Introduction

Crohn’s disease (CD) has become a challenge to global public health. After 1990, the incidence rate of CD in newly industrialized countries in Africa, Asia and South America rose sharply, while in western countries, although the incidence rate tended to be stable, it remained at a relatively high level of 0.3%1.

CD, as a non-specific inflammatory bowel disease that can affect the entire digestive tract, often alternates between recurrence and remission and is often accompanied by more serious complications (including intestinal obstruction, peri-intestinal abscess, and intestinal fistula). However, the current diagnosis of CD lacks a gold standard and requires comprehensive analysis and close follow-up of clinical manifestations, imaging findings, laboratory indicators, endoscopic examination, and histopathological results2,3. In addition, the condition of CD is complex, variable, and difficult to cure. In clinical practice, the main treatment goals are to maintain remission and achieve mucosal healing4. Its treatment plans are diverse, including immunosuppressants, thalidomide, biologics, exclusive enteral nutrition, fecal microbiota transplantation, endoscopic therapy, and surgery. The formulation and adjustment of treatment plans are mainly based on changes in disease activity5. Therefore, seeking a convenient and fast way to diagnose CD early and accurately evaluate its activity is crucial to the diagnosis and treatment of CD.

Some guidelines recommend using CT enterography (CTE) or MR enterography (MRE) to examine CD patients who are suspected or newly diagnosed, in order to evaluate the extent of the lesion and its complications. Compared to MRE, CTE has higher spatial resolution, higher consistency among radiologists, and is more pronounced in detecting intestinal obstruction and fistula6,7,8. Therefore, CTE has also become a commonly used CD evaluation method in clinical practice. Dual-energy CTE (DE-CTE) is based on the analysis of high- and low-energy data, and can provide multiple quantitative parameters. Previous studies have shown that DE-CTE’s virtual monoenergetic images (VMI) could improve intestinal wall contrast9. Moreover, monoenergetic CT values, iodine concentration, fat quantification, and effective atomic number have potential applications in the differential diagnosis of CD10, activity evaluation11,12, and efficacy evaluation13. For the value of the slope of spectral curves in CD, its application has not been fully studied, although recent studies showed some preliminary results in differentiations between activity and inactivity14,15. It may be due to the fact that commercially available software only provides monoenergetic images, material decomposition images, and effective atomic number images, while the acquisition of the slope for the energy spectrum curve requires regions of interest (ROIs) delineations on two monoenergetic images and formula calculation, which is not convenient enough.

Therefore, the purpose of this study is to explore the application value of a slope parameter image (SPI) for spectral curves based on dual-phase DE-CTE images for CD diagnosis and inflammatory activity assessment.

Materials and methods

Patients

This retrospective study was approved by the hospital ethics committee and included patients who were clinically diagnosed with CD and underwent DE-CTE and colonoscopy at Xiamen University Affiliated Zhongshan Hospital from January 2019 to March 2022. Exclusion criteria:1 Cases with respiratory and motion artifacts that interfere disease diagnosis;2 Colonoscopy and DE-CTE examination showed mismatched lesion locations;3 Colonoscopy and DE-CTE examination with a time interval greater than 2 weeks;4 History of gastrointestinal surgery;5 Serious complications such as internal fistula;6 Incomplete clinical data;7 Individuals under the age of 18. The control group included patients who underwent DE-CTE due to suspected gastrointestinal diseases, but were ultimately diagnosed without intestinal diseases. Exclusion criteria:1 Cases with respiratory and motion artifacts that interfere disease diagnosis;2 History of gastrointestinal surgery;3 Incomplete clinical data;4 Individuals under the age of 18. (Fig. 1)

Fig. 1
figure 1

Flowchart for patient enrollment.

