1) Understand the basic principles of iterative reconstruction for CT. 2) Describe commercially available iterative reconstruction techniques. 3) Review the advantages and disadvantages of iterative reconstruction. 4) Discuss the incorporation of iterative reconstruction algorithms into clinical protocols for CT angiography.

To evaluate objective and subjective image quality of CT venography (CTV) using 80 kVp with model-based iterative reconstruction (MBIR) and compare with filtered back projection (FBP) and adaptive statistical iterative reconstruction (ASIR)

This retrospective study was approved by our institutional review board. Total 44 consecutive patients (mean age: 56.1 ± 18.1) who were underwent CTV using 80 kVp for evaluation of deep vein thrombosis (DVT) were enrolled. The same raw data were reconstructed using FBP, ASIR and MBIR. Objective image quality {vascular enhancement, noise, contrast to noise ratio (CNR)}were measured in the inferior vena cava (IVC), femoral vein (FV) and popliteal vein, respectively by independent two radiologists who blinded to reconstruction method. In addition, subjective image parameters (image quality, image noise, confidence to find DVT) were assessed using a 5 point scale system by same two radiologists independently. Data were analyzed using repeated measures ANOVA and effective dose was estimated using the dose-length product (DLP)

All images reconstructed by MBIR were acceptable for diagnosis and DVT was diagnosed in 8 patients (18.2%). The mean CNR of MBIR was significantly higher than those of FBP and ASIR in the IVC, FV and popliteal vein and images in MBIR had significantly lower objective image noise (p < 0.01). However, mean vascular enhancement of MBIR was not significantly different from those of FBP and ASIR. In addition, the subjective image quality and confidence of find DVT of MBIR was significantly higher than those of the others (p < 0.01) while MBIR had the lowest score for subjective image noise (p < 0.01). The mean DLP was 364.31 ± 61. 20 mGy cm

CTV using 80 kVp with MBIR provided diagnostic acceptable image quality for evaluation of DVT with low radiation dose and it was superior to FBP and ASIR in objective and subjective image quality

CTV using 80 kVp with MBIR will be optimal protocol for diagnosis of DVT with lowest radiation dose

Assessment of deep inferior epigastric artery (DIEA) perforating branches for free tissue transfer is challenging given their small size and low contrast (or attenuation) difference relative to abdominal wall musculature. The purpose of this study was to determine if high matrix size (1024 x 1024) model based iterative reconstruction (MBIR1024) improves spatial resolution and image noise characteristics compared to 512 x 512 filtered back projection reconstruction (FBP512) and if this can translate into improved visualization of DIEA branches.

Spatial resolution was assessed with a line-pair phantom and by calculating the modulation transfer function (MTF) from a wire phantom. Sixteen subjects who underwent high resolution CTA of the abdomen and pelvis were retrospectively identified. Axial images were reconstructed using FBP512 and MBIR2014. Regions of interest (ROI) were drawn in rectus muscles, external iliac arteries (EIA) and DIEAs. Signal-to-noise ratios (SNR) and contrast-to-noise ratios (CNR) were calculated using noise within rectus muscles as a reference. Two independent readers rated the visualization of 3 DIEA branches (4-point scale) on each side in all 16 subjects using axial thick-slab MIP images reconstructed from FBP512 and MBIR1024 studies in random order. Statistical comparisons were performed using paired t-tests.

Phantom studies reveal improved spatial resolution of MBIR1024 over FBP512 (10 vs 7 line-pairs/cm, f50 of MTF equals 0.61 mm-1 vs 0.41 mm-1). MBIR1024 results in higher SNR and CNR compared to FBP512 for all ROIs tested (p < 0.001). Of specific interest, mean CNR for DIEAs vs rectus muscles was 28.7 for MBIR1024 in comparison to 15.9 for FBP512. On visualization analysis, DIEA branches scored on average higher on MBIR1024 than on FBP512 (14.06/24 for MBIR1024 and 12.25/24 for FBP512, p < 0.001).

MBIR1024 provides improved spatial resolution while simultaneously reducing image noise in comparison to FBP512. These superior image quality characteristics improve contrast between DIEAs and rectus muscles and results in better visualization of DIEA perforators for pre-surgical assessment prior to free tissue transfer.

Superior spatial resolution and image quality of high matrix size MBIR improves visualization of deep inferior epigastric arteries, which may aid in pre-surgical planning for free tissue transfer.

Evaluate the ability of a novel Spectral Detector CT (SDCT) to augment aortic attenuation of low attenuating exams and assess whether the resulting monoenergetic image set enables diagnostic evaluation of vascular structures.

