MRI of small bowel Crohn's disease: determining the ... - Springer Link

Eur Radiol (2009) 19: 1960–1967 DOI 10.1007/s00330-009-1371-0

A. Sharman I. A. Zealley R. Greenhalgh P. Bassett S. A. Taylor

Received: 18 November 2008 Revised: 11 February 2009 Accepted: 14 February 2009 Published online: 24 March 2009 # European Society of Radiology 2009

A. Sharman . I. A. Zealley Ninewells Hospital, Dundee, UK R. Greenhalgh . S. A. Taylor (*) Department of Imaging, University College Hospital, 2F Podium, 235 Euston Road, London, NW1 2BU, UK e-mail: [email protected] Tel.: +44-207-3809300 Fax: +44-20-76915752 P. Bassett Stats Consultancy, Ruislip, UK

GASTRO INTESTINAL

MRI of small bowel Crohn’s disease: determining the reproducibility of bowel wall gadolinium enhancement measurements

Abstract This study aims to determine inter- and intra-observer variation in MRI measurements of relative bowel wall signal intensity (SI) in Crohn’s disease. Twenty-one small bowel MRI examinations (11 male, mean age 40), including T1-weighted acquisitions acquired 30 to 120s postgadolinium, were analysed. Maximal bowel wall SI (most avid, conspicuous contrast enhancement) in designated diseased segments was measured by two radiologists and two trainees using self-positioned “free” regions of interest (ROIs) followed by “fixed” ROIs chosen by one radiologist, and this procedure was repeated 1 month later. Relative enhancement (postcontrast SI minus pre-contrast SI/precontrast SI) was calculated. Data were analysed using Bland–Altman limits of agreement and intra-class correlation. Inter-observer agreement for relative enhancement was poor (span-

Introduction Magnetic resonance imaging (MRI) is an emerging technique for the evaluation of small bowel Crohn’s disease. Advantages over conventional barium studies include the ability to detect extra-luminal complications as well as the absence of ionising radiation [1–4]. It is believed that strictures of a more “acute” inflammatory character have the potential to respond to anti-inflammatory therapy, whereas diseased segments dominated by “chronic” inflammatory changes, including fibrosis, will not [1]. Any test, which could reliably predict whether a diseased

ning over 120%) using a free ROI— 95% limits of agreement −0.69, 0.70 and −0.47, 0.74 for radiologists and trainees, respectively, only marginally improved by use of a fixed ROI −0.60, 0.67 and −0.59, 0.49. Intra-class correlation ranged from 0.46 to 0.72. Intra-observer agreement was slightly better and optimised using a fixed ROI—95% limits of agreement −0.52, 0.50 and −0.34, 0.28 for radiologists and trainees, respectively. Intra-class correlation ranged from 0.49 to 0.86. Relative bowel wall signal intensity measurements demonstrate wide limits of observer agreement, unrelated to reader experience but improved using fixed ROIs. Keywords Magnetic resonance imaging . Crohn disease . Observer variation

segment would or would not respond to high-potency antinflammatory medical therapies before treatment, could have a major clinical impact. It has been suggested that the degree of intravenous contrast enhancement during MRI correlates with the degree severity of activity [3, 4] and that this is due to increased perfusion and capillary permeability [4, 5]. Previous studies have reported a close relationship between the degree of bowel-wall enhancement on MRI and the degree of clinical inflammatory activity [6–9]. Signal intensity (SI) ratios (before and after enhancement) in diseased versus non-diseased bowel of

1961

greater than 1.15:1 [10] and 1.3:1 [11] have been reported to be useful for identifying active inflammation: the type of disease that might be expected to respond to anti-inflammatory therapy. It is known from the literature that inter-and intraobserver variation in functional computed tomography (CT) measurements of tumour perfusion is sufficiently great to question the reliability of such data in clinical practice [12]. There is no reason to believe that simpler single-phase measurements of enteric perfusion using MRI are immune from these observer variations. The purpose of this study was to assess the level of interand intra-observer variation in MRI measurements of contrast-enhanced bowel-wall SI in patients with known Crohn’s disease, and to assess the effect of guided region of interest placement, reader experience and software viewing platform.

