Imaging of acute superior mesenteric artery embolus using spectral CT in a canine model

Abstract

Objective:

To explore the diagnostic value of single-source dual-energy spectral CT (sDECT) imaging in an acute superior mesenteric artery embolus (SMAE) canine model.

Methods:

Pre-contrast and double-phase contrast-enhanced sDECT were performed before and after embolization in eight SMAE dog models. Monochromatic images of embolized intestine with the best contrast-to-noise ratio (CNR) were obtained and compared with the polychromatic images. CT parameters including attenuation value, iodine content, water content and thickness of the embolized intestinal segments were obtained, and normalized difference in iodine concentration (NDIC) was calculated.

Results:

The CNR in pre-contrast, arterial phase and portal venous phase at 4 h after embolization was 1.11 ± 1.23, 13.50 ± 1.54 and 10.63 ± 3.75, respectively, significantly higher than those of the polychromatic images (p < 0.05). The iodine-based images clearly revealed the embolized intestinal segments, which were highly consistent with the gross findings. The difference in attenuation values between the embolization area and non-embolization area in the monochromatic images was 105.06 ± 35.35 HU, higher than that in the polychromatic images (p < 0.001). The attenuation values and NDIC were significantly decreased at 2 h after embolization, relatively increased at 4 h and gradually decreased at 6 and 8 h. The changing pattern of thickness was similar to that of NDIC over time after embolization.

Conclusion:

sDECT can provide the optimal monochromatic images and allow increased detection rates of lesions. sDECT is a very promising tool for quantitative diagnosis of SMAE.

Advances in knowledge:

Our research provides more quantitative parameters for the assessment of SMAE by sDECT.

INTRODUCTION

Acute mesenteric arterial embolus (AMAE) is a rare but critical surgical acute abdomen, and its hospital mortality is up to 55.5% in spite of gradual improvements in treatment.¹ Since the affected site, range and severity varies, AMAE has complex and variable clinical symptoms and signs, laboratory tests and imaging manifestations, and its early diagnosis is very difficult because of overlapping manifestations with primary diseases.^2,3

Recent great advances in multislice CT (MSCT) have encouraged research into the diagnosis of AMAE. Multislice CT angiography (MSCTA) combined with double-phase contrast-enhanced scan not only can determine the position and severity of superior mesenteric artery embolus (SMAE) but also can evaluate the severity and range of ischaemic damage to the mesentery and intestinal wall. MSCTA has become the first choice to evaluate SMAE.^3–5 However, conventional single-energy MDCT is more susceptible to X-ray hardening artefacts and cannot separate materials of different densities.⁶

Single-source dual-energy spectral CT imaging (sDECT) is a new functional imaging technique using a dual-energy scan mode. Monochromatic imaging is less vulnerable to artefacts such as beam hardening and pseudoenhancement.⁷ This is useful as it improves image quality and the visualization of fine anatomic structures. sDECT allows differentiation of materials and measurement of iodine concentration in the tissues via iodine-based images. This makes it possible to analyse the iodine uptake of tissues or organs directly and effectively reflect the blood perfusion state of tissues or organs.^6,8 A recent study⁹ has investigated the performance of spectral CT in depicting the superior mesenteric artery (SMA). It is of great importance to explore the diagnostic value of sDECT in SMAE. This study aims to explore the diagnostic value of sDECT in a SMAE dog model and to evaluate the ischaemic severity using iodine-based images.

METHODS AND MATERIALS

Reagents and instrumentation

Pentobarbital sodium was purchased from Sigma-Aldrich (St Louis, MO) and was diluted to a concentration of 2.5% with 0.9% NaCl before use. The contrast agent iobitridol injection (350 mg iodine/ml) was purchased from Guerbet Asia Pacific (Guerbet, Paris, France). GE Innova 2100-IQ, Discovery™ CT750 HD and Workstation AW4.5 (GE Healthcare, Waukesha, MI) were used in this study.

