ABSTRACT

PURPOSE

Digital variance angiography (DVA), a recently developed image processing technology, provides a higher contrast-to-noise ratio (CNR) and better image quality during lower limb interventions than digital subtraction angiography (DSA). Our aim was to investigate whether the quality reserve of DVA can also be observed in uterine fibroid embolization (UFE).

METHODS

In this retrospective observational study, the CNR and image quality of DSA and DVA images from 56 patients (mean ± standard deviation age: 44.2 ± 5.3 years) who underwent UFE at our institution were assessed. For the visual evaluation of the same image pairs, the visibility of large vessels, small vessels, tissue blush, and background noise was compared by three experienced readers using a four-grade Likert scale. Data were analyzed using the Wilcoxon signed-rank test or the one-sample Wilcoxon test.

RESULTS

DVA provided significantly higher CNR than DSA (the median CNR_DVA/CNR_DSA was 1.96). In the visual comparison of DVA and DSA images, Likert scores did not significantly differ from zero (equal quality level) in any evaluated categories. The median (interquartile range) values were 0.00 (1.00) for large vessels, −0.33 (1.33) for small vessels, 0.00 (0.67) for tissue blush, and 0.00 (0.75) for background noise.

CONCLUSION

Although the visual image quality of DSA and DVA was identical, DVA provided a twofold CNR in UFE, indicating a significant quality advantage for this technology.

CLINICAL SIGNIFICANCE

The observed quality reserve may allow for dose management (reduction of applied radiation dose and/or contrast media), enhancing the safety of UFE for both patients and personnel.

Digital variance angiography (DVA) is a recently developed image processing alternative to digital subtraction angiography (DSA). The technology is based on the principles of kinetic imaging.¹ Although DVA uses the same unsubtracted acquisition as DSA, it does not use a mask but instead calculates the standard deviation (SD) of pixel intensities. This algorithm enhances the contrast media signal and suppresses background noise, thereby providing a higher contrast-to-noise ratio (CNR) and better image quality than DSA. The quality reserve of DVA has already been validated in lower limb angiography,^2-5 liver transarterial chemoembolization,⁶ prostatic artery embolization,⁷ and carotid angiography.⁸ As the quality advantage of DVA can be effectively used for dose management,^8-10 the aim of this study was to investigate whether DVA can improve the image quality of angiograms in transcatheter uterine fibroid embolization (UFE), which could serve as a basis for radiation dose management in this endovascular intervention.

Over the last 29 years, as stated in the 2015 CIRSE guideline, numerous publications have demonstrated that UFE is a viable alternative to hysterectomy for women who wish to preserve their uterus (level 1 evidence).^11-14 The 2021 ACOG guideline confirms these findings, stating that “uterine artery embolization (UAE) is recommended as an interventional procedure for the treatment of uterine leiomyomas in patients who desire uterine preservation and are counseled about the limited available data on reproductive outcomes”.¹⁵ In addition to fibroids, UAE has also been proposed as a minimally invasive alternative to hysterectomy for patients with symptomatic adenomyosis.^{16, 17}

UAE is much less invasive and a non-surgical alternative to myomectomy or hysterectomy; however, ionizing radiation is used to identify and access the uterine artery for the embolization procedure. The literature shows that radiation exposure doses remain below the threshold for any deterministic radiation risks. Despite these data, implementing the ALARA principle and minimizing the radiation dose as much as possible is of utmost importance in every interventional radiological procedure, especially in UAE, as many patients are women of child-bearing age. Many papers have discussed different dose reduction techniques, such as adjusting collimation, minimizing DSA runs, reducing frame rates, using PA projections, and employing dose optimization software.^{18, 19} The use of DVA has not yet been tested in UFE.

Methods

In this single-center retrospective observational study, 56 patients (mean ± SD age: 44.2 ± 5.3 years) were included who had previously undergone UFE at the Medical Imaging Center, Semmelweis University, Budapest, Hungary, between February 2021 and June 2022. All procedures were conducted in accordance with the 1964 Helsinki Declaration and its later amendments. The study was approved by the Regional Institutional Scientific and Research Ethics Committee, Semmelweis University, Budapest, Hungary (SE RKEB), approval no. 186/2022 on September 26, 2022. Due to the retrospective nature of the study, informed consent was waived.

Study Design

One pre-embolization posteroanterior (PA) pelvic acquisition was included from each patient. Pre-embolization acquisition was preferred as it depicts small arteries and tissue blush of the fibroid, thereby providing a better basis for performance comparison. DSA and DVA images were generated retrospectively from the same unsubtracted acquisition using the GE Advantage Workstation (GE Healthcare, Chicago, Ill., U.S.A.) and the Kinepict Medical Imaging Tool software (Kinepict Ltd., Budapest), respectively. Absolute CNR values and ratios were calculated for each image pair, and visual image quality was assessed by readers in a blinded and randomized manner using a 4-grade Likert scale.

