Evaluating the Image Quality of Monoenergetic Images From Dual-Energy Computed Tomography With Low-Concentration and Low-Flow-Rate Contrast Media for the Arterials Supply to the Nipple-Areola Complex in Breast Cancer Compared With Conventional Computed Tomography Angiography.

OBJECTIVE: The objective of this study was to evaluate the image quality of monoenergetic images (MEIs (+)) acquired from dual-energy computed tomography with low-concentration and low-flow-rate contrast media for the arterial supply to the nipple-areola complex (NAC) in breast cancer compared with conventional computed tomography angiography (CTA).
METHODS: We enrolled 25 patients (MEI (+)300 group, 300 mg/mL and 2.5 mL/s of contrast media) and 23 patients (CTA370 group, 370 mg/mL and 3.5 mL/s of contrast media) for assessing NAC blood supply angiography. The image quality of the 2 groups was evaluated objectively and subjectively.
RESULTS: The 40 keV MEI (+)300 demonstrated higher attenuation and contrast-to-noise ratio than CTA370 group (P < 0.001). The subjective image quality and visualization of the arteries were comparable between 2 groups.
CONCLUSIONS: The 40 keV MEI (+)300 acquired from dual-energy computed tomography can achieve comparable image quality of arterial supply to NAC with low-concentration and low-flow-rate contrast media in breast cancer compared with CTA370.

Breast cancer is the most frequently diagnosed cancer among women and is the leading cause of cancer death in many countries.[1] As a result of early detection, improved adjuvant therapy, and a better understanding of tumor biology, more patients with breast cancer choose breast-conserving therapies. Nipple-sparing mastectomy (NSM) is a new surgical approach characterized by the preservation of the entire nipple-areola complex (NAC) and breast skin envelope during mastectomy, thereby promoting superior esthetic outcomes and improving quality of life when combined with immediate breast reconstruction.[2] However, one potential pitfall of NSM is the uncertainty of the arterial supply to the NAC, which is important to consider to ensure perfusion to the NAC before choosing the vascular pedicle.[3] Previous anatomical studies[3-5] have revealed that the arterial supply to the NAC originates from vascular sources such as the internal thoracic artery, lateral thoracic artery, thoracoacromial artery, intercostal artery, and axillary artery. However, many studies[5-7] have also shown that the arterial supply pattern of the NAC is highly variable across individuals. Therefore, it is an increasingly common practice in NSM to plan procedures using noninvasive imaging technologies to map the vascular anatomy preoperatively to confirm the location, size, and source of the arterial supply to the NAC.

Computed tomography angiography (CTA) is considered the criterion standard for preoperative vascular imaging in vivo[8,9] because it can display the complete outline of the breast vascular map preoperatively by 3-dimensional visualization in a manner that is robust, accurate, intuitive, and of satisfactory resolution. In addition, CTA has proven to be an excellent predictor of sensitivity and positivity, and several studies[10,11] have demonstrated that routine preoperative use of CTA can shorten surgery time and reduce emotional stress for the surgeon as well as the number of postoperative complications. With the advent of rapid CTA scanning patterns, administration of contrast media with a high iodine concentration (ie, 370–400 mg/mL) and high-flow-rate injection modes have become routine clinical practices to maximize arterial enhancement in the systemic circulation.[12-14] The concomitant disadvantage of CTA is that it can increase the risk of contrast media-induced acute kidney injury,[15,16] and the relatively large quantity of high viscosity contrast media could place an increased strain on the vein wall, increasing the risk of extravasation.[17]

Previous studies[16,18-20] have shown that low–kiloelectron volt monoenergetic images (MEIs (+)) from dual-energy computed tomography (DECT) constitute a useful protocol for CTA, demonstrating superior iodine contrast enhancement with a low dose of contrast media compared with that of conventional CTA. In a study of chest CTA,[13] the authors found that the image quality of MEIs (+) at 60 keV with low-concentration contrast media (170 mg/mL) was comparable with that of single-energy computed tomography (CT) with high-concentration contrast media (350 mg/mL). To date, few comparative studies have compared the image quality between MEIs (+) obtained with low-concentration and flow-rate contrast media and conventional CTA images for the preoperative imaging of the arterial supply to the NAC.