patient received intravenous hysocine-N-butylbromide (20 mg) as an anti-peristaltic agent before MRI. The examination protocol included T1-weighted (T1w) volume-acquisitions performed before and after a handinjected intravenous bolus of 15 ml gadolinium chelate. (Omniscan, Gadodiamide, Amersham Health), followed by a 20-ml saline chaser injected over 10 s. The image sequences employed were coronal and axial T1w fat saturated volume-interpolated breath-hold (VIBE) sequences and acquisition parameters (summarised in Table 1) were tailored to individual patients to ensure whole abdomen coverage in 20– to 22-s breath-holds for each acquisition. Axial and coronal acquisitions were obtained pre-contrast and were followed by coronal, axial, coronal and axial acquisitions at 30,60,90 and 120 s after administration of intravenous contrast medium. Image review platforms

Materials and methods The study design was registered and approved by the local research and development office and the study was performed under waiver from the local ethical review board.

Unless otherwise stated (see below) image analysis was performed on workstations running Merge eFilm (Merge Healthcare,Toronto, Canada). The matrix size and actual resolution of the monitors on which the examinations were studied was 1,500×2,000 (3 megapixels).

Subjects

Readers

Small bowel MRI examinations were retrospectively selected from a cohort of 84 consecutive patients, who had attended for assessment as part of routine clinical care in our institution between January 2005 and January 2006. The study coordinator reviewed the medical records and imaging reports for the cohort and selected those with an established diagnosis of Crohn’s disease, and at least one diseased segment depicted by MRI, for inclusion in the study. Patients were excluded if there was no firm diagnosis of Crohn’s disease, or MRI had shown no reported abnormality. All patients had apparently normal renal function demonstrated by serum biochemical markers. The final study population comprised 21 patients: 11 male, mean age 38 years (range 17–58, SD 14.9), ten female, mean age 43 years (range 23–70, SD 17.4).

Four readers took part in the study. Two radiologists with 3 and 2 years’ experience, respectively, of reporting small bowel MRI (radiologist 1 and radiologist 2), and two trainee radiologists who had received training in general abdominal MRI but who had no specific experience of small bowel MRI (trainee1 and trainee 2).

MRI technique All of the MR examinations were performed on a 1.5-T Siemens Avanto unit (Siemens, Erlangen, Germany) using two multi-channel phased-array body coils. In order to distend the small bowel, all patients ingested up to 1.5 l of a high-osmolar water-based bowel preparation (0.2% locust bean gum and 2% mannitol solution) over 45 min before MRI [10]. Patients were placed in the supine position and a standard MRI small bowel protocol performed. Each

Table 1 Summary of MRI technical parameters for T1w fatsaturated volume-interpolated breath-hold (VIBE) sequences MRI factor

Coronal plane

Axial plane

Matrix FOV FOV phase Partitions per 3D acquisition Slice thickness Intersection gap Phase resolution Slice resolution Bandwidth GRAPPA TR TE Flip angle

256×256 480×480 mm 100% 48 3 mm 20% 60% 64% 300 Hz/pixel Factor 2 7.22 ms 2.21 ms 10 degrees

256×256 430×430 mm 56% 88 3 mm 20% 60% 65% 300 Hz/pixel Factor 2 7.22 ms 2.25 ms 10 degrees

1962

ing the terminal ileum), and terminal ileum (defined as the final 15 cm of small bowel)], together with any adjacent anatomical landmarks (for example, terminal ileum adjacent to bladder dome, or proximal jejunum, left upper quadrant, etc). Image snap shots illustrating the selected bowel segment were specifically not used, so as not to bias the reader in their subsequent choice of free ROI placement (see below). During the reading protocol (described below) readers were instructed to use either a “free” or “fixed” ROI within the nominated disease segment. 1. Free ROI