On an SDCT (Philips Healthcare) 49 patients underwent contrast-enhanced examinations of the chest and abdomen following routine or CTA protocols. A subset of 36 examinations were chosen which had mean aortic attenuation of less than 200 HU calculated as the average of 8 regions of interest within the aorta. Subjective evaluation of vascular enhancement and overall image noise was graded on a 5-point scale (1= Non-diagnostic, 5 = Excellent). Monoenergetic image sets of 40 to 180 keV at every 10 keV were created. An ideal monoenergetic energy level was chosen defined as the highest energy that provided mean aortic attenuation greater than 200 HU and maintained diagnostically acceptable subjective image noise and vascular enhancement. At this energy level attenuation, noise, and signal-to-noise ratios as well as subjective image quality and vascular enhancement were compared to the standard 120 kVp polyenergetic study. Paired t-test and Wilcoxon signed-rank test were used for analysis.

34 studies (94%) met criteria for successful optimal monoenergetic reconstruction, with a mean optimal energy of 57 ± 6.4 keV. Optimizing energy levels significantly increased aortic attenuation compared to the 120 kVp studies by an average of 66% (238 ± 25 vs. 144 ± 25 HU; p<.001). Image noise averaged 16% higher in the monoenergetic energy studies than the 120 kVp exams (19.6 ± 3.9 vs. 16.9 ± 4.1; p<.001), while SNR was 41% higher (13.3 ± 2.9 vs. 9.4 ±2.5; p <.001). Subjective vascular enhancement was significantly higher on the monoenergetic images (4.7 vs 2.4, p < .001). While subjective image noise increased (4.2 vs. 5.0, p < .001), the resulting image set was diagnostically acceptable.

SDCT enables retrospective creation of an optimal monoenergetic image set which achieves attenuation levels comparable with dedicated aortic angiographic exams and remains acceptable for diagnosis despite increased image noise.

SDCT enables the creation of retrospective aortic angiograms from studies that otherwise may not be diagnostic, suggesting the ability to create angiograms from routine or low contrast studies.

Previous studies demonstrated that calculated low keV monoenergetic datasets from Dual energy (DE)CT abdominal angiography imaging does not significantly improve contrast-to-noise ratio (CNR) when compared to polyenergetic images (PEI), which is due to the dramatic increase in image noise at lower keV levels. The recently introduced frequency-split technique combines the lower spatial frequency stack at low keV for high contrast with the high spatial frequency stack for image noise at high keV levels to calculate noise-reduced images at ultra-low keV levels below 70 keV. The aim of this study was to compare objective image quality indices in DECT angiography studies of the abdomen using conventional polyenergetic images (PEIs) and virtual monoenergetic images (MEIs) at different keV levels using the frequency-split technique.

20 patients (14 male; mean age 70±10 years) who underwent DECT angiography of the abdomen were retrospectively included in this study. MEIs from 40 to 120 keV were reconstructed using the frequency-split technique. Signal intensity, noise, signal-to-noise ratio (SNR) and CNR were assessed in infrarenal aorta, superior mesenteric and external iliac arteries. Comparisons between MEIs and PEIs were performed using a Mann-Whitney U test.

60 arteries were evaluated. 40 keV MEIs resulted in significantly higher signal intensity (+254% on average; all p < 0.05) compared to PEIs at slightly higher SNR (+7% on average; p > 0.05). Highest CNR values were found in 40 keV MEIs (9 ± 0.8 HU, 7 ± 1.6 HU, 9 ± 1.4 HU in the infrarenal aorta, superior mesenteric and external iliac arteries, respectively), which were significantly higher compared to PEI (7 ± 1 HU, 5 ± 1.5 HU, 7 ± 1.5 HU in the infrarenal aorta, superior mesenteric and external iliac arteries, respectively, all p < 0.05).

Combining the lower spatial frequency stack for contrast at low keV levels with the high spatial frequency stack for noise at high keV levels leads to improved image quality of ultra-low keV monoenergetic abdominal DECT datasets when compared to previous monoenergetic reconstruction techniques without the frequency-split technique.

With a frequency split approach, 40 keV MEIs provide improved objective image quality in DECT abdominal angiography compared to standard PEI. Their additional reconstruction might improve diagnostic accuracy.

To evaluate image quality and radiation dose saving potential of a CT scan protocol that uses a detector with integrated signal-transformation, a high-pitch acquisition technique, automatic kVp selection and an iterative reconstruction algorithm.

