Table 3.

Comparison of diagnostic accuracies of radiation-free imaging (ultrasonography/magnetic resonance imaging) vs conventional ionising radiation imaging for the evaluation of small bowel Crohn’s disease

Study	Country	Number of patients (n)	Design	Imaging compared	Study findings
Low et al[66], 2000	United States	26 CD	Prospective study, single centre	Contrast enhanced MR with single phase CT using findings from surgery, barium studies, endoscopy and histology as reference standard	Side-by-side comparison: MR imaging superior than helical CT in depiction of normal bowel wall, mural thickening or enhancement and overall GI tract evaluation MR images showed 55 (85%) and 52 (80%) of 65 abnormal bowel segments for the two observers, compared with helical CT which showed 39 (60%) and 43 (65%) of bowel segments affected by CD (P < 0.001, P < 0.05)
Maconi et al[67], 2003	Italy	128 CD	Prospective study, consecutive CD patients who underwent surgery immediately after diagnostic work-up	US, barium studies, CT to detect internal fistulae and intra-abdominal abscesses compared to intraoperative findings	Detecting internal fistula: comparable diagnostic accuracy of US (85.2%) and barium X-ray (84.8%) studies Sensitivity US (71.4%), X-ray (69.6%), Specificity US (95.8%), X-ray (95.8%) Detection of abscesses: US (90.9%), CT (86.4%) Overall diagnostic accuracy higher with CT than US (91.8% vs 86.9%) due to false positives with US
Parente et al[49], 2004	Italy	102 CD	Prospective study, consecutive patients with proven CD by BE and ileocolonoscopy enrolled from IBD clinic Dec 2002-July 2003 Adult cohort (≥ 18 yr)	Conventional US vs oral contrast enhanced US, compared to BE and ileocolonoscopy as gold standard	Per segment analysis: Superior diagnostic accuracy of contrast US in detecting small bowel CD. Sensitivity: conventional US 91.4%, contrast US 96.1% Good correlation of disease extent measurements with BE: US (r = 0.83), contrast US (r = 0.94) Higher sensitivity and specificity with contrast US in detecting ≥ 1 small bowel strictures: Sensitivity: US (74%), contrast US (88.8%) Specificity: US (93.3%), contrast Us (97.3%) US and contrast US more accurate in detecting internal fistulas than BE, but no significant difference in diagnostic accuracy between US and contrast US. US (80%), contrast US (86%), BE (67%) Significantly improved interobserver variability between sonographers with contrast US for detecting bowel wall thickness and disease location
Calabrese et al[55], 2005	Italy	28 CD	Prospective study, consecutive patients recruited from IBD clinic Adult cohort (age range 21-60 yr)	SICUS (performed by a sonologist of 1 yr experience) vs TUS (performed by an experienced sonologist of 10 yr experience), compared to SBE as gold standard	Sensitivity for detection of small bowel lesions: 96% TUS, 100% SICUS Greater correlation of extension of lesions between SICUS and SBE (r = 0.88) vs TUS and SBE (r = 0.64) Sensitivity for detection of ≥ 1 stricture: 76% TUS, 94% SICUS Sensitivity and specificity for assessing prestenotic dilatation: 50% and 100% for TUS, vs 100% and 90% for SICUS
Horsthuis et al[40], 2007	Amsterdam	1735 (sample size 15-440)	Meta-analysis of 33 prospective studies published between Jan 1993- Feb 2006 Adult and paediatric cohort (age range 2-86 yr)	US, MRI, scintigraphy, CT US evaluated in 11 studies, MRI in 11, scintigraphy in 9 and CT in 7 studies	Per-patient analysis: Significantly lower specificity for scintigraphy vs US. No significant difference between mean sensitivities for diagnosis of IBD Sensitivities: 89.7% US, 93% MRI, 87.8% scintigraphy, 84.3% CT Specificities: 95.6% US, 92.8% MR, 84.5% scintigraphy, 95.1% CT Per bowel segment analysis: Significantly lower sensitivity and specificity for CT compared to scintigraphy and MRI. Sensitivities: 73.5% US, 70.4% MRI, 77.3% scintigraphy, 67.4% CT. Specificities: 92.9% US, 94% MRI, 90.3% scintigraphy, 90.2% CT