Fig. 1 Coronal VIBE fat saturated sequence demonstrating a segment of diseased small bowel with enhancement following intravenous contrast injection with an ROI marked

Identification of diseased bowel segments and region of interest (ROI) placement For each patient, the study coordinator (an abdominal radiologist) reviewed the full MRI dataset and located the abnormal small bowel segment which was to be utilised in the study. In patients in whom there was more than one disease site, the segment of diseased small bowel which was subjectively considered to exhibit the most avid, conspicuous contrast enhancement was selected. The coordinator then documented the location of the selected bowel segment, detailing the anatomical segment [proximal jejunum, distal jejunum, proximal ileum, distal ileum (exclud-

Fig. 2 Coronal VIBE fat saturated sequence demonstrating multiple ROIs on a section of enhanced small bowel wall following intravenous contrast injection (e-Film platform) with SI measurements ranging from 407 to 553

Readers were instructed to measure mural SI by placing the largest possible ROI over the area in the nominated diseased segment which the reader perceived to exhibit the maximal bowel wall SI (i.e. most avid, conspicuous contrast enhancement) (Figs. 1, 2). Care was taken to exclude both adjacent mesenteric fat and the small bowel lumen. Readers were encouraged to magnify the image to aid ROI placement. ROI placement was individually performed for each time-point (pre-contrast, and 30, 60, 90 and 120 s post contrast medium) with each reader attempting to keep their ROI placement constant across the image sequences. For enhanced images in which a layered enhancement pattern was observed (Fig. 3), readers placed their ROI over the area which they perceived as exhibiting the greatest degree of enhancement regardless of whether this was on the mucosal or serosal side of the bowel wall. 2. Fixed ROI The study coordinator took an image snapshot stored as a jpeg, illustrating the slice number and exact location of the first “free” ROI placed by radiologist 1 for each timepoint and patient. These image snapshots were provided to readers before the fixed ROI section of the reading protocol, detailed below. Readers were instructed to

Fig. 3 Axial VIBE fat saturated sequence demonstrating serosal (short arrow) and mucosal (long arrow) small bowel wall enhancement following intravenous contrast injection

1963

exactly reproduce this ROI as far as possible for their own measurement of mural SI. Reading protocol 1. Inter-observer agreement (relative contrast enhancement) Firstly, all four readers were provided with the study proforma detailing the target small bowel segments and independently recorded mural SI using first free and then fixed ROIs at each time-point for each patient within the nominated diseased segment. Readers were blinded to the position of the fixed ROI until after they had completed their free ROI measurements. 2. Intra-observer agreement (relative contrast enhancement)—within software viewing platform One month after their first readings (to reduce recall bias), two readers (radiologist 1 and trainee 1) repeated their readings (again guided by a study proforma) using first free and then the fixed ROI. 3. Intra-observer agreement (relative contrast enhancement)—between software platforms To test the potential influence of viewing software, one month after their readings using the standard eFilm viewing platform, 2 readers (radiologist 2 and trainee 2) repeated the fixed ROI measurements using a second viewing software (Agfa UK, Agfa-Gevaert Group, Brentford, West London, UK ). Calculation of relative mural signal enhancement All ROI measurements obtained by readers at each timepoint were converted to the more clinically relevant “relative mural enhancement” using the equation relative mural enhancement = (post-contrast mural SI – precontrast mural SI)/pre-contrast mural SI Statistical analysis StatsDirect (Liverpool, UK) software was used for the analysis. Inter- and intra-observer agreement for relative mural enhancement measurements were calculated using the Bland-Altman method [13] and expressed as the mean differences, SD of the differences, and the 95% limits of agreement [mean difference ± (1.96×SD of the differences)]. In particular, inter-observer agreement (for each time-point, and overall) was calculated between the two radiologists and between the two trainees, for both the free and fixed ROIs. Intra-observer agreement (combining all time-points and for both free and fixed ROIs) was

calculated for both radiologist 1 and trainee 1 on the Efilm platform and for both radiologist 2 and trainee 2 (fixed ROI) across the two software platforms. In addition, the intra-class coefficient of agreement was calculated for the same inter-, and intra-observer comparisons for the data overall. In order to determine the time of maximum enhancement, the mean fixed ROI measurement for each time-point (30, 60, 90 and 120 s after i.v. contrast administration) was calculated across all readers and analysed across time-points using repeated measures ANOVA (analysis of variance).