Between January and April 2014, 55 consecutive patients underwent CT angiography (CTA) of the chest, abdomen and pelvis on a third-generation dual-source CT using an ultra high-pitch (73 cm/s) acquisition protocol, topogram-based automatic kVp selection and mAs modulation in combination with a model based iterative reconstruction algorithm (group 1). As a control group, 55 patients who had undergone CTA with comparable scan ranges on a second-generation dual-source CT in 2012 and 2013 were matched according to gender, age and BMI (group 2). In all patients, 20 vascular segments were analyzed for attenuation and image noise by two readers on 3.0mm slices. Signal-to-Noise ratio (SNR) and Contrast-to-Noise ratio (CNR) were calculated for all segments. Dose-Length-Product (DLP) was documented to calculate effective dose.

There were no significant differences in age, weight or BMI between groups. In group 1, tube potential was 90kVp, 80kVp and 70kVp in 29 (53%), 24 (43%) and 2 (4%) patients. In group 2, tube potential was 140kVp, 120kVp and 100kVp in 7 (13%), 29 (53%) and 19 (34%) patients. Across all vascular segments, SNR and CNR were considerably higher in group 1 (SNR: 29.6 ± 3.3 vs. 21.0 ± 3.0; CNR: 26.7 ± 3.9 vs. 18.3 ± 2.9, both p's < 0.01). DLP, on the other hand, was substantially lower in group 1 (230 ± 63 vs. 391 ± 166 cm*mGy, p<0.001).

In third-generation dual-source CT scanners, the combination of an integrated detector design, a high-pitch acquisition technique, automatic kVp selection and iterative-reconstruction algorithms results in dose reductions of at least 40% in comparison with earlier scanner generations at preserved image quality.

This study highlights the importance of a rapid and successful translation of the considerable engineering progress of the last years into radiation dose reduction and thus patient benefit.

1) Understand the basic principles and technical basics of dual energy CTA. 2) Describe two components of dual energy imaging including material decomposition and virtual monochromatic spectral imaging. 3) Review the experimental studies and discuss the potential clinical application to vascular systems.

1) Understand the basic principles and technical basics of dual energy CTA. 2) Describe two components of dual energy imaging including material decomposition and virtual monochromatic spectral imaging. 3) Review the experimental studies and discuss the potential clinical application to vascular systems.

The aim of our study was to reduce the radiation dose exposure and contrast medium volume in the evaluation of abdominal aorta disease, using low-kV setting CT-angiography (CTA) protocol.

A total of 84 patients(35 women and 49 men;mean age 66.5years;range,37-86years) with abdominal aorta disease, from June 2011 to December 2013,were prospectively enrolled. All patients underwent 256MDCT scan examination of abdominal aorta(Brilliance-iCT,Philips,NL).Forty-six patients were evaluated using low-dose radiation protocol (100kV;automated tube current modulation) and ultra low-contrast volume (40ml; 4ml/s;350mgI/ml). A control group of 38 patients underwent standard CTA protocol(120kV;automated tube current modulation) with standard contrast volume(80ml). Intravessels density measurements(HU) were performed manually drawing a region of interest(ROI) in the lumen of abdominal aorta,renal arteries and common iliac arteries. The radiation dose exposure(dose-length product, DLP;CT dose index, CTDIvol) and the signal-to-noise-ratio(SNR) were also calculated. The data were then compared and statistically analyzed.

All exams were presented high diagnostic quality, permitting thus the correct visualization of the main aortic branches and vascular wall. Higher density measurements were obtained in low-kV protocol as compared to control group, in abdominal aorta(mean attenuation value 343HU vs 320HU), renal arteries(332HU vs 298HU) and common iliac arteries(325HU vs 320HU). No significant noise increase was observed in the study group (mean signal to noise ratio,SNR 15,2) in comparison to control group (SNR 18,6). A significant(p<0,05) overall reduction of 65% in radiation dose exposure was obtained using low-kV protocol(DLP335mGy*cm,CTDIvol5,8mGy) as compared to control group (DLP973mGy*cm;CTDIvol19,4mGy).

Low-kV CTA by using 256 MDCT scanner permits to significantly reduce the radiation dose exposure (over 65%) and the amount of contrast media volume injected, maintaining high diagnostic quality as compared to standard protocol, reducing also the risk of renal damage.

Low kV protocol with low contrast media volume reduces the radiation exposure, preserving renal function and providing an effective tool for the evaluation of patients with abdominal vascular disease.