Lee et al[22], 2009	South Korea	30 CD	Prospective study, single centre, consecutive patients with known or suspected CD enrolled Adult cohort (age range 18-44 yr)	MRE, CT, SBFT for detection of active small bowel inflammation and extra enteric complications with ileocolonoscopy as reference standard	No significant difference between CTE, MRE and SBFT for the detection of active terminal ileitis. Sensitivity CTE (89%), MRE (83%), SBFT (67%-72%) Significantly higher sensitivity for MRE (100%) and CTE (100%) compared to SBFT (32% reader 1, 37% reader 2) for the detection of extra enteric complications
Siddiki et al[68], 2009	United States	33 CD	Prospective blinded study, single centre, consecutive patients with suspected active small bowel CD April 2005-May 2008 Adult cohort (age range 20-63 yr)	MRE, CTE compared with ileocolonoscopy	No significant difference between sensitivity of MRE (90.5%) and CTE (95.2%) in detecting active small bowel CD In 8 cases (24%) MRE and CTE identified active small bowel inflammation not detected at ileocolonoscopy MRE significantly lower image quality score than CTE
Ippolito et al[69], 2009	Italy	29 CD	Prospective study, Single centre, symptomatic patients with proven CD and suspected relapse, recruited from outpatient clinic Adult and paediatric cohort (age range 14-70 yr) Mean age 43.8 yr	Contrast MRE and contrast multi-detector CTE	Complete agreement between MRE and CTE in classification of disease activity (k = 1) Good level of agreement between MRE and CTE for wall thickening and mucosal hyperenhancement (k = 1), comb (k = 0.9) and halo signs (k = 0.86) CTE superior to MRE in detecting fibrofatty proliferation (P = 0.045) MRE depicted higher number of fistulas than CTE but non-significant (P = 0.083)
Schreyer et al[70], 2010	Germany	53 CD	Retrospective study, Single centre, Patients with advanced CD and acute abdominal pain attending the emergency department Adult cohort	Conventional CT, MRE	No significant difference in image quality between CT and MRE No significant difference in diagnosis of small bowel inflammation between CT (69.4%) and MRE (71.4%) CT detection of lymph nodes significantly higher than MRE No significant difference in detection of fistulae (CT n = 25, MRE n = 27) or abscesses (CT n = 32, MRE n = 32)
Panés et al[41], 2011	Spain	N/A	Systematic review of 68 prospective studies, minimum 15 patients per study	US, CT, MRI for diagnosis of CD, assessment of disease extent and activity, detection of complications	Sensitivity for diagnosis of suspected CD and evaluation of disease activity: US 84%, MRI 93% Specificity for diagnosis of suspected CD and evaluation of disease activity: US 92%, MRI 90% CT similar accuracy to MRI for assessment of disease activity and extension. US accuracy lower for disease proximal to terminal ileum US, CT, MRI all high accuracy for detection of fistulas, abscesses, stenosis. US higher false positive for abscesses
Fiorino et al[43], 2011	Italy	44 CD	Prospective study, Single centre, consecutive patients with ileocolonic CD requiring endoscopic or radiological evaluation Enrolled 2006-2009 Adult cohort (> 18 yr) Mean age 44 yr	CTE and MRE to assess disease activity and complications in ileocolonic CD, using ileocolonoscopy as reference standard	MRE significantly superior to CTE in detecting internal strictures: sensitivity (92% vs 85%), accuracy (95% vs 91%), specificity (90% vs 51%) Overall no significant difference in sensitivity and specificity of MRE and CTE in localising CD, bowel wall thickening, bowel wall enhancement, enteroenteric fistulas, detection of abdominal nodes, perivisceral fat enhancement Per segment analysis, MRE significantly superior to CTE in detecting ileal wall enhancement, with higher sensitivity (93% vs 81%) and accuracy (88% vs 81%), but lower specificity (72% vs 81%). MRE significantly superior in localising rectal disease, with higher accuracy (93% vs 85%), specificity (100% vs 50,9%) but lower sensitivity (72% vs 81%)
