Case 3. (a) T2-weighted SE, (b) pregadolinium Tl-weighted TurboFLASH, and (c) postgadolinium Tl-weighted TurboFLASH images of the pancreas. Pancreatic parenchyma (arrow) has intermediate signalintensityon T2-weightedspin-echosequencesand low signal intensityon pregadoliniumTurboFLASHsequences.Followinggadolinium, uniform pancreaticenhancementis demonstrated. The apparent mass in the pancreatic body (curved arrow) is due to phase encoding artefact from the aorta.
Assessment of Pcripancreatic Inflammatory Changes and Fluid Collections
(i) Spin-Echo Imaging Peripancreatic inflammatory changes were identified as hypointense strands within the hyperintense retroperitoneal and mesenteric fat on PDW images (cases 1, 3 and 6). These corresponded to areas of increased density within
1 13
MR AND CT IN ACUTE PANCREATITIS
Fig. 2 - Case 3. T2-weighted SE image of the pancreas. A fluid-debris level is demonstrated in the pancreatic head cyst (arrow).
(a)
(b)
A (a)
(c) (b) Fig. 3 - Case 2. (a) Contrast-enhanced CT and (b) T2-weighted SE images in a patient with pseudocyst. Calcification in the pancreatic head is not identified on MRI. M R demonstrates the complex nature of the pseudocyst better than CT.
Fig. 4 - Case 7. (a) Contrast-enhanced CT, (b) T2-weighted SE, and (c) postgadolinium T l-weighted TurboFLASH images of the pancreas. An area of presumed necrosis in the pancreatic head (curved arrows) is distinguished from peripancreatic fluid (straight arrow) only on the TurboFLASH sequences. Gall-stones are identified in the gall-bladder (g).
114
CLINICAL RADIOLOGY
(a) (a)
(b)
(b) Fig. 5 - Case 4. (a) Contrast-enhanced CT and (b) postgadolinium T1weighted TurboFLASH images in a patient with gross pancreatic
necrosis. There is exact correlation between CT and MRI in the demonstration of viable pancreatic parenchyma (arrows). The fluid appears hypointense on the TurboFLASH MRI sequence. Splenicvein patency (s) is confirmed on MRI. the peripancreatic fat on CT. Fluid collections had moderately high signal intensity on P D W images becoming markedly hyperintense on T2W images (Figs 3b, 4b, 6b and 7b). Both sequences, particularly the T2-weighted images, identified a greater degree of heterogeneity within these collections than was evident on the corresponding CT study (Fig. 3b). In one patient (case 5) the signal intensity within a multiloculated fluid collection was identical to that of fluid-filled small bowel on T2W scans (Fig. 6). However, the two were easily differentiated on P D W images. In the patient with infected necrosis (case 4), a gas-fluid level was readily identified (Fig. 8).
(ii) TurboFLASH Imaging Inflammatory changes and fluid collections all appeared hypointense on precontrast T u r b o F L A S H sequences (Figs 4c, 5b and 6c). The heterogeneity of the collections was not as evident as that seen on T2W imaging. The gas-fluid level was difficult to identify on precontrast images of the pancreatic abscess. In two
(c) Fig. 6 - Case 5. (a) Contrast-enhanced CT, (b) T2-weighted S'E, and (c) postgadolinium Tl-weighted TurboFLASH images. Fluid-filled small bowel (small arrow) is isointense to fluid in a multiloculated paracolic collection (curved arrow) on the T2-weighted SE sequences. Enhancement of the smallbowelwallon postgadoliniumTurboFLASH images differentiatesthe two. patients (cases 5 and 6), fluid-filled small bowel was isointense on precontrast images to adjacent fluid collections. However, postcontrast images showed enhancement of the small bowel wall which allowed differentiation between the two (Fig. 6c). Other Findings
Splenic vein patency was confirmed in all cases by signal void on T2W and pregadolinium T u r b o F L A S H
MR AND CT IN ACUTE PANCREATITIS
! 15
(a)
(a)
(b) Fig. 7 - Case 4. (a) Contrast-enhanced CT and (b) T2-weighted SE images in a patient with infected pancreatic necrosis. A gas-fluid level is demonstrated on MRI.