Image acquisition

All procedures followed institutional protocols. UFE was performed on a GE Innova IGS 5 angiography system by an experienced interventional radiologist with over 20 years of experience. A 4F UF (Cordis, Miami Lakes, FL, U.S.A.) catheter was introduced via right femoral access. Aortography was performed to assess arterial filling of the fibroids. A Medrad Avanta (Bayer AG, Leverkusen, Germany) automated injector was used to inject 20 mL of iodinated contrast media (Ultravist 370, Bayer) at a flow rate of 10 mL/s. A 4F Cobra 1 Glidecath (Terumo, Leuven, Belgium) was positioned in the left uterine artery, followed by the right uterine artery. Hand injections of contrast media (3–6 mL) were performed for the selective angiograms into the uterine arteries. Standard PA pelvic acquisitions (2 fps) were obtained on both sides before and after embolization. DSA runs were saved on the GE workstation, and the unsubtracted files were later used to generate stacked DSA and DVA images as described above.

Contrast-to-noise ratio analysis

Regions of interest (ROIs) were defined on vessels and background regions using ImageJ (v.2.0.0-rc-68/1.52e, Creative Commons License, NIH). As adjacent regions of blood vessels often contained signals from small arteries or tumor blush, background ROIs were placed outside the fibroid area. Vascular and background ROIs were paired accordingly.

The CNR values were calculated for all ROI pairs individually using the following formula [21], where Mean_v and Mean_b represent the mean pixel intensity values of the vascular and background ROIs, respectively, and Std_b is the background SD:

 CNRDVA/CNRDSA ratios (R) were also calculated for each corresponding DVA and DSA ROI.

Visual evaluation

Evaluations were conducted by three interventional radiologists with at least 5 years of experience in UFE. The readers were not involved in the treatment of enrolled patients.

A randomized, paired evaluation was performed with corresponding DSA and DVA image pairs. The readers were blinded to the imaging modality. The diagnostic value of the acquisitions was compared based on the visibility of large vessels, small vessels, tissue blush (if visible), and the extent of background noise (Figure 1).

Diagnostic value was graded using the following 4-grade bidirectional Likert scale:

0. Identical

1. Slightly better/less noise

2. Clear-cut advantage/less noise, no interference with structures

3. Better in every aspect/less noise, no interference, background clear

Each image pair was evaluated only once during the survey, and scores were automatically collected in a database for later processing.

Radiation dose and total fluoroscopy time measurements

Radiation dose (total dose-area prod­uct - DAP) and total fluoroscopy time were gathered from the radiation dose information provided for each patient in the “X-ray Radiation Dose Report” of the GE Innova IGS 5 angiography system. Data are presented as median (interquartile range).

Statistical analysis

Calculations of CNR and R medians, along with interquartile ranges (IQR), were performed using Excel 2016 (Microsoft, Redmond, WA). CNR values were compared using the Wilcoxon signed-rank test (Prism 8.4.2, GraphPad).

For visual evaluation scores, the mean and standard error of the mean were calculated. Due to the non-Gaussian distribution of the data, the median and IQR were also determined. To assess potential differences between modalities, image pair scores were compared with 0 (equal quality level) using the one-sample Wilcoxon test. Interrater agreement was characterized by Kendall’s W value. The level of significance was set at P < 0.05 for all tests.

Results

Patients

Patients (n = 56, mean ± SD age: 44.2 ± 5.3 years) with previously diagnosed uterine fibroids received UFE treatment between February 2021 and June 2022 at the Medical Imaging Center, Semmelweis University, Budapest, and were retrospectively enrolled for image analysis in a consecutive manner.

Contrast-to-noise ratio results

A total of 695 ROI pairs were analyzed from 56 pre-embolization image pairs. The results of the CNR measurements are summarized in Table 1. The median CNR of DVA images was significantly higher than that of DSA images [29.55 (IQR: 24.96) vs. 16.23 (IQR: 13.24), Wilcoxon signed-rank test, P < 0.001). The R (CNRDVA/CNRDSA) value was 1.96 (IQR: 0.88) (Figure 2).

Visual evaluation results

Readers evaluated 56 DSA-DVA image pairs using the 4-grade bidirectional Likert scale, where 0 represented identical image quality. According to the score settings, negative values indicated an advantage for DSA, whereas positive values indicated an advantage for DVA (Table 2). The median (IQR) Likert scores were 0.00 (1.00) for large vessels, −0.33 (1.33) for small vessels, 0.00 (0.67) for tissue blush, and 0.00 (0.75) for background noise (Figure 3). None of these values were significantly different from zero (one-sample Wilcoxon test).