Full size image

DE-CTE protocol

The preparation and patient scanning were performed followed previous guidelines16,17,18,19. Patients fastened for 6 h and cleaned the intestines before the CT examination. Oral isotonic mannitol solution (2.5%): 500 ml/15min × 4 times 1 h before scanning; intravenous injection of 20 mg of scopolamine 15 min before scanning. Patients were scanned using a 256-row Revolution CT (GE HealthCare, USA) for DE-CTE. Patients were placed in a supine position with breath holding, with a scanning range from the diaphragmatic crest to the pubic symphysis. The scanning was performed in GSI mode, with a fast-switching voltage of 80/140 kVp, a tube current of 250 mA, a layer thickness and spacing of 0.625 mm, and a pitch of 0.992:1. Reconstruction: a layer thickness and spacing of 1.25 mm, a reconstruction algorithm of adaptive statistical iterative reconstruction (Veo) with a weight of 50%. The iodine contrast agent (iodoprolol, 370 mgI/ml, 1.5-2.0 mL/kg) was injected through the elbow vein at a rate of 3–4 ml/s. The scanning adopted an intelligent bolus triggering technique, with a threshold of 100 HU setting at the abdominal aorta. When the threshold was achieved, the arterial phase (AP) scanning is performed after a delay of 8 s, and after a delay of 30 s, the portal vein phase (PVP) scanning is performed.

SPI generation

Transferred the AP and PVP image data to the post-processing workstation (GE HealthCare, Advanced Workstation 4.7), and reconstructed VMI40–100 with every 10 keV interval within the range of 40–100 keV, while also reconstructing 120 kVp-like images. The dicom. format data was then batch converted into nii.gz format using image processing software (SimpleITK20). An algorithm based on a slope calculation formula written in Python was also used to generate the SPI of the 40–100 keV energy spectrum curve (SPI40-100), with VMI40 and VMI100 inputs (Fig. 2). The slope calculation formula was: the slope parameter (SP) of the energy spectrum curve between 40 and 100 keV (SP40–100) = (HU40keV–HU100keV)/60.

Fig. 2
figure 2

The Portal-vein-phased SPI40–100 (a) and the colonoscopy image (b) of a CD patient. The arrow on figure a pointed to the affected intestinal bowel segment. (b) illustrates a colonoscopy image of the distal ileum, showing multiple longitudinal ulceration of the mucosa.

Full size image

Endoscopy

Two senior gastroenterologists with over ten and twenty years of experience in endoscopic procedures performed endoscopic examinations on CD patients, and used Simplified Endoscopic Activity Score (SES-CD) to jointly evaluate the severity of endoscopic activity21 (Table 1). After reaching a consensus, the final score was obtained. The experiment divided the ileum into the terminal segment, right colon, transverse colon, left colon, and rectum. For multiple lesions, the most severe lesion was used as the ROI. The range of lesions and intestinal stenosis were evaluated, with scores ranging from 0 to 3 for each evaluation item. The total score for a single intestinal segment was between 0 and 12 points. 0–6 points are divided into non-moderate-to-severe inflammatory activities, and 7–12 points are divided into inflammatory moderate-to-severe activity groups22.

Table 1 Definitions of simple endoscopic score for Crohn’s disease.
Full size table

DE-CTE data measurements

All measurements of spectral quantitative parameters were conducted using the ITK-SNAP software23. ROI were drawn and measured on VMI40 ~ VMI100 and SPI40–100 images. For the CD group, ROIs were placed on the diseased intestinal wall and adjacent creeping fat. For the control group, ROIs were placed on the normal intestinal wall and adjacent mesenteric fat. All ROIs were selected to avoid adjacent blood vessels, mesenteric fat, and intestinal contents. The normal intestinal wall ROI of the control group was placed at the end of the ileum, while the lesion ROIs of the CD group were placed at the most significantly enhanced area, corresponding to the most severe lesion under endoscopy. The ROI of the stratified enhanced lesion was placed in the mucosal layer, while for uniformly enhanced lesions, the ROI was placed in the center of the intestinal wall24 (Fig. 3). The creeping fat ROI in the CD group was located within 5 mm of the lesion wall at the most prominent location of the mesenteric lesion, as much as possible along mesenteric vessels such as the ileocolic artery and right colon artery. The ROI of mesenteric fat in the control group was selected from the mesenteric margin fat within a range of 5 mm around the end of the ileum. The size of the area of interest should be controlled within 3–6 mm². ROI selection and measurement were repeated three times, and the average values were taken as the final results.