We hypothesized that MEIs (+) acquired from DECT can improve the attenuation of the arterial supply to the NAC with low-concentration and low-flow-rate contrast media. In our study, DECT was performed for a MEI (+)300 group (300 mg/mL, 2.5 mL/s of contrast media) and a CTA370 group (370 mg/mL, 3.5 mL/s of contrast media) with breast cancer, and the attenuation and SD of the 5 arterials supply to the NAC were measured in the MEI (+)300 and CTA370 groups individually. The purpose of our study was to evaluate the imaging quality of MEIs (+) acquired from DECT with low-concentration and low-flow-rate contrast media for the arterial supply to the NAC in breast cancer compared with conventional CTA.

MATERIALS AND METHODS

Patient Characteristics

Our prospective, single-center study was approved by the ethics committee of our hospital, and written informed consent was obtained from all participants. A total of 48 patients with histopathologically proven primary breast tumors who had been admitted to our hospital between June 2019 and January 2020 were included. The MEI (+)300 group (25 participants) underwent DECT of the thorax for the staging of lung metastasis. The CTA370 group (23 participants) underwent DECT of the thorax for preoperative NAC blood supply angiography. Inclusion criteria included a first diagnosis of breast cancer and a lack of history of chemotherapy or radiation therapy in the breast space. The exclusion criteria for DECT were based on the clinical guidelines for contrast-enhanced CT at our institution: impaired kidney function (glomerular filtration rate, <30 mL/min), history of severe contrast media allergy, pregnancy, and inability to give informed consent for the CT examination.

DECT Image Acquisition

Image data were acquired on a 2.5 generation dual-source CT unit (SOMATOM Drive; Siemens Healthineers, Forchheim, Germany) in dual-energy mode using a tin filter for the high-voltage tube. An automatic exposure control system (CARE Dose4D; Siemens Healthineers) was used for all scans. All patients were scanned craniocaudally in the supine position with the bilateral arms elevated in close contact with the head. The whole chest was scanned from the superior aperture of the thorax to the inferior edge of the costophrenic angle during a deep-inspiratory breath hold, covering the breast and axillary areas. For contrast-enhanced scanning, contrast media were administered through the right or left ulnar vein by a dual-head injector. The ulnar vein contralateral to the suspected breast lesion was selected to avoid beam hardening artifacts of the axillary vein. The scanning and injection protocols for the 2 groups are described in Table 1.

TABLE 1

Scanning and Injection Protocols in 2 Groups

Parameter	MEI (+)300 Group (n = 25)	CTA370 Group (n = 23)
Scanning parameters
Tube voltage, kVp	80 (Tube A)/Sn140 (tube B)	80 (Tube A)/Sn140 (tube B)
Blending factor	0.6	0.6
Reference tube current time product, mAs	71 (Tube A)/60 (tube B)	71 (Tube A)/60 (tube B)
Rotation time, s	0.28	0.28
Pitch	0.55	0.55
Collimation, mm	64 × 0.6	64 × 0.6
Reformatted section thickness, mm	1.5	1.5
Reformatted section increment, mm	1.5	1.5
ADMIRE level	2	2
Kernels (mediastinum)	Q30f	Q30f
Threshold of bolus-tracking (region of interest in ascending aorta), HU	100	120
Delay time, s	10	10
Injection parameters
Contrast media	Iohexol (Yangzijiang, Jiangsu, China)	Iopromide (Bayer AG, Berlin, Germany)
Iodine concentration, mg/mL	300	370
Ion type	Nonionic	Nonionic
No. benzene rings	1	1
Viscosity (37°C), cP	6.1	10
Osmotic molecular concentration (37°C), osm/kg	0.64	0.77
Volume, mL/kg	1.2	1.5
Flow rate, mL/s	2.5	3.5
Iodine content, mg/kg	360	555
Injection mode	2 Phase, bolus injection	2 Phase, bolus injection
Physiological saline volume, mL	30	30
Physiological saline flow rate, mL/s	2.5	3.5

ADMIRE indicates Advanced Modeled Iterative Reconstruction.

