Evaluation of precalculated attenuation correction map for preclinical cardiac PET/MR using a ¹H/²³Na surface coil

Abstract

Background

In preclinical PET/MR, attenuation correction (AC) uses a mix of pre-calculated attenuation maps accounting for animal cradles and organ segmentation obtained from whole-body MR using a volume coil. The development of X-nuclei such as ²³Na or ³¹P may benefit from high sensitivity surface coils, which could lead to inaccurate PET quantification. In this study, we evaluated the benefit of a pre-calculated attenuation map including a dedicated cradle enclosing a surface coil for ¹⁸F-FDG PET imaging.

Materials and methods

We developed a 3D-printed cradle (DC) embedding a 20 mm surface coil made of PLA (polylactic acid), coated with a thin cap of epoxy. An attenuation map was generated using computed tomography and integrated into PET reconstruction. To validate AC, we compared PET images to those obtained using a conventional cradle and a volume coil (CC). Various image quality metrics were evaluated in various phantoms including a NEMA NU-4 (recovery coefficients (RC), uniformity (%STD), spillover ratio (SOR)), a homogenous phantom (slice and inter-slice uniformity) and a resolution phantom (spatial resolution). Finally, cardiac ¹⁸F-FDG PET images acquired with the 2 cradles were compared in Sprague-Dawley rats.

Results

RC obtained using DC and CC configurations were not significantly different. However, the %STD was significantly increased with the DC (5.07 ± 0.18) vs. CC (3.29 ± 0.53, p = 0.0002) leading to a decreased CNR with DC. The SOR was similar between the two cradles. In homogenous phantom, there was a non-significant 1.58% underestimation of the PET signal near the surface coil and 5% overestimation on the opposite side when using the DC vs. CC. Uniformity between slice was significantly higher in DC than in CC (3.15 ± 1.2% for DC vs. 2.24 ± 0.7% for CC, p = 0.02) while uniformity inter-slice was similar in DC and CC (2.25 ± 0.83% for DC vs. 1.93 ± 0.44% for CC, p = 0.2). Spatial resolution was similar between DC and CC (axial: 1.80 vs. 1.63 mm, tangential: 1.62 vs. 1.79 mm and radial resolution: 1.53 vs. 1.76 mm for DC and CC respectively). In-vivo, cardiac standardized uptake values were similar between the two cradles (3.55 ± 0.89 and 3.59 ± 1.02 for DC and CC respectively, p = 0.81).

Conclusion

Pre-calculated attenuation map using a standardized positioning of MR surface coil provided similar ¹⁸F-FDG PET images compared to a conventional PET/MR system with a volume coil, allowing its usage for combined PET and X-nuclei MR.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40658-025-00832-6.

Background

Preclinical positron emission tomography (PET) allows in vivo characterization of various animal models of human diseases, including cardiovascular diseases [1]. The strength of PET imaging relies on its high sensitivity and its quantification capabilities of radiopharmaceuticals uptake. On the other hand, MR offers high-resolution structural imaging with a better soft tissue contrast than computed tomography (CT) [2–4], and enables various forms of structural and functional imaging within one single session. Especially, x-nuclei imaging reveals underlying changes in physiological processes, whereas ¹H imaging focuses on spontaneous contrast between tissues with different relaxation properties [5]. Sodium (²³Na), phosphorus (³¹P), and fluorine (¹⁹F) are the most commonly used x-nuclei, and integration of spectroscopic imaging in PET systems may enhance the potential of hybrid PET/MR [6].

In the context of x-nuclei MR imaging, the use of a transmit and receive (Tx/Rx) surface coil may be preferred as they offer a compact design, lower electronic noise and a 4-fold cheaper cost compared to volume coils. In most PET/MR scanners, attenuation correction is usually based on organ segmentation obtained from a whole-body MR acquisition using a volume coil and pre-defined attenuation coefficients assigned to the segmented regions [7]. However, the presence of a surface coil within the PET/MR field-of-view (FOV) introduce additional sources of attenuation that may not be accounted for in MR-based AC maps. Regarding the impact of animal cradles, made of proton-free materials, most scanners include pre-calculated attenuation maps accounting for the cradles to improve the accuracy of PET quantification. In the perspective of using a surface coil in order to combine PET with MR spectroscopy, we hypothesized that standardizing the positioning of the surface coil and including it with the cradle in a pre-calculated attenuation map would enable accurate PET quantification.