Fig. 3
figure 3

Figures showing region of interest (ROI) delineation. (a) The ROI of the uniformly enhanced lesion was placed in the center of the intestinal wall. (b) The ROI of the stratified enhanced lesion was placed in the mucosal layer.

Full size image

Statistical analysis

Data statistical analysis was performed using SPSS 26.0 software. Continuous variables that followed a normal distribution were represented as mean ± standard deviation, and pairwise comparisons were conducted using independent sample t-tests. Continuous data that did not follow a normal distribution were represented as M (Q1, Q3), and pairwise comparisons were conducted using the Mann Whitney U test. Receiver operating characteristic (ROC) analysis was used for evaluating the performance of quantitative parameters in CD diagnosis and inflammatory activity evaluation. The optimal threshold was obtained through the Youden index (Youden index = sensitivity + specificity − 1), and the corresponding sensitivity and specificity were recorded. The area under the curve (AUC) between the parameters was compared using the Delong test. P < 0.05 indicated a statistical difference.

Results

Patients and radiation dose

This study ultimately included 76 CD patients and 53 controls, with a total of 167 intestinal segments meeting the experimental requirements. There were 37 males (69.81%) and 16 females (30.19%) in the control group, with a median age of 33 (24, 47.25) years old. The clinical information of patients in the CD group is shown in (Table 2). There was no statistical difference in gender and median age between the two groups (P > 0.05). Among CD patients, intestinal lesions affected the ileocolon segment were the most (65% [85.53%]), and perianal lesions were found in 39 patients (39% [51.32%]). The average CTDIvol and DLP for AP were 7.36 ± 0.03 mGy and 364.46 ± 33.04 mGy·cm, respectively. The average CTDIvol and DLP for PVP images were 7.36 ± 0.02 mGy and 369.69 ± 32.96 mGy·cm, respectively.

Table 2 Clinical characteristics of patients with Crohn’s disease.
Full size table

Comparison of energy spectrum parameters between the CD group and the control group

The monoenergetic CT values of intestinal wall (diseased and normal) and mesenteric fat (normal and creeping fat) in both groups decreased with the increase of keV level in dual phases; In dual phases, the SP40–100 and HU40–100keV values of the lesioned intestinal wall and creeping fat in CD patients were higher than those of normal intestinal wall and mesenteric fat in controls, respectively, and the differences were statistically significant (P < 0.001). Representative images of the control group are shown in (Fig. 4).

Fig. 4
figure 4

A 33-year-old female with no apparent clinical diagnosis in control group. (a) depicts the slope parameter mapping (SPI40–100) during arterial phase, with a corresponding SP40–100 values for ROI of 1.2. (b) depicts CT value of virtual monoenergetic image of 40 keV (HU40keV) during arterial phase, with a corresponding HU40keV value for ROI of 104.6 HU. (c) illustrates a colonoscopy image of the distal ileum revealing no mucosal abnormalities.

Full size image

Comparison of energy spectrum parameters between non-moderate-to-severe group and moderate-to-severe group of the CD patients

The monoenergetic CT values of the diseased intestinal wall and creeping fat in both the non-moderate-to-severe group and the moderate-to-severe group decreased with the increase of keV in dual phases. The dual-phase SP40-100 values and the HU40-100keV of the diseased intestinal wall and adjacent creeping fat were higher in the moderate-to-severe group compared to the non-moderate-to-severe group, and the differences were statistically significant (P < 0.001). Representative images are shown in Figs. 5 and 6.