DECT Image Reconstruction

Reconstructed CT image data were postprocessed on a syngo.via workstation (syngo.via VB20A, Dual Energy; Siemens Healthineers). In the MEI (+)300 group, noise optimized MEIs (+) were reconstructed at 40, 50, 60, 70, and 80 keV. In the CTA370 group, standard linear-blended polyenergetic image were reconstructed by applying a blending factor of 0.6 (M_0.6; 60% of the low-kilovolt spectrum and 40% of the high-kilovolt spectrum). Multiplanar reformations were generated for the optimal MEI (+)300 and CTA370 groups in the same workstation. Multiplanar reformations were reconstructed in the coronal plane with 20-mm section thickness and no intersection gap with the maximum intensity projection method. All image data sets, including reformatted images and axial images, were archived to a picture archiving and communication system for interpretation.

Objective Image Analysis

For quantitative evaluation, parameters were measured by 2 radiologists (X.W., with 7 years of experience in breast and chest diagnostic imaging, and X.Z., with 2 years of experience in postreconstruction imaging). We measured the attenuation (Hounsfield units) and SD by placing a manually defined region of interest on all axial images for the 5 arterials supply of the NAC (internal thoracic artery, lateral thoracic artery, thoracoacromial artery, intercostal artery, and axillary artery) and the ipsilateral pectoral muscle with breast cancer. Regions of interest were drawn as large as the vessel lumen, and wall calcifications that might cause artifacts were avoided. The mean attenuation and SD were calculated for each patient by averaging the values from the 5 arterials supply of the NAC. The contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) were calculated according to the following formulas:

Subjective Image Analysis

Subjective imaging data were performed by the same 2 radiologists who obtained the quantitative parameters. They were unaware of the DECT scanning parameters and reviewed the multiplanar reformations MEI (+)300 image with the highest CNR and SNR among the MEIs (+)300 and CTA370 groups at the same window width and window level after objective image analysis 1 month later independently. When evaluating the 2 groups of images, source blood vessels that either extended directly or had branches or perforators winding distally into the NAC regions were defined as the dominant vessels for NAC perfusion.[5] Then, each dominant vessel was identified and ascribed to a corresponding source vessel that included the internal thoracic artery, the lateral thoracic artery, the thoracoacromial artery, the intercostal artery, and other arteries likely to be the source vessels for NAC perfusion. The image quality was assessed with a subjective 5-point scale system[16] based on the overall clarity of the anatomic vascular details. Image quality was scored as follows: 5, sharp anatomical structure and satisfactory details; 4, less clear anatomical structure and details; 3, decreased confidence in details but anatomical structure still relatively clear; 2, poor details and anatomical structure decreased; and 1, unacceptable anatomical structure and details. The visualization of the arterials supply to the NAC was assessed on a 3-point scale: 0, inadequate visualization; 1, adequate visualization; and 2, excellent visualization.

Statistical Analysis

Statistical analyses were performed using commercially available statistical software (SPSS software, version 25.0; Armonk, NY). Interobserver variability and statistical analyses were calculated for every pairwise combination of observers. A 2-way random intraclass correlation coefficient model with absolute agreement, single measures, and 95% confidence interval was used. Quantitative data (age, height, weight, body mass index, maximal tumor diameter, radiation dose, attenuation, SD, SNR, and CNR) are presented as the means ± SDs and were compared using the independent samples t test. The characteristics of the 2 groups (menstruation state, prevalence of diabetes and hypertension, tumor position, and concomitant signs) were assessed using the χ2 or Fisher exact test. For qualitative analysis, the differences between the subjective scores of 2 groups were tested with the Mann-Whitney U test. The level of significance was defined as P < 0.05.

RESULTS

The 2 groups' demographics, tumor characteristics, and radiation dose data are summarized in Table 2. There were no significant differences with respect to patient age, weight, height, body mass index, prevalence of diabetes and hypertension, menstruation state, maximal tumor diameter, position, or concomitant signs (P > 0.05) between the MEI (+)300 group and CTA370 group. According to our protocol, the mean cumulative CT dose index volume and the mean dose length product were not significantly different between the MEI (+)300 group (5.35 ± 1.10 mGy, 160.6 ± 71.94 mGy cm, respectively) and the CTA370 group (5.02 ± 1.87 mGy, 156.3 ± 74.7 mGy cm) (P > 0.05). The MEI (+)300 group intake (195 mg/kg) has less iodine content than CTA370 group.