To achieve this goal, we designed a dedicated cradle integrating the surface coil and included attenuation correction for both the cradle and the surface coil during the reconstruction process. ¹⁸F-FDG PET results were compared to images obtained using a conventional cradle and a volume coil, in both phantoms and rodents.

Materials and methods

PET-MR system characteristics

All experiments were performed using a dedicated preclinical hybrid PET/MR system (BioSpec 70/18, Bruker BioSpin MRI GmbH, Ettlingen, Germany) associating a 7T magnet and an MR-compatible PET detector. The PET insert consists of three rings of detector blocks, each incorporating eight monolithic LYSO crystals (50 × 50 × 10 mm²) coupled to a 12 × 12 array of silicon photomultipliers (SiPMs), each with an active area of 3 × 3 mm² and spaced at a 4.2 mm pitch. This configuration provides an axial and transaxial field of view (FOV) of 15 cm and 8 cm respectively [8]. The PET insert achieves a maximum sensitivity of 11.0% at the center of the FOV and an average energy resolution of 17% [9]. Experiments were performed using alternatively a conventional cradle provided by the manufacturer, and a dedicated 3D-printed cradle (DC) embedding a 20 mm surface coil made of PLA (polylactic acid), coated with a thin cap of epoxy (NUMP, Verson, France).

Phantom study

A series of 3 different phantoms designed for quality control of preclinical imaging systems was used in the study: (i) a fillable PMMA homogeneous phantom (NMI PHAN IM QA, Bruker, Ettlingen, Germany) (ii) a NEMA NU-4 complying with the NEMA NU 4-2008 standard [8], and (iii) a co-registration phantom (Fig. 1). The phantoms were positioned in exactly the same way in both the DC and the CC.

Fig. 1.

Phantoms used in the study, a Homogeneous phantom, b NEMA NU-4 phantom, c Resolution phantom

The NMI PHAN IM QA phantom was used to evaluate within and across slices uniformity in reconstructed images. The NEMA NU-4 phantom has an inner diameter of 30 mm, consists of three distinct regions, and was used for evaluating the performance of small-animal PET scanners, including the assessment of uniformity, spillover ratio (SOR) and recovery coefficients (RC). The first region included two cylinders respectively filled with air and non-radioactive water allowing for the evaluation of the SOR. SOR was evaluated as an indicator of possible incorrect classification of air/tissue interfaces potentially affecting quantification. The second region consists of a uniform hot cylinder, which was also used to evaluate uniformity in the reconstructed images (STD (%)). The third region consists of five rods of different diameters (1, 2, 3, 4, and 5 mm), which are used to calculate the RC [10].

Finally, the co-registration phantom (NMI PHAN IM coregistration, H171745, Bruker Biospin) containing three capillaries (each with a diameter < 1 mm, filled with a ¹⁸F-FDG solution), was used to evaluate the radial, tangential, and axial spatial resolution in reconstructed images.

Attenuation correction method

A CT scan of the DC cradle (see Fig. 2) including the surface coil was acquired in helical mode (voltage: 120KV, current: 100 mA, slice thickness: 0.625 mm, exposure time: 912s, pitch: 0.53125) using a PET/CT scanner (Discovery RX VCT 64, GE Healthcare, Milwaukee, WI) and reconstructed using a standard kernel (reconstruction diameter: 193 mm, slice thickness: 0.2 mm, matrix: 512 × 512).

Fig. 2.

Reconstructed CT of the dedicated cradle with surface coil (DC)

The sheet supporting the cradle in the CT scan, the flexible tube components, and various CT reconstruction artifacts, such as metal shadows, were manually removed (see Fig. 2). The AC map consists of two primary components: the cradle body (including the embedded coil) and the animal transporter system (ATS) adapter. The CT values of the cradle body were converted into linear attenuation coefficients (LACs) at a PET energy of 511 keV, following the method proposed by Burger et al. [11], with the adapted LAC of 0.093 cm⁻¹ for 0 Hounsfield units. However, this approach is not ideal for dense materials. Therefore, the LACs corresponding to the ATS adapter and the metallic components of the surface coil were replaced with values provided by the manufacturer (Bruker BioSpin GmbH & Co. KG, Ettlingen, Germany) as derived geometrically from the CAD design, and with material properties taken from tables published by the National Institute of Standards and Technology (NIST) [12].

