Between June 2016 and May 2017, a total of 45 consecutive patients with CD who underwent spectral detector-based DECTE were enrolled by using a text search function in our picture archiving and communication system (PACS). Of 45 patients, six were excluded on the basis of the following exclusion criteria: four patients without information regarding CDAI, or in whom CDAI was calculated more than 30 days from the CT examination; one patient whose bowel wall was unable to be evaluated due to remnant positive contrast material in bowel loops; one patient with underlying CD who was diagnosed with concomitant sigmoid colon cancer. Finally, 39 patients with CD were enrolled in this retrospective study (Fig. 1). They were 25 men and 14 women with a mean age of 33.9 years (range, 20–68 years).

Crohn's disease activity flow sheets from electronic medical records of our institution were used to assess CD activity. CDAI scores that had been evaluated within 30 days from the dates of CTE examinations were recorded. Upon examination by experienced gastroenterologists, CDAI score was determined by the sum of eight variables after adjustment with each weighting factor. CDAI scores ranged from 0 to approximately 600 points. On the basis of previous studies, the state of CD activity was subdivided into four groups: remission of disease (CDAI scores less than 150), mild (CDAI scores of 150–219), moderate (CDAI scores of 220–450), and severe (CDAI scores > 450) (21). Thereafter, the patients were divided into two groups: remission (CDAI < 150) and active (CDAI ≥ 150) groups. The assessed CDAI scores were those closest to the examination dates of the DECT. The mean interval between the date of CDAI and the date of DECT examination was 11 days (range, 10–21 days). Assessments of white blood cell count, platelet, hemoglobin, hematocrit, inflammatory markers (e.g., erythrocyte sedimentation rate), C-reactive protein (CRP), and serum levels of total protein and albumin were also performed.

All patients were asked to ingest 1350 mL of a low Hounsfield unit (HU) enteric contrast agent (Easymark, Taejoon Pharmaceutical Co. Ltd., Seoul, Korea) 1 hour prior to CT scanning. All spectral detector-based DECTE examinations were performed by using a spectral detector-based DECT scanner (IQon Spectral CT, Philips Healthcare, Cleveland, OH, USA) with a tube voltage of 120 kVp. Automated tube current modulation (DoseRight, Philips Healthcare) was applied. An intravenous contrast agent (iohexol; Bonorex 350, Central Medical Service, Seoul, Korea) with a total amount of 1.6 mL/kg of body weight was administered at a rate of 3–5 mL/sec for a fixed 30 seconds by using a power injector system. Saline chase was performed at the same rate for 10 seconds. Triphasic CT consisting of precontrast, enteric phase, and portal venous phase was performed for all patients. For the enteric phase, a 35-second delay was used after the aorta reached 100 HU by using the bolus-tracking method. For the portal phase, a fixed 70-second delay was used after contrast administration. Portal-phase images were scanned in the conventional polychromatic mode; enteric-phase images were scanned in the spectral mode, in which spectral-based images (SBIs) can be generated. The SBIs contained the raw data of both detector layers and were used to reconstruct any dual-energy images, such as virtual monoenergetic images or virtual noncontrast images, and to perform spectral analysis, such as iodine concentration calculation.

For enteric phase images, conventional polychromatic images were reconstructed at 120 kVp by using an iterative reconstruction algorithm, iDose4 (Philips Healthcare). In addition, virtual monoenergetic images were reconstructed at 40 keV. All conventional and virtual monoenergetic images were reconstructed on both axial and coronal planes with 3-mm slice thickness and 3-mm reconstruction interval. For the measurement of the iodine concentration, iodine concentration maps were generated by using dedicated software (IntelliSpace Portal, Philips Healthcare) networked with a PACS workstation. Polychromatic enteric phase images reconstructed with iDose4 and iodine concentration maps were used for further image analysis.

CT dose index volume (CTDIvol) and dose length product (DLP) were recorded for each CT examination to analyze the radiation exposure of spectral detector-based DECTE. The effective radiation dose was calculated by multiplying DLP by the conversion factor (0.015 mSv/mGy·cm) (22).