Reduction of iodinated contrast load for CTA is of importance in azotemic patients, especially those who are not dialysis dependent, and may be achieved with the improved iodine absorption of lower kVp techniques. We compared quantitative and qualitative scan data in two abdominopelvic CT Angiography (AP-CTA) protocols, one with a 50% reduced dose of iodinated contrast @ 80 kVp, the other with a standard dose @ 120 kVp. Radiation dose was also compared between the two techniques.

A chart review was performed yielding 103 patients who underwent AP-CTA on 64 or 256 channel CT scanners with 320-350 mg I/mL contrast: 49 with 30-60 mL (mean 48) at 80 kVp, and 54 controls with 90-100 mL (mean 98) at 120 kVp. Objective image quality parameters included arterial attenuation, noise, and contrast to noise ratio (CNR) measured at 6 levels in the aorta and iliac arteries. Subjective assessment of image quality (IQ), enhancement intensity (EI) and image noise (IN) was scored blindly at the same levels by three attending radiologists on a 3 point scale. Iterative reconstruction (iDose, Philips Healthcare, Cleveland, OH) was utilized in 18 and 49 of the 120 and 80 kVp groups respectively. Radiation dose (CTDIvol and DLP) was recorded in each case.

The aggregate of 6 level CNR for the 120 kVp group was 12.4 ± 4.6, compared to 11.8 ± 7.0 for the 80 kVp group (p =0.60). The 120 kVp technique scored better for subjective IQ (p=0.042) and IN (p=0.004) but not for EI (p=0.205). No study had to be repeated. Radiation dose was significantly higher in the 120 kVp than 80 kVp group [DLP 863 ± 344 and 482 ± 206 (mGy*cm) respectively (p<0.0001)]. In the 80 kVp group, serum creatinine pre-and post AP-CTA (means 3.7 and 3.4 mg/dl respectively) showed no significant change (p=0.172).

Based upon objective CNR and subjective EI scores, the 80 kVp technique with 50% lower iodine contrast dose allowed satisfactory AP-CTA studies without neprotoxicity in azotemic patients. Use of iterative reconstruction in all 80 kVp group cases contributed to a 43% mean radiation dose reduction. Negative but potentially reversible sequelae of this major drop in radiation dose included increased IN and reduced subjective IQ.

In azotemic patients, 80 kVp technique using 50% lower iodinated contrast dose and iterative reconstruction allows safe and satisfactory AP-CTA studies with major radiation dose reduction.

To evaluate the diagnostic performance of a triple bolus dual-phase acquisition protocol (unenhanced/combined artero-venous-excretory phases) for the preoperative assessment of kidney anatomy in renal laparoscopic living donors.

Twenty consecutive patients, referring to our department to undergo CT evaluation prior to living donor nephrectomy, were enrolled in our single-center prospective randomized comparison of 2 CT-acquisition protocols. Ten patients (Group A) underwent standard quadri-phase CT-examination (unenhanced, arterial, venous, and delayed phases) during a single injection bolus of 100 ml of contrast medium whereas the last 10 patients (Group B) underwent a dual-phase CT protocol (unenhanced and combined artero-venous-excretory phases). Combined artero-venous-excretory phase (Renal Triple-Rule-Out) was performed with a triple split-bolus injection protocol (CM: 30+30+40 ml, @4mL/sec) and an optimized time delay triggered to obtain both artery, veins and renal pelvis opacification at the same time. CT-images were quantitatively and qualitatively compared by two blinded independent readers. The 2 protocol were also compared in terms of diagnostic performance using the surgical assessment during nephrectomy, as gold standard.

All CT examinations were considered technically adequate and no complications occurred. Significantly higher vascular attenuation values (renal arteries and veins) were obtained in Group A. No significant differences were noted in terms of image quality with either axial source images or 3D reconstructions. Likewise, no significant differences were found among the two protocols in terms of noise. No beam-hardening artifacts due to renal pelvis opacification affecting image interpretation were found. No significant differences were found among the two groups with regard to diagnostic performance. Overall dose reduction of 60% was achieved in Group B.

Renal Triple-Rule-Out CT Protocol by using a triple-bolus injection protocol is feasible and effective in the preoperative planning of laparoscopic living kidney donors, without compromising image quality and diagnostic performance with a substantial reductions of radiation dose.

Renal-Triple-Rule-Out CT protocol may allow a significant reduction in radiation burden in renal laparoscopic living donor without affecting an accurate pre-treatment planning.

A recently proposed sign at CTA, the "gravitational gradient" (GG), quantifies the antero-posterior inhomogeneity of luminal contrast enhancement under the hypothesis that it may signify slow flow in large vessels. We thoroughly investigated the GG in infrarenal AAA as a source of hemodynamic information and assessed its relevance with respect to rapid AAA growth.