Fig. 5
figure 5

Male, 37 years old, with a comprehensive clinical diagnosis of CD, non-moderate-to-severe group. (a) depicts the slope parameter mapping (SPI40–100) during arterial phase, with a corresponding SPI40–100 value for ROI of 2.1. (b) depicts CT value of virtual monoenergetic image of 40 keV (HU40keV) during arterial phase, with a corresponding HU40keV value for ROI of 181.2Hu. (c) shows a colonoscopy view of the terminal ileum, showing irregular ulceration of the mucosa, slightly longitudinal, obvious congestion and edema of the surrounding mucosa, and polypoid bulge. SES-CD score: 4, or mild.

Full size image
Fig. 6
figure 6

Male, 29 years old, with a comprehensive clinical diagnosis of CD, moderate-to-severe group. (a) depicts the slope parameter mapping (SPI40-100) during arterial phase, with a corresponding SPI40-100 value for ROI of 2.9. (b) depicts CT value of virtual monoenergetic image of 40 keV (HU40keV) during arterial phase, with a corresponding HU40keV value for ROI of 239.0 HU. (c) is a colonoscopy map of the terminal ileum, showing polypoid bulge, red surface, and covered with foul moss and purulent secretions, surrounding mucosal congestion and edema, SES-CD score: 11, moderate to severe.

Full size image

ROC analysis of identifying CD using spectral parameters

The ROC analysis of the identification of CD using energy spectrum parameters of intestinal wall and mesenteric fat is shown in (Fig. 7). The AUCs of HU40–100keV and SP40–100 of the intestinal wall in the diagnosis of CD ranged from 0.893 to 0.948 in the AP and from 0.858 to 0.932 in the PVP. The AP parameters have a higher AUC than the corresponding PVP parameters. The AP HU40keV has the highest AUC among all parameters of the intestinal wall, reaching 0.948. At a cut-off value of 167.64 Hu, the sensitivity is 82% and the specificity is 100%; The AUC of SP40–100 is the second highest (0.946), with a sensitivity of 89.8% and a specificity of 92.5% at a cut-off value of 1.79. For the energy spectrum parameters of mesenteric fat, AP parameters still have higher AUC (0.811 ~ 0.969) than PVP parameters (AUC: 0.811 ~ 0.942), and AP SP40–100 has the highest AUC (0.969). At a cut-off value of -1.42, the sensitivity for diagnosing CD is 88%, and the specificity is 98%. The AP SP40–100 of mesenteric fat also has the highest AUC among all spectral parameters.

Fig. 7
figure 7

ROC curves of dual-phase HU40–100keV and SP40–100 in the diagnosis of CD. (a) Arterial phase parameters of the intestinal wall; (b) Portal phase parameters of the intestinal wall; (c) Arterial phase parameters of mesenteric fat; (d) Portal phase parameters of mesenteric fat.

Full size image

ROC analysis of spectral parameters for evaluating CD activity

Fig. 8
figure 8

ROC curves of dual-phase HU40 –100keV and SP40–100 for differentiation between non-moderate-to-severe CD group and moderate-to-severe activity group. (a) Arterial phase parameters of the diseased intestinal wall; (b) Portal phase parameters of the diseased intestinal wall; (c) Arterial phase parameters of creeping fat; (d) Portal phase parameters of creeping fat.

Full size image

The ROC analysis of the energy spectrum parameters of the diseased intestinal wall and creeping fat in distinguishing non-moderate-to-severe and moderate-to-severe activity of CD is shown in (Fig. 8). For diseased intestinal wall and creeping fat, the AP parameters have a higher AUC than the PVP parameters. The AUCs of HU40–100 keV and SP40–100 of the diseased intestinal wall in the differentiation of moderate-to-severe activity ranged from 0.855 to 0.908 in the AP and from 0.819 to 0.860 in the PVP. The AP HU40keV and HU50keV have the highest AUCs among all parameters of the intestinal wall, reaching 0.908. The AUC of AP SP40–100 is 0.902. Among the energy spectrum parameters of creeping fat, AP SP40–100 has the highest AUC (0.935), with a sensitivity of 79.7%, a specificity of 93.5%, at a cut off value of -0.99. The AP SP40–100 of creeping fat also has the highest AUC among all spectral parameters.