Despite moderate interrater agreement levels, ratings were significantly associated with large vessels (W = 0.568, P < 0.001) and small vessels (W = 0.502, P < 0.01). However, agreement was only slight and not significant for tissue blush (W = 0.285, P = 0.766) and background noise (W = 0.349, P = 0.378).

Radiation dose and total fluoroscopy time

Total DAP was 57.0 (21–284) Gy·cm², and total fluoroscopy time was 736 (360–1570) sec.

Discussion

Our aim was to investigate whether the previously described advantages of DVA can also be observed in UFE intervention. Therefore, we compared the CNR and visual performance of DSA and DVA images in this retrospective observational study. Our results show that DVA provides a significantly higher (about twofold) CNR than DSA, but there is no difference in the visibility of large vessels, small vessels, tissue blush, and background noise. The poor interrater agreement in the latter two categories might reflect that the judgment of tissue blush and background noise is even more subjective. These findings are partly inconsistent with previous observations, as earlier studies demonstrated that DVA was always superior to DSA in both parameters.^2-10 However, in the present study, angiography conditions were different, as the catheter was in the uterine artery, very close to the target area, ensuring a highly selective injection of contrast media, and the acquisition was performed at a standard radiation level. Under these conditions, DSA provides excellent visual representation, which cannot be outperformed (ceiling effect). Nevertheless, the improved CNR value clearly indicates the quality reserve of DVA.

Previous studies have demonstrated that the quality reserve of DVA can be effectively utilized for dose management. A reduction of dose/frame value by 70% provided non-inferior or superior image quality in lower limb interventions compared with full-dose DSA acquisitions.²⁰ A subsequent randomized controlled trial showed that applying a similar low-dose protocol reduced total DSA-related DAP by 63% and total procedural DAP by 46% without compromising image quality or the diagnostic value of angiograms.²¹ The quality reserve of DVA can also be used to reduce contrast media, as DVA provided non-inferior image quality in carotid angiography compared with full-dose DSA when only 50% of the contrast media amount was used.⁸ Our preliminary unpublished observation suggests that an 80% reduction in contrast media achieved through dilution still provides excellent image quality in UFE using DVA images, whereas the concomitant DSA images under the same conditions appear poor.

Our finding may have important clinical implications if further studies prove the relevance of the increased CNR and increased quality reserve. UFE is a good alternative for the treatment of uterine fibroids, as it presents less burden and less risk for patients than surgical solutions. Nevertheless, this endovascular intervention requires several X-ray angiography acquisitions and repeated injections of iodinated contrast media. These steps carry their own risks, including possible acute and long-term side effects of radiation and potential impairment of kidney function. Obviously, the dose management capabilities, especially the radiation dose reduction ability of DVA, could be very beneficial in UFE, as patients are often of reproductive age. In addition, lower radiation exposure would reduce the risk of radiation-induced occupational hazards for medical staff. The reduction of contrast media usage could also be advantageous by lowering the risk of contrast-induced nephropathy.

Our results reveal the potential of DVA for dose management in UFE; nevertheless, further clinical studies are required to validate these claims. Such a study has already been initiated at our center. The radiation dose from our center serves as a baseline for such a study; our data fall well within the range of recent literature [DAP (Gy·cm²; median, range): Nocum et al.²²: 113.1 (21.9–792); Lacayo et al.²³: 74.8 (0.32–795); our data: 57 (21–284); total fluoroscopy time (minutes, median, range): Nocum et al.²²: 11.1 (6.2–33.6); Lacayo et al.²³: 13.5 (5.7–104); our data: 12.2 (6.0–26.2)].

Our study has some limitations. Due to its retrospective observational nature, the acquisition protocol was predefined and optimized for DSA; therefore, we could not detect any differences in the visual performance of DSA and DVA images. The full validation of DVA in UFE requires prospective clinical trials in which the protocol is appropriately modified to achieve dose management and DVA images are available for the interventional radiologist in real-time in the operating room.²⁴

In conclusion, our data show that DVA has a substantial quality reserve in uterine artery angiography compared with the traditionally used DSA technology. Although a visual advantage was not observed in the current clinical setting, the twofold CNR of DVA images provides a solid basis for prospective clinical trials, where the dose management capabilities of DVA can be validated in the endovascular treatment of fibroids and adenomyosis. These trials aim to achieve a 70% reduction in dose/frame value while maintaining non-inferior or superior image quality, as already demonstrated in lower limb interventions. Thus, our study indicates that DVA has the potential to reduce the applied radiation dose during UFE for both patients and personnel.