Animal models

Eight healthy, adult, hybrid dogs (male or female, weight: 15–22 kg) were provided by our hospital (First Affiliated Hospital of Harbin Medical University). All the protocols were approved by the Institutional Animal Care and Use Committee and were conducted according to the Institutional guidelines for Care of Laboratory Animals.

A 5 Fr catheter was delivered to the middle colic artery originating from the SMA trunk by the Seldinger technique, and then the 2.0 × 2.0 × 2.5 mm self-made gelatin sponge particles were injected. Angiography was performed to demonstrate the embolization situation. After embolization, the catheter was removed and local compression haemostasis was performed.

The experimental protocol is shown in Figure 1. sDECT scans were performed at five time points (before embolization, 2, 4, 6 and 8 h after embolization). Animals were sacrificed for pathological examinations at 4 and 8 h after embolization (n = 4 at each time point).

Figure 1.

Scheme of the experimental protocol. sDECT, single-source dual-energy spectral CT; SMA, superior mesenteric artery.

CT examination

The animals were fixed on the CT scan bed in a dorsal position. Pre-contrast and double-phase contrast-enhanced sDECT were performed with a scan range from the diaphragmatic dome to the pubic symphysis. The contrast agent was injected at a dose of 1.5 ml kg⁻¹ and an injection rate of 2.5 ml s⁻¹. The scan was triggered with SmartPrep technique, and the monitoring level was set at the abdominal aorta above the opening of coeliac trunk. After reaching a predefined threshold of 120 HU, the scan was started by a delay of 6.5 s [arterial phase (AP)] and by a delay of 25 s [portal venous phase (PP)]. The following acquisition parameters were employed: tube voltage of 80 and 140 kVp with instantaneous (0.5 ms) switch; automatic tube current–exposure time adjustment for tube current; a slice thickness of 64 × 0.625 mm; a pitch of 0.984; and a tube rotation speed of 0.6 s per rotation.

Pathological examination

The abdomen was opened to observe the abdominal cavity, mesentery and intestine. The animal was sacrificed by injecting excessive pentobarbital. Two embolized intestinal segments (jejunum in the upper left abdomen and ileum in the right next abdomen), and one normal intestinal segment (duodenum) were excised, then were soaked and fixed in 10% formalin solution, embedded with paraffin and haematoxylin–eosin (HE)-stained. The shape and structural changes in various intestinal layers were examined under high-power microscope. Pathological observation was performed independently by a pathologist.

Image evaluation

The polychromatic images (quality control images) and monochromatic images at 101 continuous energy levels from 40 to 140 keV were reconstructed at a slice thickness of 1.25 mm and a slice interval of 1.25 mm with the data obtained by plain and double-phase contrast-enhanced scan. The post-processing was performed on Workstation AW4.5. Data processing and analysis were performed with sDECT viewer software package. CT images were reviewed and analysed by two radiologists.

Two embolized intestinal segments corresponding to pathological observation sites and normal intestinal segments in the same slice were selected from the polychromatic images and sDECT images in different phases, and values of three points in the cross-section were measured and averaged. The optimal monochromatic energy needed to observe the embolized intestine was obtained using the optimal CNR function. Data from all groups were recorded and averaged. The averaged number of all the groups was used as the optimal kiloelectron volts (keV) and the images of the optimal keV at this energy level were saved.

Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of monochromatic images and polychromatic images obtained in pre-contrast, AP and PP scans were compared. The following formulae were used:

$$\text{SNR}=\raisebox{1ex}{${\text{ROI}}_{\text{intestine}}$}\!\left/ \!\raisebox{-1ex}{${\text{SD}}_{\text{background}}$}\right.,$$

$$\text{CNR}=\raisebox{1ex}{${\text{ROI}}_{\text{embolized}}-{\text{ROI}}_{\text{normal}}$}\!\left/ \!\raisebox{-1ex}{${\text{SD}}_{\text{background}}$}\right..$$

Wherein, SD_background was the mean of the standard deviations of the CT attenuation values of three regions of interest (ROIs) in the subcutaneous fat layer. The spectral curve was drawn automatically by the sDECT Viewer software package (GE Healthcare).