Differences of spectral parameters among aphthous ulcers, large ulcers and very large ulcers

All parameters showed significant differences among bowel segments with aphthous ulcers (54 segments), large ulcers (75 segments) and very large ulcers (29 segments) (all P < 0.05) (Table 3). For intestinal wall and creeping fat at both phases, very large ulcers had the highest values, large ulcers had the second highest, while aphthous ulcers had the lowest (all P < 0.05).

Table 3 Differences of spectral parameters among aphthous ulcers, large ulcers and very large ulcers.
Full size table

Discussion

This study indicated that the SP40–100 obtained based on SPI and the HU40–100 keV of the intestinal wall (or diseased intestinal wall) and the adjacent mesenteric fat (or creeping fat) can be used for CD diagnosis and CD activity evaluation. Therein, AP SP40–100 for mesenteric fat has the highest diagnostic efficacy.

DE-CTE provides various quantitative images such as monoenergetic images, iodine maps, fat maps, and effective atomic number maps, which have been proven to have clinical application value in CD evaluation. However, due to the limitation that commercial software does not provide SPI for energy spectrum curves like the abovementioned images, obtaining the slope of energy spectrum curves requires more steps, which is also one of the possible reasons why the research on energy spectrum curve slope in CD evaluation is still insufficient. This study used the slope formula-based algorithm to obtain SPI, making it easy to measure the slope of the energy spectrum curves. In addition, previous literature on quantitative analysis of energy spectrum parameters has mainly focused on the intestinal wall, with relatively little research on mesenteric fat. This study compared the energy spectrum parameters of the intestinal wall and adjacent mesenteric fat in the diagnosis and activity evaluation of CD. Moreover, recent studies14,25 also used slope of the energy spectrum curves to distinguish the active and inactive CD, while we focused on the differentiation between moderate-to-severe activity CD and non-moderate-to-severe activity group with a larger sample size.

In addition, other studies mostly use the CDAI score as a reference standard for evaluating CD activity, which is subjective and cannot accurately assess the condition to the affected intestinal segment, and may not fully reflect the true status of CD disease. Our study used the SES-CD score as the activity reference standard, which has good consistency with the CDEIS score and a higher degree of objectivity26.

For the intestinal wall, this study showed that the dual-phase SP40–100 and HU40–100 keV of the CD group and moderate-to-severe CD group were significantly higher than those of the control group and non-moderate-to-severe CD group, respectively. The dual-phase SP40–100 and HU40–100 keV of the intestinal wall had good performance in diagnosing CD and evaluating the CD activity. In the intestinal wall of CD patients, due to the congestion and edema mainly caused by various inflammatory cell infiltration and non-caseous granuloma formation27, iodine contrast agent accumulates more between cells, thus the monoenergetic CT value and the slope of energy spectrum curve correspondingly increase. Previous studies also found that the enteric-phase (dual-phase in our study) slope of the energy spectrum curve can be used to evaluate the CD activity of in the affected intestinal wall, consistent with our study28. In different inflammation activity status, the degree of mucosal damage, intestinal wall congestion and edema, and fibrosis changes varies. As the activity of CD enhances, the blood supply to the diseased intestinal wall usually increases, and congestion becomes more pronounced. Wang et al.29 found that the slope of the energy spectrum curve in intestinal wall was not so valuable in analyzing the inflammation activity, which is inconsistent with our study. On the one hand, it may be due to the small sample size of their study; on the other hand, they employed a different method for ROI delineation. They delineated 70 -80% of the intestinal wall for all diseased intestinal segments. However, due to the diverse ways of contrast enhancement in CD intestinal wall, the enhancement may be layered or uniformly consistent. Therefore, our study adopted different measurement methods for the diseased intestinal wall with different enhancement methods.