images (Fig. la), and vessel enhancement on postgadolinium images (Fig. lc). This was of particular value in one patient (case 4) where CT had demonstrated a compressed splenic vein and occlusion could not be excluded (Fig. 5). In one patient with alcohol-related pancreatitis (case 5), CT evidence of fatty change in the liver was also identified on postgadolinium TurboFLASH images as a region of reduced parenchymal liver enhancement. In case 7, multiple gall-stones were identified in the gallbladder on T2W images (Fig. 4b). These were also seen on CT. Coronal postgadolinium scans were of value in identifying the relationships of inflammatory masses and fluid collections (Fig. 8), and in demonstrating the spread of inflammation into the mesentery. DISCUSSION At present, CT is the technique of choice for imaging severe acute pancreatitis [1]. Precontrast scans demon-
(b) Fig. 8 - ( a ) Case 5. Postgadolinium Tl-weighted TurboFLASH sequence demonstrating the relationship of a large right anterior pararenal collection (short arrow) to the pancreas (long arrow). (b) Case 3. Value of coronal imaging. Postgadolinium Tl-weighted TurboFLASH image, demonstrating the relationship of a pancreatic head cyst (curved arrow) to the transverse colon (long arrow) and the portal vein (short arrow).
strate the location and extent of pancreatic and peripancreatic inflammatory changes and fluid collections. Gallstones and biliary tract dilatation may be identified. Dynamic postcontrast CT has been shown to be accurate in the assessment of pancreatic necrosis [2-5] and vascular complications [7]. The need for repeated CT examinations during the course of the patient's illness is also well established [4]. Consequently, the accumulated radiation dose in relatively young patients can be significant. Early experience with MRI of the pancreas has been limited by artefacts from respiration, vascular pulsation and bowel peristalsis, related to the long scanning times
1 16
CLINICAL RADIOLOGY
with conventional spin-echo techniques. Also, the absence of a suitable small bowel contrast medium has been a limitation in assessing peripancreatic inflammatory changes. However, the advent o f breath-hold multislice T u r b o F L A S H sequences has largely overcome these problems. In a recent study comparing six different M R sequences [8], it was found that fat-suppressed fast low angle shot (FLASH) and fat-suppressed spin-echo (FSSE) sequences produced the best images of the pancreas, although non-suppressed F L A S H was considered an acceptable alternative if fat suppression was not available. Furthermore, normal postgadolinium appearances of the pancreas have been established, with the gland parenchyma showing uniform enhancement which is maximal in the first 2 min after injection [9,10]. Previously reported M R I findings in acute pancreatitis have not included patients with severe necrotizing disease. An early study at 0.08 T [11] demonstrated enlargement and loss of definition of the gland. Calculated Tl relaxation values were of no help in differentiating inflamed from normal pancreatic tissue. All fluid collections and pseudocysts were identified. A more recent study at 1.5 T [12] using combinations of FSSE and breath-hold ~multislice F L A S H and T u r b o F L A S H sequences before and after G d - D T P A , demonstrated enlargement and loss of definition of the pancreas in four patients with acute pancreatitis. Peripancreatic inflammation was manifest as low signal intensity strands within the hyperintense surrounding fat, best identified on F L A S H sequences. Delineation of the pancreas was best on FSSE sequences but the peripancreatic inflammatory changes were not appreciated. Uniform enhancement of the gland was seen in all cases. In the present study, a combination of axial dual echo T2W and breath-hold T u r b o F L A S H sequences before and after G d - D T P A was performed in seven patients with severe acute pancreatitis. The findings were compared with those demonstrated on recent dynamic contrastenhanced CT scans. In all cases, viable pancreatic tissue was identified on postgadolinium T u r b o F L A S H scans and the extent of enhancing tissue corresponded exactly to that seen on the respective postcontrast CT studies. M R I failed to detect the single case of pancreatic head calcification. In one patient (case 2), M R I was superior to CT in characterizing the complex nature of associated fluid collections. This feature is o f importance, since it has been shown that complex fluid collections are less likely to respond to percutaneous catheter drainage than simple fluid collections [13]. The potential for T u r b o F L A S H M R sequences in distinguishing inflammatory changes from presumed necrosis was well illustrated by case 7, where both postcontrast CT and T2W M R I failed to differentiate the two. In a single case of pancreatic abscess, M R I was able to identify gas within the collection, which was subsequently confirmed on CT performed for interventional purposes. The multiplanar imaging capability of M R I was also found to be of value in selected cases, allowing better definitic~n of the anatomical relationships of pancreatic masses and fluid collec-
tions. M R I was equivalent to, and in one case, better than CT in assessing vascular patency. Although there were no cases of splenic or portal vein thrombosis in this study, current experience indicates this complication would be identified by M R I [14]. The major limitation of M R I at present is availability. The small number of cases in this series is partly due to the difficulty in transferring severely ill and often ventilated patients from one hospital to another for M R imaging. However, once such limitations are overcome, it would appear from the results from this study, that M R I has a valuable place in the investigation of patients with severe acute pancreatitis. Acknowledgements. We are grateful to: Mr M. J. McMahon for referring his patients; Medical Illustration at Leeds General Infirmary; and Mrs Susan Lawrence for typing the manuscript.