Two radiologists, blinded to the clinical and endoscopic findings, reviewed all DECTE images by consensus. The analyzed CT items were location and number of involved bowels; wall thickness (mm); enhancement pattern of involved bowels; subjective degree of enhancement; presence and degree of mesenteric fat infiltration, comb sign, lymphadenopathy, fistula, and obstruction; and presence of fibrofatty proliferation, mesenteric abscess, or gallstone. The enhancement pattern was divided into three subtypes: single homogeneous, bilaminar, or tri-laminar patterns. The degree of mesenteric fat infiltration was assessed with a 4-point scale: 0, absent; 1, mild; 2, moderate; 3, severe. The degree of comb sign was assessed with a 4-point scale: 0, absent; 1, minimal; 2, segmental dilatation with mild tortuosity; 3, diffuse dilatation with severe tortuosity. The presence of fibrofatty proliferation was evaluated. The degree of lymphadenopathy was graded by using a 4-point scale: 0, absent; 1, regional; 2, distant; 3, diffuse. When lymph nodes enlarged greater than 1 cm in long-axis diameter and were detected only around the involved segment, we considered them as regional lymphadenopathy. If enlarged lymph nodes were found in a distant area not in a regional area, distant lymphadenopathy was recorded. If lymph node enlargements were shown in both regional and distant areas, diffuse lymphadenopathy was considered. The degree of fistula was graded with a 4-point scale: 0, absent; 1, perianal; 2, enteric; 3, both. The degree of obstruction was assessed by using a 4-point scale: 0, absent; 1, partial low-grade obstruction-mild dilatation with occasional air-fluid levels and some colonic gas; 2, partial high-grade obstruction-moderate small bowel dilatation more than air-fluid levels and colonic gas; 3, complete obstruction-marked dilated loops of small bowel with air-fluid levels and little or no colonic gas (23). Fibrofatty proliferation was considered present when there was mesenteric fat proliferation around the involved bowel that resulted in a widening of the interloopal space. The degree of bowel enhancement was assessed both qualitatively and quantitatively. For the qualitative assessment, bowel wall enhancement was graded by using a 4-point scale: 0, normal; 1, minor enhancement, bowel wall attenuation greater than normal small bowel but significantly less than nearby vascular structures; 2, moderate enhancement, bowel wall attenuation increased but somewhat less than nearby vascular structures; 3, marked enhancement, bowel wall signal approached that of nearby vascular structures (24). For the quantitative analysis, two radiologists independently applied ROIs to the diseased bowel wall that showed the strongest enhancement in the most conspicuous part of small and large bowels on iodine concentration maps generated from enteric-phase CT (Fig. 2). Then, they measured attenuation and iodine amount (mg·I/mL) on conventional polychromatic images and iodine quantification maps, respectively. Free-hand ROIs were drawn as large as possible on the diseased bowel to avoid intraluminal air or fluid and perienteric fat. The mean size of ROIs was 5 mm² (range, 3–10 mm²). ROI analysis was also applied to precontrast images by using the “copy and paste” function for the measurement of relative enhancement. Relative enhancement on conventional polychromatic image was calculated by using the formula:

Relative enhancement = [(HU_ent − HU_pre) / HU_pre] × 100, where HU_ent indicates HU on enteric-phase images and HU_pre indicates HU on precontrast images.

Clinical and CT imaging characteristics were evaluated by the Kolmogorov-Smirnov test for normal distribution. Univariate statistical tests were used to evaluate if there was a significant difference of CTE features between remission and active groups. The chi-squared test or Fisher's exact test were used for discontinuous variables; Student's t test was used for continuous variables. In addition, Pearson's correlation test was used analyze inter-observer agreement for the degree of enhancement by two radiologists, as well as to analyze correlations between quantitative DECT parameters and CDAI. Finally, to find the most discriminant CT features, stepwise multiple linear regressions analyses were performed by using CT variables that showed significant correlations with CDAI on univariate analysis. p < 0.05 was considered to indicate statistical significance; a 95% confidence interval was reported for each variable. Statistical analysis used SPSS version 22.0 (IBM Corp., Armonk, NY, USA) and MedCalc for Windows (version 8.0.0.1, MedCalc Software, Mariakerke, Belgium).