Phantom acquisitions and reconstructions

All scans were performed with a dedicated preclinical 7T hybrid PET/MR system (BioSpin 70/18, Bruker, Ettlingen, Germany). The NMI PHAN IM QA phantom, NEMA Nu-4, and co-registration phantom were filled with 3.82, 3.7 and 5.85 MBq of ¹⁸F-FDG respectively, diluted with water. Two successive acquisitions were performed for each phantom, respectively with the CC and the DC configuration.

A 30-min PET acquisition was performed in list-mode format followed by the reconstruction of a static image from the whole dynamic data. The energy and coincidence windows were set to 357–664 keV and 5 ns respectively. After correction for attenuation, scatter (using dual energy window), decay, normalization, dead-time, random events, sensitivity, and point spread function (PSF), the sinograms were reconstructed using a MAP (Maximum a Posteriori) 0.5 mm algorithm (20 iterations) in a 180 × 180 mm² matrix. Partial volume correction (PVC) was applied during the reconstruction as described previously [13]. For the NEMA phantom the following image reconstruction parameters were tested: (i) 5, 10, 15, 20, 25, 30, 35, and 40 iterations, (ii) MAP (with 1 subset), OSEM 3D (Ordered Subset Expectation Maximization, with 16 subsets), and MLEM (Maximum Likelihood Expectation Maximization, with 1 subset) algorithms. The three reconstruction algorithms were initially evaluated to identify the most appropriate algorithm for optimal image quality. In a second step, the comparison was used to assess whether the reconstruction algorithm itself has an impact on the accuracy and robustness of attenuation correction when applying the pre-calculated AC map for the DC.

Phantom PET image processing

Homogeneous phantom

Within and across slices PET image uniformity was assessed by drawing five ROIs positioned at the center and at four cardinal locations (12, 3, 6, and 9 o’clock), and then projected to the other adjacent slices within the uniform region. For each slice (i) and 5 transverse locations (j), the mean activity concentration was obtained () to calculate the uniformity within a single slice (IU_i) and axial uniformity between slices (IU_AXIAL) for each of the 5 ROIs locations. All measurements were made in accordance with the report of the AAPM task group 126 [14].

The integral uniformity for slice i = N was calculated using this equation:

2

where and were the maximum and minimum of the five values measured in slice N, respectively.

The axial uniformity between Slices ( for a single transverse location was calculated using the following equation:

3

where and are the maximum and minimum of the values measured in all slices at transverse location M, respectively.

The CC–DC image subtraction was performed, and percent differences were calculated by defining specific regions of interest (ROIs) near and away from the surface coil.

NEMA NU 4 phantom

All measurements were made on the NEMA NU 4 phantom in accordance with NEMA NU 4 standards. Uniformity (STD) was calculated in reconstructed PET images using the following equation:

4

Regions of interest (ROIs) were defined on the air and water inserts to determine the spill-over ratio as follows:

5

Circular regions of interest (ROIs) were drawn around each rod, and the recovery coefficient (RC) was calculated as follows:

6

The contrast-to-noise ratio (CNR) was used to evaluate the trade-off between spatial resolution, recovery coefficients and image noise. The CNR was calculated as follows [15]:

7

Co-registration phantom

To calculate the spatial resolution using the coregistration phantom, profiles were generated through the maximum pixel value of each capillary in the radial, tangential, and axial directions using AMIDE software (amide.exe 1.0.6 http://amide.sf.net/) [16]. All segmentations were executed using the 3D Slicer 4.13 software (http://www.slicer.org/). Every single graph was produced using Python within the 6.4.12 Jupyter Notebook environment (https://jupyter.org/).

Animal studies

Six (n = 6) male Sprague-Dawley rats (weight 600–800 g), purchased from Janvier labs (Saint-Berthevin, France) were housed under controlled conditions (temperature 21 °C ± 1 °C, humidity 60% ± 10%, 12-h light/12-h dark cycle, and free access to food and water). The experimental procedures were approved by the regional animal ethics committee (CENOMEXA) and the Ministry of Higher Education, Research, and Innovation (#29403) and conducted in accordance with Directive 2010/63/EU of the European Union.