REFERENCES
1 Balthazar EJ. CT diagnosis and staging of acute pancreatitis. Radiologic Clinics of North America 1989;27:19-37, 2 Kivisaari L, Somer K, Standertskjold-NardenstamGG, Schroder T, Kivilaasko E, Lempinen M. A new method for diagnosis of acute haemorrhagic-necrotizingpancreatitis using contrast enhanced CT. Gastrointestinal Radiology 1984;9:27 30, 3 Larvin M, Chalmers AG, McMahon MJ. Dynamic contrast enhanced computed tomography: a precisetechnique for identifying and localising pancreatic necrosis. British Medical Journal 1990;300:1425-1428. 4 Balthazar EJ, Robinson DL, Megibow AJ, Ranson JHC. Acute pancreatitis; value of CT in establishing prognosis. Radiology 1990;174:331-336. 5 Johnson CD, Stephens DH, Sarr MG. CT of acute pancreatitis; correlation between lack of contrast enhancement and pancreatic necrosis. American Journal of Roentgenology 1991;156:93-95. 6 Documents of the NRPB. Patient Dose Reduction in Diagnostic Radiology 1990;1:15. 7 VujicI. Vascularcomplicationsofpancreatitis. Radiologic Clinicsof North America 1989;27:81 91. 8 SemelkaRC, SimmFC, Recht MP, DeimlingM, Lenz G, Laub GA. MR imagingof the pancreas at high fieldstrength; comparison of six sequences. Journal of Computer Assisted Tomography 1991;15:966971. 9 Chezmar JL, Nelson RC, Small WC, Bernadino ME. Magnetic resonance imagingof the pancreas with gadolinium-DTPA: Gastrointestinal Radiology 1991;16:139-142. I0 Hamed MM, Harem B, Ibrahim ME, Taupitz M, Mahfouz AE. Dynamic MR imaging of the abdomen with gadopentate dimeglumine: normal enhancement patterns of liver, spleen, stomach and pancreas. American Journal of Roentgenology 1992;158:303 307. 11 Smith FW, BaylissAP, HusseyJK, Robertson EM, Weir J, Crosher GA. Low-field(0.08 T) magneticresonanceimagingof the pancreas; comparison with computed tomography and ultrasound. British Journal of Radiology 1987;62:796 802. 12 Semelka RC, Kroeker MA, Shoenut JP, Kroeker R, Yaffe CS, Micflikier AB. Pancreatic disease: prospective comparison of CT, ERCP and 1.5 T MR imaging with dynamic gadolinium enhancement and fat suppression. Radiology 1991;181:785 791. 13 Lee MJ, Rattner DW, LegemateDA, Saini S, Dawson SL, Hahn PF, Warshaw AL, Mueller PR. Acute complicatedpancreatitis: redefining the role of interventional radiology. Radiology 1992;183:171 174. 14 Ward J, Martinez D, Chalmers AG, Ridgway JP, Robinson PJ, Smith MA. Rapid dynamic contrast enhanced magnetic resonance imaging of the liver and portal vein. British Journal of Radiology 1993;66:214-222.