TABLE 2

Patient Demographics, Tumor Characteristics, and Radiation Dose of 48 Patients With Breast Cancer in MEI (+)300 and CTA370 Groups

Characteristics	MEI (+)300 Group (n = 25)	CTA370 Group (n = 23)	P
Age, y	50.92 ± 9.25	50.58 ± 10.06	0.90
Weight, kg	56.58 ± 4.73	56.33 ± 7.26	0.89
Height, cm	154.11 ± 6.20	154.98 ± 5.12	0.98
BMI, kg/m2	23.90 ± 2.19	23.89 ± 3.80	0.98
Diabetes	2 (8.0%)	1 (4.3%)	1.00
Hypertension	3 (12.0%)	2 (8.7%)	1.00
Menstruation state
Premenopausal women	13 (52.0%)	13 (56.5%)	0.75
Postmenopausal women	12 (48.0%)	10 (43.5%)
The maximal tumor diameter, cm	3.06 ± 0.95	2.60 ± 1.57	0.23
Mass position
Outer upper quadrant	14 (56.0%)	10 (43.5%)	0.85
Outer lower quadrant	5 (20.0%)	6 (26.1%)
Inner upper quadrant	3 (12.0%)	3 (13.0%)
Inner lower quadrant	3 (12.0%)	4 (17.4%)
Concomitant signs
Nipple depression/skin thickening	3 (12.0%)	0 (0%)	0.24
No. cases of positive lymph nodes	6 (24.0%)	1 (4.3%)	0.10
DLP	5.35 ± 1.10	5.02 ± 1.87	0.66
CTDI vol	160.6 ± 71.94	156.3 ± 74.7	0.17

Data are presented as mean ± SD.

BMI indicates body mass index; CTDI vol, CT dose index volume; DLP, dose length product.

The mean intraclass correlation coefficient, calculated to quantify interobserver variability, was 0.937 (0.831–0.978). As the x-ray energy decreased, the mean attenuations (Fig. 1A), SDs (Fig. 1B), SNRs (Fig. 1C), and CNRs (Fig. 1D) of the 5 arterials supply of NAC gradually increased. The highest CNR and SNR were observed at 40 keV among the MEIs (+)300 and standard linear-blended polyenergetic image. Thereafter, we used the 40 keV MEI (+)300 for comparison. Compared with the CTA370 group, the 40 keV MEI (+)300 yielded significantly higher attenuation in the 5 arterials supply of that NAC (428.72 ± 125.80 vs 225.00 ± 50.15, P < 0.001) but also resulted in higher noise (53.37 ± 23.54 vs 18.79 ± 11.53 P < 0.001). Both the SNR and CNR were higher in the 40 keV MEI (+)300 (SNR, 64.78 ± 28.50 vs 26.29 ± 8.68; CNR, 55.92 ± 27.15 vs 18.79 ± 7.46; both P < 0.001) than in the CTA370 group (Table 3).

FIGURE 1

Quantitative indexes across the range of virtual monochromatic energy levels (40–80 keV) and PEI reconstructed from DECT in the MEI (+)300 group: (A) attenuation, (B) SD, (C) SNR, and (D) CNR. Maximum attenuation, SD, SNR, and CNR occurred at 40 keV MEI (+)300. PEI, standard linear-blended polyenergetic image (60% of the low-kilovolt spectrum and 40% of the high-kilovolt spectrum).

TABLE 3

Objective Analysis of Images Between the 40 keV MEI (+)300 Group and CTA370 Group