PET acquisitions and reconstructions

The animals were anesthetized with isoflurane (Forene, AbbVie, Rungis, France), with a 3% concentration for induction, followed by a maintenance dose of 1.5 ± 0.5%, administered in a 50% O₂ and 50% N₂O mixture. Then, a catheter was placed in a tail vein, and the animal was positioned in the PET scanner. Rats were initially positioned in the CC cradle. Simultaneously with the intravenous injection of 15–20 MBq of ¹⁸F-FDG, a 40-min dynamic PET acquisition (list mode) was started with the following temporal distribution: 1 × 5s, 12 × 10s, 6 × 30s, 5 × 60s, 5 × 300s, 1 × 295s. At the end of the acquisition, the animal was removed while under anesthesia, the CC cradle was replaced by a DC cradle, the resonator replaced by the surface coil, and the animal was repositioned. A new dynamic PET acquisition was then performed for 40 min with the same temporal distribution (without reinjection). A static PET was then reconstructed using a 0.5 mm MAP algorithm (20 iterations, 1 subset), with corrections for attenuation of the cradles, decay, normalization, dead-time, random events, sensitivity, PVC, and PSF modeling.

PET image analysis

To assess the effect of positioning a surface coil close to the heart, an automated volume of interest (see Fig. 3) encompassing the whole left ventricle was generated for each animal using Carimas 2.10 software (Turku, Finland). The mean pixel values were extracted for each VOI, and the SUVmean was calculated assuming a density of 1 g/mL as follows:

Fig. 3.

Automatic segmentation of PET images

8

Statistical analysis

Continuous data were compared using a paired test. Depending on the results of the normality test, either the Wilcoxon signed-rank test or the paired t-test was applied, using GraphPad Prism version 10.0.0 for Windows (GraphPad Software, Boston, Massachusetts, USA; www.graphpad.com). A p-value < 0.05 was considered statistically significant.

Results

Phantom studies

Homogeneous phantom

The true activity concentration in the phantom was 0.0601 MBq/mL, accurately recovered in the CC configuration but underestimated by about 1.33% in the DC configuration.

There was no significant difference between the mean pixel activity concentration in the homogeneous phantom images obtained with DC and CC (59.3 ± 5.5 kBq/mL vs. 0.0601 ± 3.7 kBq/mL). The CC–DC image subtraction demonstrated a 1.58% lower PET signal on the surface coil side (0.934 ± 1.688 kBq/mL) and a 5% higher PET signal on the opposite side (2.977 ± 1.858 kBq/mL, see Fig. 4) when using the DC compared to CC cradle.

Fig. 4.

Images of the homogeneous phantom: a with CC, b with DC, and c subtraction (CC minus DC) image

Uniformity within a slice was significantly higher in DC compared to CC (p = 0.02) while uniformity between slices did not show a significant difference between the two cradles (p = 0.2) although it remained below the 5% threshold (the limit for image uniformity evaluation) (see Fig. 5).

Fig. 5.

PET image uniformity evaluation for the two cradles: a within-slice uniformity, b axial inter-slice uniformity

NEMA Nu-4 phantom study

Recovery coefficients

Figure 6 depicts the RC values for each rod size, evaluated for two voxel sizes (0.25 mm and 0.5 mm) in NEMA images acquired with both cradles. The RC was higher for the DC compared to the CC. In terms of resolution, the 0.25 mm MAP showed a higher RC than the 0.5 mm MAP (see Fig. 6a). However, to get an optimal compromise between noise and resolution, the resolution was set to 0.5 mm for the next reconstructions.

Fig. 6.

Recovery coefficient (a) as a function of resolution and rod size, b as a function of uniformity for both DC and CC in the PET image reconstructed using MAP 0.5 mm algorithm

The RC values at different iteration updates were analyzed for both cradles for all rods (see Fig. 6b). It was observed that the RC values increased with the uniformity, independently of the rod size, with a convergence occurring after 3.6% in CC and after 5.2% for DC for all rod sizes (see Figure S1).

Spillover ratios and impact of scatter correction

Figure 7 depicts the spillover ratio (SOR) for both water- and air-filled tubes of the NEMA phantom, with and without scatter correction, for the two cradles as a function of the number of reconstruction iterations. The SOR values for both water and air were not significantly different between the two cradles. As expected, the impact of scatter correction was greater in the water-filled tube compared to the air-filled tube. Scatter correction did not significantly affect the SOR in either water or air for both cradles.