Results Inter-observer variability Overall 95% limits of agreement were a little narrower using the fixed ROI than when using the free ROI, but even then were relatively wide (spanning around 120%) (Table 2). The data show that overall a difference of up to 70% between observer measurements of relative mural signal change is within the expected 95% limits of agreement. In other words, 95% of all repeated measurements of bowel wall enhancement would be expected to vary by as much as ±70%. Figure 4a, b shows the BlandAltman plots of agreement for experienced readers using free vs fixed ROIs. The clustering of the data-points provides insight into the distribution of agreement levels throughout the experiment. It can be seen that for both ROIs (free and fixed), the difference between observer measurements of relative mural signal change was no more than 40% (0.4) for most measurements but there were four (free ROI) and two (fixed ROI) measurements in which the difference was over 100%. The intra-class correlation coefficients were also a little better with the fixed ROI than the free (Table 2) but reached 0.72 at best (fixed ROI for trainees). There was little difference in the 95% limits of agreement or intra-class correlation coefficients between radiologist and trainee readers, suggesting prior experience of small bowel MRI did not hold any particular advantage (indeed agreement was arguably a little better between the trainees). Furthermore, there was no clear pattern in interobserver variation depending on the time post gadolinium injection. Intra-observer agreement (within software viewing platform) For the radiologist reader, the levels of intra-observer agreement were not much different from the overall level of inter-observer agreement (Tables 2, 3) for both methods (free and fixed ROI), although slightly superior using the fixed ROI method. However, for the trainee reader intra-

1964

Intra-class coefficient

Difference in measured change in Mural SI (Radiologist 1 – Radiologist 2) 2

95% Bland-Altman limits of agreement

(−0.61, (−0.41, (−0.68, (−0.76, (−0.62, (−0.51, (−0.50, (−0.68, (−0.60, (−0.59,

0

SD of difference

0.33 0.17 0.35 0.35 0.37 0.26 0.25 0.31 0.32 0.28

0.60 0.72

a

-1

0.68) 0.30) 0.68) 0.61) 0.82) 0.50) 0.47) 0.54) 0.67) 0.49)

1

± 95% limits of agreement

1.0

1.5

2.0

Difference in measured change in Mural SI (Radiologist 1 – Radiologist 2) 1.8

b

0.04 0.06 0.02 0.07 0.1 0.00 −0.01 0.07 0.03 −0.05


Mean difference between measurements of relative change in mural SIa Fixed ROI

0.5

Mean change in mural SI ((Radiologist 1 + Radiologist2) / 2)

1.3


0.51 0.46

0.8

(Post-contrast mural SI – pre-contrast mural SI)/pre-contrast mural SI

-0.2

a

Overall

120

90

60

Free ROI

30

Radiologist Trainee Radiologist Trainee Radiologist Trainee Radiologist Trainee Radiologist Trainee

0.01 0.13 0.02 0.18 0.06 0.12 −0.04 0.10 0.01 0.13

Mean difference between measurements of relative change in mural SIa

0.30 0.31 0.33 0.31 0.46 0.32 0.31 0.31 0.35 0.31

SD of difference

(−0.61, (−0.48, (−0.63, (−0.43, (−0.84, (−0.50, (−0.66, (−0.50, (−0.69, (−0.47,

0.58) 0.74) 0.67) 0.79) 0.96) 0.74) 0.57) 0.71) 0.70) 0.74)