Jensen et al[71], 2011	Denmark	50 CD	Prospective, multicentre study, patients with symptomatic pre-existing CD requiring small bowel imaging for treatment decisions	MRE and CTE compared with gold standard of ileoscopy or surgery	No significant difference between MRE and CTE for detection of small bowel CD MRE: sensitivity 74%, specificity 80% CTE: sensitivity 83%, specificity 70% No significant difference for detection of small bowel stenosis. MRE: sensitivity 55%, specificity 92%. CTE: sensitivity 70%, specificity 92%
Chatu et al[50], 2012	United Kingdom	143 CD	Retrospective study, single tertiary centre, all symptomatic patients with known or suspected CD who underwent SICUS retrospectively were reviewed June 2007-Dec 2010 Adult cohort Mean age 36 yr	SICUS compared with SBFT, CT, histological findings from ileocolonoscopy or surgery, and CRP, using final diagnosis as the reference standard	Sensitivity of SICUS in detecting active small bowel CD in known or suspected cases 93%, specificity 99%, positive predictive value 98%, negative predictive value 95% Agreement between SICUS with SBFT (k = 0.88), CT (k = 0.91), histological findings (k = 0.62), CRP (k = 0.07)
Pallotta et al[51], 2012	Italy	49 CD	Prospective study, consecutive patients, adult and paediatric CD who underwent resective bowel surgery Jan 2000-Oct 2010 Mean age 37.7 yr (Age range 12-78 yr)	Conventional transabdominal US and SICUS compared to intraoperative and histological findings to assess CD complications	SICUS ability to: Detect at least one stricture: Sensitivity 97.5%, specificity 100%, k = 0.93 Detect two or more strictures: Sensitivity 75%, specificity 100%, k = 0.78 Detect fistulas: Sensitivity 96%, specificity 90.5%, k = 0.88 Detect intra-abdominal abscesses: Sensitivity 100%, specificity 95%, k = 0.89
Qiu et al[44], 2014	China	290 CD	Systematic review with meta-analysis including six studies, all prospective with enrollment of consecutive CD patients	MRE and CTE in detecting active small bowel CD and complications	Pooled sensitivity MRE in detecting active small bowel CD: 87.9%, specificity 81.2% Pooled sensitivity CTE in detecting active small bowel CD 85.8%, specificity 83.6% No significant difference between MRE and CTE in detecting fistula, stenosis and abscesses.
Kumar et al[52], 2015	United Kingdom	67 CD	Retrospective study, Single tertiary centre. Adult cohort (age 18.8-68.9 yr) CD patients requiring resective bowel surgery within 6 mo of SICUS/MRE investigation being performed June 2007-December 2012	SICUS and MRE compared to intraoperative findings	Sensitivity of SICUS and MRE in detecting: Strictures: 87.5%, 100% Fistulae: 87.7%, 66.7 Abscesses: 100%, 100% Bowel dilatation: 100%, 66.7% Bowel wall thickening: 94.7% and 81.8% Compared with surgery, high level of agreement of SICUS, MRE in: Localising strictures: k = 0.75, 0.88 Fistulae: k = 0.82, 0.79 Abscesses k = 0.87, 0.77 High level of agreement between SICUS and MRE in identifying stricturing disease (k = 0.84), number and location of strictures (k = 0.85), fistulae (k = 0.65), mucosal thickening (k = 0.61)
Aloi et al[53], 2015	Italy	25 CD	Single tertiary centre for paediatric IBD Paediatric cohort with known or suspected small bowel CD	MRE, SICUS, CE for diagnosis of small bowel CD	Jejunum: Specificity CE significantly lower (61%) than MRE. No significant difference in sensitivity: SICUS 92%, CE 92%, MRE (75%) Proximal and mid-ileum: Specificity CE significantly lower. No significant difference in sensitivity: MRE 100%, CE 100%, SICUS 80% Terminal ileum: Sensitivity of SICUS and MRE (94%, 94%) higher than CE (81%), CE more specific

CD: Crohn’s disease; MRE: Magnetic resonance enterography; CT: Computed tomography; CTE: Computerised tomography enterography; SICUS: Small intestine contrast-enhanced ultrasonography; CE: Capsule endoscopy; BE: Barium enteroclysis; SBFT: Small bowel follow-through; US: Ultrasonography; MRI: Magnetic resonance imaging.