Parameters	40 keV MEI (+)300 Group	CTA370 Group	P
Attenuation
Axillary artery	591.83 ± 87.71	256.74 ± 48.54	<0.001
Thoracoacromial artery	424.05 ± 103.69	216.09 ± 45.71	<0.001
Lateral thoracic artery	368.80 ± 89.53	214.74 ± 45.04	<0.001
Internal thoracic artery	348.02 ± 105.92	215.52 ± 45.85	<0.001
Intercostal artery	410.91 ± 74.16	221.91 ± 55.75	<0.001
Mean attenuation	428.72 ± 125.80	225.00 ± 50.15	<0.001
SD
Axillary artery	44.70 ± 14.89	18.34 ± 8.9	<0.001
Thoracoacromial artery	50.05 ± 24.09	16.56 ± 12.75	<0.001
Lateral thoracic artery	63.82 ± 31.44	15.39 ± 8.00	<0.001
Internal thoracic artery	49.7 ± 18.54	18.82 ± 8.67	<0.001
Intercostal artery	58.44 ± 21.96	24.82 ± 15.87	<0.001
Mean noise	53.37 ± 23.54	18.79 ± 11.53	<0.001
SNR	64.78 ± 28.50	26.29 ± 8.68	<0.001
CNR	55.92 ± 27.15	18.79 ± 7.46	<0.001

Data are presented as mean ± SD.

A total of 58 and 50 source vessels that provided the dominant blood perfusion to NAC tissues were identified in the 40 keV MEI (+)300 group and CTA370 group, respectively, and the average number of dominant arteries per breast was 2.32 ± 0.75 and 2.17 ± 0.72 (P = 0.494), respectively (Table 4). Among the source vessels in the 2 groups, 24 (41.4%) and 22 (44.0%) were identified as the internal thoracic artery (Fig. 2), 18 (31.0%) and 16 (20.0%) as the lateral thoracic artery (Fig. 3), 11 (18.9%) and 8 (16.0%) as the thoracoacromial artery, and 5 (8.6%) and 4 (8.0%) as the intercostal artery. No vessels from the axillary artery were identified as contributing to NAC perfusion in this study (P = 0.981).

TABLE 4

Subjective Analysis of Images Between the 40 keV MEI (+)300 Group and CTA370 Group

Parameters	40 keV MEI (+)300 Group	CTA370 Group	P
No. dominant arteries	58	50
Axillary artery	0	0	0.981
Thoracoacromial artery	11	8
Lateral thoracic artery	18	16
Internal thoracic artery	24	22
Intercostal artery	5	4
Average dominant arteries per breast	2.32 ± 0.75	2.17 ± 0.72	0.494
Image quality	4.64 ± 0.49	4.78 ± 0.42	0.287
Visualization	1.76 ± 0.44	1.82 ± 0.39	0.583

Data are presented as mean ± SD.

FIGURE 2

Reformatted maximum-intensity projection images of a 52-year-old woman and a 47-year-old woman with right breast cancer obtained by CTA370 (A) and 40 keV MEI (+)300 (B), respectively. The quality of the CTA370 image is slightly better than that of the 40 keV MEI (+)300. The visualization of the internal thoracic artery was comparable between CTA370 and 40 keV MEI (+)300. The arrow points to the internal thoracic artery.

FIGURE 3

Reformatted maximum-intensity projection images of a 50-year-old woman with right breast cancer and a 54-year-old woman with left breast cancer obtained by CTA370 (A) and 40 keV MEI (+)300 (B), respectively. Visualization of the thoracoacromial artery (arrow a), lateral thoracic artery (arrow b), and axillary artery (arrow c) was excellent in both groups.

The mean ± SD image quality scores of the 40 keV MEI (+)300 group and CTA370 group were 4.64 ± 0.49 and 4.78 ± 0.42, respectively. This difference was not significant (P = 0.287), indicating that the quality of the images from the 40 keV MEI (+)300 group was comparable with that of the images from the CTA370 group (Table 4). The mean visualization score of the arterials supply to NAC in the CTA370 group was slightly higher than that in the 40 keV MEI (+)300 group (1.82 ± 0.39 and 1.76 ± 0.44, respectively), but this difference was not statistically significant (P = 0.583) (Table 4).

DISCUSSION

Our study demonstrated that 40 keV MEI (+)300 acquired from DECT can improve image quality of the arterial supply to the NAC with low-concentration and low-flow-rate contrast media in breast cancer. We observed greater intravascular attenuation, SNR, and CNR with 40 keV MEI (+)300 acquired from DECT; at the same time, the image noise was also increased, which resulted in a comparable subjective image quality score and visualization of the 5 arterials supply to the NAC compared with those of conventional CTA370.