For mesenteric fat, the results showed that the dual-phase SP40-100 and monoenergetic CT values of adjacent mesenteric fat in the CD and moderate-to-severe CD patients were higher than those in the controls and non-moderate-to-severe CD, respectively. All parameters can be used to diagnose CD and evaluate activity, therein arterial-phased SP40-100 has the highest diagnostic value and is higher than all DE-CTE parameters of the intestinal wall mentioned above. In CD, creeping fat itself has the inflammatory characteristics, and under endoscopy, it is manifested as an abnormal increase in the number of adipocytes, a significant decrease in volume, an increase in non-fat components, enhanced vascular proliferation, and infiltration of macrophages and lymphocytes. In addition, previous studies have shown that it can also produce various inflammatory factors, pro-inflammatory factors, and adipokines, which play important roles in the course of CD30,31,32,33,34. Due to the inflammatory response, the vascular permeability of creeping fat increases, and as the activity increases, the inflammatory response becomes more severe. Therefore, due to vascular proliferation and an increase in non-fat components, creeping fat compared to normal mesenteric fat, moderate-to-severe CD compared to non-moderate-to-severe CD, had higher HU40–100keV and SP40–100. Feng et al.35 found that the slope of energy spectrum curve for the mesenteric fat had clinical value in the diagnosis and evaluation of CD activity, which is consistent with us. However, there is a slight difference in the diagnostic threshold compared to this study. Possible reasons include that the ROI size of the lesion varies. In this study, the ROI delineated was smaller than that of the literature in order to avoid mesenteric blood vessels as much as possible during the data measurement process. It may also be derived from the different concentrations of contrast agents and injection plans used in this experiment. Li et al.25 also found that the slope of the energy spectrum curve for creeping fat can be used to evaluate the CADI-determined CD activity, and its AUC is the highest among the explored energy spectrum parameters, similar to the results of this study. However, the creeping fat delineated by them was determined by pathology, and the energy spectrum parameters were measured in the enteric phase (similar to the PVP in this study).

In this study, the efficacy of monoenergetic CT values and the slope of energy spectrum curve in CD diagnosis and inflammation activity evaluation was relatively high. This may be due to the fact that the data in this study were first measured on the SPI and then copied to the other monoenergetic images. The SPI has a relatively high CNR, displaying the lesion segment more obvious and may be more accurate during measurement. In addition, this study found that all quantitative parameters had slightly higher diagnostic efficacy in diagnosing and evaluating the activity of CD patients in the AP compared to the PVP, consistent with the findings of Xiao et al.36 on the lesion of the intestinal wall. It may be due to acute congestion caused by mesenteric vascular dilation in CD, which is particularly pronounced in the moderate-to-severe active period. The lesion can approach or reach a peak of enhancement in the intestinal wall at AP, while the degree of enhancement in the PVP tends to flatten. During the mild active and remission periods, the diseased intestinal wall at the PVP often shows progressive enhancement due to significant non caseous granulomas and fibrosis changes. The normal intestinal wall often shows progressive enhancement. But the specific reasons still need to be further explored.

This study has the following limitations. (1) Single center and small sample size limited the universality of our results. (2) Due to the limited number of intestinal segments affected during the remission period, the remission and mild activity periods were grouped together for analysis, and the diagnostic value of DE-CTE quantitative parameters for the remission and mild activity periods in CD patients was not evaluated. (3) This study used SES-CD as the reference standard. Although many studies have suggested that it can serve as the gold standard for evaluating activity, it can only reflect mucosal changes and cannot observe extraintestinal lesions, which still has certain limitations. (4) Previous studies have shown that other DE-CTE parameters including iodine concentration can be used to quantitatively evaluate the activity of CD, but this study did not include it, which should be further explored. (5) Our study did not compare the diagnostic efficacy between visuoperceptual evaluation with conventional or dual-energy CT images and the quantitative measurements on slope parameter mapping, which should be further investigated.

In summary, in dual-phase DE-CTE, the SP40–100 measured on SPI has high application value in the diagnosis and activity evaluation of CD, and the AP SP40–100 of the mesenteric fat adjacent to the affected intestinal wall has the best performance.