The GG was measured in 67 consecutive pre-repair CTAs and used to classify patients as high (≥1.4) or normal (<1.4) GG (Figure). We compared the two groups for a) cardiovascular (CV) status, including ejection fraction; b) CTA scan and contrast delivery parameters, including timing; c) aortic and AAA geometry; d) characteristics of the blood flow patterns using computational fluid dynamics in 5 high and 5 matched normal GG patients; and e) AAA growth >0.4cm/yr in a 3-year period. We also assessed GG change in post-repair CTA available in 10 high GG patients.

18% of AAAs (n=12) exhibited a high GG≥1.4. A high GG was independent of patient CV status, CT scan/contrast delivery parameters, and AAA/aortic geometry (all p>0.05). The only significant differences between patients with high vs normal GG were a sudden drop in aortic enhancement between the renal arteries and aortic bifurcation (43±38 vs 10±39 HU drop, p<0.05; Figure), and all CFD-derived parameters, namely extent of retrograde flow or stasis (10±6% vs 2±2% luminal area affected, p<0.05, Figure), helicity (5±2 vs 12±5 m2/s2, p<0.05), and vorticity (72±25 vs 109±27 s-1, p<0.05). In the 10 high GG patients that underwent repair, both the high GG and sudden drop in aortic enhancement across the AAA resolved (pre vs post GG=1.9±0.5 vs 1.0±0.1, p<0.05; HU drop=42±42 vs 20±20, Figure). Finally, a high GG was associated with a relative risk of 2.7 (95% CI: 1.3-5.3, p<0.05) for AAA growth >0.4cm/yr.

Inhomogeneous luminal contrast enhancement in infrarenal AAA CTA as detected by the GG is associated with complex hemodynamic patterns and rapid aneurysm enlargement. Its potential use to detect patients at risk of rapid AAA growth and rupture should be further explored.

The Gravitational Gradient is a readily identifiable imaging sign in AAA CTA that is associated with disturbed blood flow patterns and clinically significant disease progression. It may have potential use in surveillance and elective repair algorithms toward reducing rupture rates.

Aim of this study was to quantitatively assess the magnitude and direction of respiratory movement of the aorta and side branches.

This prospective study was approved by the institutional review board; informed consent was obtained from all patients. We performed a quantitative three-dimensional subtraction analysis of computed tomography during inspiration and expiration to establish the respiratory geometric movements of the aorta and side branches. 60 patients (42 men, 18 women, mean age 70 ± 9 years) with aortic disease were included. During breath-hold expiration and inspiration respectively, one millimeter (mm) slice thickness non-contrast enhanced and contrast-enhanced computed tomography of the aorta were performed respectively. By means of dedicated multiplanar reformation image subtraction software using the spine as reference point, position of relevant anatomic sections from both datasets were analyzed. These included the diaphragm domes, anterior thorax wall, ascending thoracic aorta (AAo), the origin of the left subclavian artery (LSO), descending thoracic aorta at level of tenth thoracic vertebra (DAo) as well as the origin of the renal arteries (RAO).

With inspiration, the regions of interest of the aorta and side branches moved in anterior, medial and caudal direction compared to the expiration state. Threshold for vessel displacement was at least five mm anterior-posterior thoracic excursion or ten mm diaphragm dome movement. Mean 3-dimensional movement (± standard deviation) was 8.9 ± 3.6 mm (AAo), 11.1 ± 3.9 mm (LSO), 4.9 ± 2.5 mm (DAo) and 1.4 ± 1.1 mm (RAO). There was significantly less movement in the DAo compared to LSO (p < 0.001). Correlation coefficient between extent of LSO displacement and thorax excursion was 0.78.

The aorta and side branches undergo considerable respiratory movement. This finding may be important for thoracic and complex thoraco-abdominal endograft designs as well as fusion image guidance during endovascular aortic repair.

Respiratory movement is an important contribution to understanding the aortic dynamics; it has implications especially for planning and implantation of endovascular thoraco-abdominal aortic repair.

1) To illustrate steps in image post-processing for the interpretation of CTA images. 2) To highlight elements that can be used to optimize workflow for multiplanar reformatted images, maximum intensity projections, and three-dimensional volumes.

During this presentation, the most current post processing techniques available for CT angiography will be reviewed. The audience will understand the application of these tools for optimal image interpretation. Imaging and workflow protocols will be introduced, with the emphasis on improving patient care, as part of a multidisciplinary team.