With water–iodine as base substance pair, the perfusion-reduced or perfusion-absent area was visualized in the iodine-based images and compared with the pathological results. The thickness, attenuation values, water content and iodine concentration of the selected intestinal segments were measured before embolization and after embolization. The CT enhancement value and the difference in iodine concentration (DIC) were calculated by subtracting the pre-contrat values from the contrast-enhanced values. The thickness change was calculated by subtracting the thickness at the last scan time point from the thickness at the current time point. In order to avoid the interference of various factors other than blood flow in the assay of iodine concentration, we normalized the DIC using a new indicator—normalized difference in iodine concentration (NDIC) according to the report of Lv et al,¹⁰ by dividing the DIC of observed tissues by the iodine concentration in the aorta at the SMAE level. CT enhancement value–time curve, NDIC–time curve and thickness of intestine wall–time curve were drawn, respectively. Measurement was performed in the PP in which intestinal walls were homogeneously enhanced, which definitely had the additional advantages in depicting mesenteric veins, allowing better assessment of abnormalities of the bowel wall itself and providing greater accuracy in the detection of perforation, abscess formation and peritonitis.⁴

Ethics Statement

The study was provided by Animal Experiment Center, the First Affiliated Hospital of Harbin Medical University. All the protocols were approved by the Institutional Animal Care and Use Committee and were conducted according to the Institutional guidelines for Care of Laboratory Animals.

Statistical analysis

SAS 9.1.3 statistical software (SAS Institute Inc., Cary, NC) was used for data analysis. The quantitative data were expressed as mean ± SDs. Pairwise t-test was employed to compare polychromatic and monochromatic images. Independent samples t test was performed on NDIC differences between the two groups. One-way analysis of variance with replicate measures was used for differences on multiple time points. p < 0.05 indicated that the difference was statistically significant.

RESULTS

Superior mesenteric artery embolus model and basic CT findings

The SMAE model was successfully established in eight animals. Gross observation revealed that the intestine within the embolization area was dark purple or pale, the mesenteric vessels were sparse and intestinal peristalsis had weakened or disappeared. In the animals sacrificed at 4 h (Group I), lesions (prominent congestion and oedema) were mainly found in the mucosal layer and (or) submucosal layer. In the animals sacrificed at 8 h (Group II), lesions of six segments involved the whole range of changes, and the borders of various layers were unclear, while lesions of the other two segments only affected the submucosal layer. Transmural infarction was not seen. The NDIC average numbers of Group I and Group II were 0.053 ± 0.029 and 0.005 ± 0.004, respectively, in the PP (t = −3.14; p = 0.048).

At the 8-h time point after embolization, two cases developed mesenteric oedema with an elevated spectral curve, and the corresponding CT images showed that the oedematous mesentery was vague and the density of this area was increased (Figure 2a,b). The enhancement of all the embolized intestinal walls was decreased, or even disappeared, and the height and slope of the spectral curve were reduced (Figure 3).

Figure 2.

(a) Spectral curve of oedematous mesentery (blue) and normal omentum (green). (b) A 62-keV monochromatic image showed that the oedematous mesentery was vague and the density was increased. HU, Hounsfield units.

Figure 3.

Spectral curves of embolized intestinal wall (a). Blue and pink represent the spectral curves of normal duodenal descending part and partial jejunum, respectively; green represents the spectral curve of ischaemic ileum; light blue represents the spectral curve of embolized jejunum. Iodine-based image (b). The colour of region of interest corresponds to the colour of the spectral curve. The iodine-based image shows the iodine distribution of corresponding intestinal segments based on brightness. HU, Hounsfield units.

Comparison of monochromatic images and polychromatic images

The optimal CNR has no significant difference between various groups (Table 1), and the averaged number for visualizing the embolized intestine was 61.88 ± 5.42 keV. Therefore, the optimal monochromatic energy level was set as 62 keV to compare the changes of CT parameters in the subsequent study. At the 4-h time point after embolization, SNRs of monochromatic images obtained in pre-contrast, AP and PP scans were higher than those of polychromatic images except for SNR in PP scan, while CNRs of monochromatic images were significantly greater than those of polychromatic images (Table 2).