0.3

Reader Time post gadolinium injection (s)

Table 2 Inter-observer variability in mural SI measurements according to ROI method, reader and time point

0.0

-0.7 0.00

0.75

1.50

2.25

Mean change in Mural SI ((Radiologist 1 + Radiologist2)/ 2)

Fig. 4 a Bland-Altman plot of agreement between the radiologist 1 and radiologist 2 using the free ROI method. Note there is no clear evidence of any relationship between the level of agreement and the magnitude of the change in relative mural SI. b Bland-Altman plot of agreement between the radiologist 1 and radiologist 2 using the fixed ROI method. Note that the 95% limits of agreement are a little narrower when compared with those obtained using the free ROI method (a)

observer agreement was in general better than interobserver agreement, particularly for the fixed ROI method (95% limits of agreement spanning around 60%, intra-class correlation coefficient 0.85). Figure 5a, b shows the Bland-Altman plots for the inexperienced readers using the free and the fixed ROI methods. Again, for both methods (free and fixed ROI), the difference between the observer’s measurements of relative mural signal change was no more than 40% for most measurements. However, there were four (free ROI) and three (fixed ROI) measurements in which the difference was over 50% (Table 3).

0.49 0.79

a ± 95% limits of agreement

(−0.73, 0.74) −0.34, 0.28

0.2

-0.2

-0.6

0.26 0.16

-1.0 0.0

0.4

0.8

1.2

1.6

Mean change in Mural SI (Trainee 1 first read + Trainee 1 second read) / 2)

Difference in measured change in Mural SI (Trainee 1 first read - Trainee 1 second read) 0.45

b


(Post-contrast mural SI – pre-contrast mural SI)/pre-contrast mural SI, all time-points combined

−0.52, 0.50 (−0.33, 0.42) 0.37 0.19 0.00 0.04 Radiologist Trainee

-0.05

a

−0.01 −0.03

0.20

0.86 0.85

SD of difference Mean difference between measurements of relative change in mural SIa Free ROI

0.6


Intra- class coefficient

Difference in measured change in Mural SI (Trainee 1 first read - Trainee 1 second read)

SD of difference Mean difference between measurements of relative change in mural SIa Fixed ROI

Table 3 Intra-observer variability in mural SI measurements using a single software viewing platform according to ROI and reader



1965

-0.30

-0.55

-0.80 0.2

0.7

1.2

1.7

Mean change in Mural SI ((Trainee 1 first read + Trainee 1 second read) / 2)

Fig. 5 a Bland-Altman plot showing intra-observer agreement for trainee 1 using the free ROI method. Note there is no evidence of any relationship between the level of agreement and the magnitude of the change in relative mural SI. b Bland-Atman plot showing intra-observer agreement for trainee 1 using the fixed ROI method. Note that the 95% limits of agreement are marginally narrower when compared with those obtained using the free ROI method (a)

Intra-observer agreement (between software platforms) Using the fixed ROI method the level of intra-observer agreement and intra class correlation coefficients between measurements taken on different software platforms was of similar magnitude to the level of intra-observer agreement using a single viewing platform (Table 3), particularly for the radiologist reader (Table 4). Determination of maximal contrast enhancement The mean measured relative change mural SI differed significantly according to time post gadolinium injection

1966

Table 4 Intra-observer across two software viewing platforms (using the fixed ROI method) according to reader Reader experience

Mean difference between measurements of relative change in mural SI made on Efilm and Agfa viewing platformsa

SD of difference



Radiologist Trainee

0.03 −0.04

0.14 0.36

(−0.32, 0.26) (−0.75, 0.67)

0.86 0.56

a

(Post-contrast mural SI – pre-contrast mural SI)/pre-contrast mural SI, all time-points combined

(p=0.002), with the peak being after 60 s. At 60 s post contrast medium, the mean mural SI across all patients using the fixed ROI data (averaged over the four readers) was 497 units (Table 5).