Table 1 shows the demographic and clinical findings. Median duration of CD diagnosis was 3.84 years (range, 0.95–8.81 years). Treatments for each patient at the CT examination date were: three on 5-aminosalicyclic acid (5-ASA), 11 on steroids, 5 on an immunosuppressant, 16 on combination treatment of 5-ASA or steroids or immunosuppressant, and 4 on infliximab therapy. Five patients had undergone operations previously: four had undergone right hemicolectomy and one had undergone small bowel segmental resection. The mean CDAI scores were 86.6 with a range of 12.0–305.64. Based on CDAI score, 27 (69.3%) patients were classified as remission group, seven as CD with mild activity, and five as CD with moderate activity. Therefore, 27 patients were placed in the remission group and 12 were in the active group. The mean CRP concentration was 1.60 mg/dL (range, 0.88–2.31 mg/dL).

The mean CTDIvol, DLP, and effective radiation dose for precontrast images were 2.59 mGy (range, 2.38–2.80 mGy), 139.13 mGy·cm (range, 126.11–152.14 mGy·cm), and 2.11 mSv (range, 1.17–3.77 mSv), respectively. The mean CTDIvol, DLP, and effective radiation dose for each enteric and portal venous phase were 9.89 mGy (range, 9.03–10.74 mGy), 534.13 mGy·cm (range, 480.91–587.35 mGy·cm), and 8.00 mSv (range, 1.18–14.37 mSv), respectively. Therefore, the mean effective dose for all three phases was 18.11 mSv (range, 16.45–20.04 mSv).

Table 2 shows the overall DECTE findings in 39 patients. The mean number of lesions and wall thickness were 3.54 mm and 7.45 mm, respectively. Mean relative enhancement and iodine concentration of the lesions were 204.19% and 2.42 mg·I/mL, respectively. The inter-reader agreement for the measurement of quantitative iodine concentration was very good (correlation coefficient = 0.838) (Fig. 3). Approximately 50% (16/39, 41.0%) of the patients showed homogeneous enhancement of the diseased bowel, while the remaining half (23/39, 59.0%) showed a bilaminar pattern of enhancement. Moderate (n = 5) to severe (n = 8) mesenteric fat infiltration was observed in 13 patients. The comb sign was positive in 23 patients (58.9%). In terms of lymphadenopathy, there was no involvement in 17 patients, regional involvement in 17, distant involvement in 3, and diffuse involvement in 2. Fifteen patients had perianal (n = 11), enteric (n = 1), or both (n = 3) fistulas. Partial low-grade (n = 17) or high-grade (n = 3) obstruction was found in 20 patients. None of the 3 patients with high-grade obstruction underwent surgery for the treatment of obstruction. They all recovered from obstruction with conservative medical treatment. Fibrofatty proliferation, mesenteric abscess, and gallstone were present in 18 (46.15%), 5 (12.8%), and 1 (2.6%) patient, respectively. Table 3 shows the results of statistical comparisons between remission and active groups.

Table 3 shows the results of univariate analyses of DECTE findings between remission and active groups. For qualitative CTE findings, there were significant differences between remission and active groups in terms of enhancement pattern (p = 0.012), subjective degree of enhancement (p < 0.001), mesenteric fat infiltration (p = 0.007), comb sign (p < 0.001), and obstruction (p = 0.023). For quantitative CTE features, patients in the active group showed a significantly larger number of lesions (p = 0.002), thicker bowel wall (p = 0.001), stronger relative enhancement (p = 0.004), and higher iodine concentration (p = 0.001) in the involved bowel, compared with those in the remission group. Table 4 shows the results of correlation analyses between quantitative DECTE findings and CDAI. First, there was no significant correlation between CDAI and CRP concentration (p = 0.112). There were significant correlations between CDAI and all quantitative CT parameters, such as numbers of lesions (correlation coefficient r = 0.573, p < 0.001), bowel wall thickness (r = 0.477, p = 0.002), relative degree of enhancement (r = 0.541, p < 0.001), and iodine concentration (r = 0.744, p < 0.001). Representative examples are shown in Figures 4,5,6,7.

Multivariate linear regression analyses included the following variables: number of lesions, bowel wall thickness, pattern of bowel wall enhancement, degree of enhancement, mesenteric fat infiltration, comb sign, fistula, relative degree of enhancement, and iodine concentration. Among these variables, iodine concentration measured on the iodine map of spectral detector-based DECTE remained the sole independent variable associated with CDAI after adjustment for other confounders in multivariate analysis (p = 0.001). The linear regression equation for CDAI (y) and iodine concentration (x) was y = 53.549x + 55.111, with R² = 0.726.