Table 1.

Optimal kiloelectron volts for visualizing the embolized intestinal segments in different scan phases

Time	Pre-contrast, mean ± SD	Arterial phase, mean ± SD	Portal venous phase, mean ± SD
2 h (n = 8)	62.31 ± 4.18	62.25 ± 6.80	60.75 ± 3.14
4 h (n = 8)	63.13 ± 4.43	61.44 ± 3.39	62.37 ± 2.78
6 h (n = 4)	60.88 ± 2.95	62.75 ± 3.80	62.50 ± 6.16
8 h (n = 4)	59.50 ± 5.61	60.63 ± 5.25	63.25 ± 2.06
F	0.17	0.66	1.05
p-value	0.239	0.595	0.417

SD, standard deviation.

Table 2.

Comparison of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) between polychromatic and monochromatic images in different scan phases at 4 h after embolization

Value of CNR or SNR	Polychromatic images, mean ± SD	Monochromatic images, mean ± SD	T	p-value
SNR in pre-contrast	7.17 ± 0.52	10.28 ± 2.27	−4.24	0.004
CNR in pre-contrast	0.29 ± 0.57	1.11 ± 1.23	−2.63	0.034
SNR in AP	8.77 ± 1.47	10.67 ± 1.55	−3.11	0.017
CNR in AP	10.77 ± 3.25	13.50 ± 1.54	3.05	0.019
SNR in PP	9.31 ± 1.70	9.25 ± 1.52	0.19	0.856
CNR in PP	8.58 ± 3.94	10.63 ± 3.75	3.07	0.018

AP, arterial phase; PP, portal venous phase; SD, standard deviation.

Attenuation values, iodine and water contents of intestinal walls between the embolization area and non-embolization area

In the iodine-based images, all embolized intestinal segments were highly consistent with the gross observations (Figures 3b and 4). Comparison of attenuation values, iodine concentrations and water content of intestinal walls within the embolization area and non-embolization area in the PP scans 4 h after embolization are shown in Figure 5a–d. The attenuation values and iodine concentration of intestinal walls in the embolization area were significantly lower than those in the non-embolization area (p < 0.001), and a more significant difference in attenuation values was observed in monochromatic images than in polychromatic images. There was no statistically significant difference in the water content of intestinal walls at various embolization time points (F = 1.94, p = 0.194).

Figure 4.

The iodine-based images of lower abdomen at 2 h (a), 4 h (b) and 8 h (c), respectively. The iodine-based images show the iodine distribution of corresponding intestinal segments based on brightness. The arrows show changes of intestinal thickness and enhancement degree after embolization over time. The thickness and enhancement degree were increased and then decreased. The “ring” sign was observed at 4 h. L, window level of the iodine concentrations; W, window width of the iodine concentrations.

Figure 5.

Comparison of attenuation values in polychromatic images (a), of attenuation values in monochromatic images (b), iodine concentrations (c) and water content (d) of intestinal walls within the embolization area and non-embolization area (control) in the portal venous phase scans 4 h after embolization. HU, Hounsfield units.

Changes of the CT values over time after embolization

After embolization, the CT attenuation value–time curve of the intestinal wall was similar to the NDIC–time curve (Figure 6a,b). The attenuation values were significantly decreased at 2 h after embolization, slightly increased at 4 h and gradually decreased at 6 and 8 h; there were statistically significant differences between 4-h and other time points (p = 0.043, 0.025 and 0.001, respectively).

Figure 6.

Change of CT attenuation values [in Hounsfield units (HU)] (a), normalized difference in iodine concentration (NDIC) (b) of the embolized area in the intestinal walls and change of the thickness of intestinal wall of the embolized area (c) after embolization. AP, arterial phase; PP, portal venous phase.

At 2 and 4 h, the CT enhancement values in monochromatic images were higher than those in polychromatic images; therefore, monochromatic images were more sensitive in recognizing the slight enhancement of intestinal walls. The thickness of intestinal wall also demonstrated a dynamic change over time after embolization (Figure 5a–c). The changing pattern of thickness was similar to that of NDIC over time after embolization (Figure 6c).