Discussion In order to be clinically useful, MRI measurements of mural enhancement in Crohn’s disease must be shown to be robust both between and within individual observers. To our knowledge, although other workers have assessed this in sub analyses of their main study, ours is the first study to specifically address this issue. We found wide inter- and intra-observer limits of agreement within the expected measurement range. Using the free ROI method, which simulates real clinical practice, 95% of measurements would be expected to fall within a range of around ±70% difference between different readers. This is a clinically important observation because the average maximum SI of the bowel wall was just 497 units 60 s post contrast. If measures of bowel wall enhancement are to be a useful tool in the assessment of Crohn’s activity then in order to be detected reliably (1) the difference between diseased segments which will and will not respond to anti-inflammatory therapy must be very large, and (2) any response to therapy must be accompanied by relatively large changes in bowel wall enhancement. We found no evidence that prior experience of small bowel MRI conferred any advantage with regard to reproducibility of bowel-wall enhancement measurements when compared with the performance of trainees. Table 5 Mean mural signal (fixed ROI) across all readers according to time post gadolinium injection Time post gadolinium injection (s)

Mean of measured relative change in mural SI (SD)a

p valueb

30 60 90 120

0.82 0.90 0.82 0.84

0.002

(0.35) (0.4) (0.4) (0.31)

(Post-contrast mural SI – pre-contrast mural SI)/pre-contrast mural SI, all time-points and readers combined b Repeated measures ANOVA a

Unsurprisingly, both inter- and intra-observer variation was improved by using the fixed ROI method, where readers attempted to precisely copy a pre-determined ROI. This indicates that if measures of bowel-wall enhancement are to be used in a clinical setting then perhaps good practice would be for a snapshot of an initial ROI position to be stored and used to guide comparative measurements on subsequent examinations for that patient, ideally to be performed by the same operator. However, even in this instance our results indicate that 95% of repeated measurements would be expected to fall within a range of around ±40% due to intra-observer variation alone. Reassuringly, we did not find use of a different softwareviewing platform had any major influence on observer agreement. Nevertheless it would seem sensible to use the same viewing platform whenever possible when comparing across patients or during patient follow-up. Several factors contribute to the wide variability between measurements. Even those measurements made by readers attempting to replicate the ROI position on a guide image were still subject to an only slightly improved degree of reproducibility compared with those derived from unguided, independent ROI positioning. Contributory factors include the heterogeneous layering enhancement of the bowel wall (albeit a relatively uncommon enhancement pattern) and the small diameter of both the bowel wall and subsequent ROI, both leading to increased variability. Our findings are surprising when we consider those studies have suggested using specific contrast enhancement ratio thresholds to indicate “acute” inflammation, and perhaps cast doubt on the reproducibility of such data in clinical practice. For example, variability of ±70% makes it unlikely that SI ratios as low as 1.15–1.3:1, as reported in previous studies [10, 11] could be used to reliably distinguish between active and inactive disease. Of course, in reality there are many other factors that must be taken into consideration when assessing disease activity, other than contrast enhancement. For example, wall thickening, increased mural signal, perienteric fluid, prominent mesenteric vessels, lymphadenopathy all point to active disease. Contrast enhancement is rarely used in isolation due the known limits in accuracy. Our study has several limitations. We used two “experienced” (radiologists) and two “inexperienced” (trainees) observers and accept this may not be representative of readers in general. Importantly, we used “relative

1967

mural enhancement” as a surrogate measure of mural perfusion. In reality quantitative assessment of perfusion is much more complex, and must take into consideration both image noise (to calculate signal to noise ratio) and a reference tissue [14]. It is, however, very likely that the major source of observer variation will be in the measurement of mural SI, and it is doubtful that asking our readers to perform relatively complex and timeconsuming calculations would have changed the overall message of our study. We asked readers to record data from a single ROI rather than use an average of more than one as advocated by others workers [6, 11]. By using the ROI placements of one experienced reader as a template for the fixed ROI method readings, we potentially biased results in favour of this reader. However, the fact that intra-observer variation for this reader was no better than the others reassuringly points against any significant bias from this source.