Dual-energy CT provides a wide range of postprocessing functions for evaluating DECT angiography data sets.[21] Previous studies have shown that DECT angiography has potentially greater diagnostic accuracy in the evaluation of vascular stenosis than conventional CTA.[22,23] In addition, DECT angiography can be beneficial for virtual bone removal to improve vascular visualization[24] and in reconstructing virtual, unenhanced images, which has been shown to reduce radiation exposure.[25] To minimize patient exposure to ionizing radiation, our DECT protocols operate using an online dose modulation system (CARE Dose4D; Siemens Healthineers) that adapts the tube current to the patient's anatomy.[26] Low–kiloelectron volt MEIs (+) derived from DECT angiography data allow an improved balance between iodine attenuation and image noise and a reduced amount of contrast media a priori[27] by assessing the CNR and SNR, and have shown superior image quality in DECT angiography. Several studies have shown that the iodine load for abdominal CTA can be reduced by up to 50% with the use of MEI at 40 to 60 keV (reduction in iodine dose, −49%,[28] −27%,[29] and −28%[30]) compared with single-energy CT while providing an equivalent or improved CNR as well as superior subjective image quality. In accordance with those studies, the optimal kiloelectron volt was 40 in the present study, closer to the K-edge of iodine (33 keV); therefore, the attenuation of the arteries could be increased even at a lower concentration of contrast media. We chose a low flow rate to reduce the risk of iodine extravasation. In this study, no extravasation was observed in all patients.

In vast majority of studies, the dominant blood supply to the NAC seems to be attributed to the internal thoracic artery and/or the lateral thoracic artery. One study[31] showed that the internal thoracic artery is the most important breast blood supply source, and even within the same individual, blood supply types are diverse and asymmetrical. Later, another study,[4] based on the results from a microdissection technique, demonstrated that the lateral thoracic artery is the most dominant blood supply source of the NAC, followed by the internal thoracic artery. In our study, the internal thoracic artery and the lateral thoracic artery accounted for 72.4% (42/58) and 76.0% (38/50) of the dominant blood supply to the NAC in the 2 groups, respectively, consistent with previous studies. In the 2 groups, we observed no difference in the number of average dominant arteries per breast for different contrast media protocols.

Recently, a noise-optimized MEI (+) reconstruction algorithm was developed to perform spatial frequency-based recombination, which results in the highest possible image contrast by reducing the image noise at lower energy levels.[32] In our study, image noise was increased in the 40 keV MEI (+)300 group compared with that of the CTA370 group, which is consistent with previous results.[16] However, subjectively, there was no statistically significant or clinical difference in the perceived image noise between the 2 groups, especially in the evaluation of the arteries. This may be due to that it is difficult to visually separate additional image spots when arterial attenuation is significantly improved.[33] These data show that image noise should not be an obstacle to the use of low kiloelectron volt. Computed tomography angiography can be performed effectively with a low concentration and flow rate of contrast media without a clinically significant noise penalty, which is of great benefit to patients with breast cancer with impaired renal function or suboptimal venous access. Studies involving patients with specific breast cancers will help provide further insight into the relative value of the 2 DECT angiography techniques with different concentrations of contrast media.

This study has several important limitations. First, we limited this investigation to the analysis of intravascular attenuation, SD measurements, and subjective scoring, comparing a 40 keV MEI (+)300 group and a CTA370 group, where the imaging parameters were selected to match the radiation dose. We did not attempt to assess diagnostic accuracy or vascular findings during the clinical operation. Second, we compared 40 keV MEI (+)300 with CTA370 images, which were derived from DECT rather than single-energy CT; the conclusions obtained here were identical to those obtained for single-energy CT-derived images. Third, our study included a relatively small patient cohort, and we did not evaluate the traditional monoenergetic reconstruction algorithm, which has shown poor results at low kiloelectron volt levels.

In summary, our study demonstrated that 40 keV MEI (+)300 acquired from DECT can improve image quality of the arterial supply to the NAC with low-concentration and low-flow-rate contrast media in breast cancer. Duel-energy CT with low-concentration and low-flow-rate contrast media can be used not only to routinely screen lung metastases of breast cancer but also to provide valuable information about the blood supply to the NAC to aid in the design of custom treatments through MEIs (+). This method can reduce the iodine dose for patients with breast cancer.