DISCUSSION

There are many animal models currently employed for AMAE research. In this study, hybrid dogs were used as experimental animals because of their similar manifestations to human beings when the ischaemic intestine was >65 cm.¹¹ Among various modelling methods, the intervention technique is commonly used, and the optional embolus includes gelatin sponge, cyanoacrylate, anhydrous ethanol and autothrombus.^12–14 Gelatin sponge with a particle size of 2.0 × 2.0 × 2.5 mm was employed as an embolus in this study to ensure accurate trunk embolization and to prevent entry of embolus into Grade 2 branches.

The experimental animals were grouped mainly according to the method described in the pathological study conducted by Ma et al.¹⁴ Their study showed that the intestinal wall lesions within 4 h only affect the mucosal and submucosal layers, and the intestinal changes are reversible and the intestinal activity can be restored by early blood circulation reconstruction. After 4 h, the lesions gradually invade the whole range of intestinal walls, and the intestinal changes become irreversible. The pathological changes at 4 h in this study are coincident to their report, while the lesions at 8 h are less severe than their report and the change in the muscular layer is mainly the infiltration of inflammatory cells.

At present, there are two scan modes for DECT, namely the double-tube and double-energy scan mode and the high/low voltage instantaneous scan mode. The high/low voltage instantaneous scan mode is used by Discovery CT750 HD, and sDECT is achieved by instantaneous (0.5 ms) switch of 80 and 140 kV and analysis of projection data. Compared with traditional CT, sDECT provides four new analysis tools: (1) monochromatic image, (2) spectral curve, (3) substance separation and quantification, and (4) effective atomic number.¹⁵

The monochromatic images at the continuous energy levels of 40–140 keV can be acquired with just one sDECT scan, and the images with the optimal CNR for visualizing pathological tissues and normal tissues can be acquired by Optimal CNR software (GE Healthcare). In this study, using normal intestine as control, the optimal monochromatic energy level for visualizing the embolized intestine was found to be 62 keV. There were no significant differences in optimal CNRs at different ischaemic time points and in different scan phases. This complies with the principle that low-energy images can increase the contrast between different tissue structures.¹⁵ The CNR of the optimal monochromatic images was significantly higher than that of polychromatic images (i.e., traditional MSCT images), which was beneficial in improving the image density resolution and the visualization capacity of lesions. As observed in this study, the difference in density between the embolized intestine and normal intestine in the optimal monochromatic images was much more significant, and the change of the contrast-enhanced CT attenuation values of embolized intestine was greater than that in polychromatic images. These findings would be conducible to the detection of early lesions. Compared with traditional CT images, optimal sDECT images had better CNR and SNR. Ischaemic intestinal segments could be visualized easily, and greater CT attenuation values could also be obtained.

The spectral curve is another important tool of sDECT and represents the change laws of attenuation values of the ROI at different keV. Different tissues and organs have different spectral curves, which are helpful in evaluating the nature of lesions and comparing the similarity or difference between lesions.^15,16 This study demonstrates that the normal mesentery had a special spectral curve (i.e., an upward-arched curve shadow), and the spectral curve of oedematous mesentery was elevated and flattened with a decreased slope. This provides an objective indicator for evaluating mesenteric oedema.

The contrast-enhanced spectral curve of normal intestine was a downward-arched curve shadow with a negative slope, reflecting the characteristics of iodine in the intestinal wall. The absolute height and slope of the spectral curve of embolized intestine were decreased, reflecting the decreased iodine concentration and reduced blood supply in the corresponding intestine. The severity of ischaemia could be evaluated from the contrast to the spectral curve of normal intestine.