In conclusion, in patients with Crohn’s disease, measurements of bowel wall SI at small bowel MRI are subject to wide limits of both inter- and intra-reader agreement, which may substantially limit their utility when applied to the development of quantitative measures of inflammatory activity in diseased bowel segments. Routine use of a fixed ROI by a single observer may obviate some of the variability in patients on active treatment being monitored using MRI. We recommend caution be applied when considering the use of intravenous contrast enhancement as a quantitative indicator of disease activity in patients with small bowel Crohn’s disease. Acknowledgements This work was undertaken at UCLH/UCL and a proportion of funding was received from the Department of Health’s NIHR Biomedical Research Center’s funding scheme

References 1. Masselli G, Brizi GM, Parrella A, Minordi LM, Vecchioli A, Marano P (2004) Crohn disease: magnetic resonance enteroclysis. Abdom Imaging 29:326–324 2. Gourtsoyiannis N, Papanikolaou N, Grammatikakis J, Maris T, Prassopoulos P (2000) MR imaging of the small bowel with a true-FISP sequence after enteroclysis with water solution. Invest Radiol 35:707–711 3. Schunk K, Kern A, Oberholzer K et al (2000) Hydro-MRI in Crohn’s disease: appraisal of disease activity. Invest Radiol 35:431–437 4. Horsthuis K, Lavini Mphil C, Stoker J (2005) MRI in Crohn’s disease. J Magn Reson Imaging 22:1–12 5. van Oostayen JA, Wasser MNJM, van Hogezand RA, Griffioen G et al (1997) Doppler sonography evaluation of superior mesenteric artery flow to assess Crohn’s disease activity: correlation with clinical evaluation, Crohn’s disease activity index, and alpha(1)-antitrypsin clearance in feces. AJR Am J Roentgenol 168:429–33

6. Sempere GA, Martinez Sanjuan V, Medina Chulia E et al (2005) MRI evaluation of inflammatory activity in Crohn’s disease. AJR Am J Roentgenol 184:1829–1835 7. Marcos HB, Semelka RC (2002) Evaluation of Crohn’s disease using half-Fourier RARE and gadoliniumenhanced SGE sequences: initial results. Magn Reson Imaging 18:263–268 8. Florie J, Wasser MN, Arts-Cieslik K, Akkerman EM, Siersema PD, Stoker J (2006) Dynamic contrast-enhanced MRI of the bowel wall for assessment of disease activity in Crohn’s disease. AJR Am J Roentgenol 186:1384–1392 9. Laghi A, Borrelli O, Paolantonio P, Dito L et al (2003) Contrast enhanced magnetic resonance imaging of the terminal ileum in children with Crohn’s disease. Gut 52:393–397 10. Madsen SM, Thomsen HS, Munkholm P, Dorph S, Schlichting P (1998) Active Crohn’s disease and ulcerative colitis evaluated by low-field magnetic resonance imaging. Scand J Gastroenterol 33:1193–2000

11. Koh DM, Miao Y, Chinn RJ et al (2001) MR imaging evaluation of the activity of Crohn’s disease. AJR Am J Roentgenol 177:1325–1332 12. Goh V, Halligan S, Hugill J, Bassett P, Bartram CI (2005) Quantitative assessment of colorectal cancer perfusion using MDCT: inter- and intraobserver agreement. AJR Am J Roentgenol 185:225–231 13. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307–310 14. Semelka RC, Shoenut JP, Silverman R, Kroeker MA, Yaffe CS, Micflikier AB (1991) Bowel disease: prospective comparison of CT and 1.5-T pre- and postcontrast MR imaging with T1weighted fat-suppressed and breathhold FLASH sequences. J Magn Reson Imaging 1:625–32