Compared with traditional CT images which mainly rely on grey level and intensities, spectral CT is helpful in differentiating lesions and normal tissues in terms of shape, height and slope of the curves. Spectral CT is more objective and sensitive in examining oedematous mesentery and intestinal wall ischaemia.

sDECT can present the CT attenuation coefficient of a tissue or organ based on the material decomposition images, generate the distribution profiles of the corresponding base substances and thus complete the quantitative assay of substance concentration. Water–iodine is a common base substance pair in clinical practice, because they cover the density range of all substances in human tissues and organs while iodine is also a basic component of CT contrast agents. The iodine concentration can reflect the blood perfusion of tissues and organs, which has been confirmed in the studies of pulmonary embolism and ischaemic cardiomyopathy.^5,16–19 As found in this study, iodine-based images can be used to accurately evaluate the reduced or absent perfusion of intestine and showed good consistency with the pathology results. The NDIC value of the intestinal walls decreased significantly after embolization. Meanwhile, the NDIC value of the embolized intestine walls was significantly different from that of the non-embolized intestine walls, which suggested that NDIC could be used as an indicator in the diagnosis of intestine wall ischaemia.

There was significant difference for NDIC between Groups I and II. The lower NDIC represented the more severe pathological changes. When the lesions invaded the muscular layer, NDIC approached to zero, which meant no blood perfusion in the intestine, even though transmural infarction did not appear. But, it did not apply to the early embolization. From the CT enhancement–time curve and NDIC–time curve, it could be seen that within the first 2 h of embolization, NDIC significantly decreased and even approached to zero temporarily. According to the previous studies,^13,20 the CT enhancement value after SMA embolization decreases progressively over time. We think that the difference is mainly associated with the animal model. In this study, a trunk embolus model was used. There was no intestinal perfusion in the early stage as there was no bypass circulation. However, slight enhancement of intestine (at 4 h after embolization in most cases) may be due to developing bypass circulation. Thereafter, the lesions become aggravated over time and intestinal enhancement was decreased progressively until it was absent. SMAE mainly occurs at the middle colic artery originating from the trunk; therefore, we believe that the model used in this study is closer to the clinical condition.

The change of intestinal wall thickness is one of the main imaging features of SMAE and is also one indicator for evaluating lesion severity. One finding of this study is that the changing pattern of intestine wall after embolization is similar to that of NDIC. This study revealed that the dynamic change of intestinal wall thickness was positively correlated with NDIC. The embolized intestine walls did not thicken within the first 2 h, and the corresponding NDIC was very low. From 2 to 4 h, the embolized intestine walls gradually thickened and the corresponding NDIC increased. The pathological change was mainly the congestion and oedema of the mucosal and submucosal layers, which often appeared after the generation of bypass circulation. After 4 h, intestine walls gradually thinned and NDIC decreased correspondingly. This finding was consistent with previous reports^3–5 that the thinning of the intestinal wall was often observed in the later stages, probably caused by injury to the nerves in the submucosal layer or injury to the muscle layer. The thickness of the intestinal wall is related to the perfusion state of the intestine. In this study, several scans were performed, and intenstinal walls could be evaluated using the difference between two scans. There is no uniform evaluation standard for thickened or thinned intestinal wall within a single scan in clinical practice; therefore, NDIC is a significant parameter for evaluating the severity of intestinal pathological changes.

This study has the following limitations: the dog experimental model was used, and the value of sDECT for SMAE diagnosis in human beings needs to be validated. This study has a relative small sample size, and hence the diagnostic specificity and sensitivity of sDECT and traditional MSCT were not determined and compared.

CONCLUSION

sDECT can provide optimal monochromatic images with good contrast, increase the density contrast between lesions and normal tissues and allow improved detection rates of lesions. The iodine-based images acquired with water–iodine as base substance pair can clearly show the blood perfusion-reduced or perfusion-absent areas of the intestine, which is helpful in evaluating the severity and range of ischaemia. sDECT provides more quantitative indicators, and NDIC could be a valuable parameter for assessment of the severity of ischaemia. In summary, sDECT is a very promising tool for quantitative diagnosis of SMAE. Further studies with a large sample size are warranted to set up a reference value of NDIC in examining mesenteric ischaemia and to determine the specificity and sensitivity of spectral CT in diagnosis of SMAE.