Fig. 7.

a Spillover ratios as function of iteration updates for, b water and c air. The blue and red curves are superimposed on the orange and green curves respectively

Contrast-to-noise ratio

Figure 8a depicts the CNR values as a function of rod size. CNR was decreased with the DC compared to the CC cradle. In line with the RC values, the uniformity also influenced CNR, a convergence of the CNR values occurring after 3.6% in CC and after 5.2% for DC (see Figs. 8b, S2).

Fig. 8.

CNR as function of (a) rod size, b uniformity for the 2 cradles

Comparison of reconstruction algorithms

The MAP algorithm showed an RC between 0 and 1, a lower %STD, a higher CNR and a higher SOR compared to MLEM and OSEM (see Fig. 9). As shown in Fig. 9a, the OSEM algorithm (20 iterations, 16 subsets) produced RC values above 1, which is likely due to noise-induced pixel overestimation and Gibbs artefacts. In addition, MAP images appeared less noisy than OSEM and MLEM (see Fig. 10). Therefore, the MAP algorithm was selected as the optimal algorithm and used for in-vivo reconstructions.

Fig. 9.

Comparison of OSEM, MLEM, and MAP based on: a RC, b %STD, c SOR in water, and d SOR in air

Fig. 10.

Representative reconstructions using OSEM (1itération, 16 subsets), MLEM (20 iterations), and MAP (20 iterations) of the NEMA NU 4 phantom. a rods level, b uniform region, c transverse slices at the level of water and air-filled cylinders

Resolution phantom

Only minimal differences in resolution were observed between the two cradles for the radial, tangential, and axial components of spatial resolution (see Table 1).

Table 1.

FWHM of radial, tangential, and axial image resolution

	CC	DC
Axial FWHM (mm)	1.63	1.80
Tangential FWHM (mm)	1.79	1.62
Radial FWHM (mm)	1.76	1.53

Effect of reconstruction corrections

The application of corrections for scatter, random, decay, partial volume, point spread function and attenuation resulted in an increased RC for both cradles (see Fig. 11).

Fig. 11.

Effect of: Scatter correction, randoms correction, decay correction, partial volume correction, point spread function and attenuation correction

Animal studies

As illustrated in Fig. 12, there was no difference (p = 0.8103) in cardiac ¹⁸F-FDG uptake between the two cradles, with left ventricular SUVmean of 3.55 ± 0.89 and 3.59 ± 1.02 for DC and CC respectively (see Fig. 13).

Fig. 12.

Coronal slices of MAP reconstructed images (20 iterations, 1 subset, all corrections including AC for the cradle and surface coil, have been applied, except for SC) of rat injected with 17 MBq of ¹⁸F-FDG. (a) and (b) image represent those acquired with CC and DC, respectively

Fig. 13.

SUVmean analysis based on the images acquired with the two cradles

Discussion

In this study, we compared the pre-calculated attenuation maps respectively dedicated to (i) a conventional cradle and to (ii) a custom cradle supporting a dual ¹H/²³Na surface coil. Our results showed that associating the standardization of surface coil positioning using a dedicated cradle with a precalculated attenuation map did not alter ¹⁸F-FDG PET quantification, allowing the use of a surface coil for triple modality imaging associating ¹H MR, ²³Na MR and PET within the same imaging session.

The MAP 0.5 mm algorithm showed the best compromise for spatial resolution, recovery coefficient, and image noise, without difference between the two cradles. In addition, the comparison between the 2 cradles found a significant difference in integral uniformity while axial uniformity between slices did not show significant difference between the two cradles. Although %STD was higher for the DC compared to the CC, leading to a lower CNR with DC while the SOR was similar for both cradles. Finally, no difference in cardiac ¹⁸F-FDG uptake was found in vivo using the two cradles.

Simultaneous PET/MR systems present a unique opportunity to enhance our understanding of biological processes, especially in rodent models, as it reduces the duration of anesthesia and allows both PET and MR assessment under the same physiological conditions. Büscher et al. evaluated the performance of a PET/MR system associating a low counting rate MR-compatible PET insert and a 7-T MR with a volume coil. Compared to a state-of-the-art preclinical PET scanner, image quality was high enough to allow the quantification of the myocardium-to-background ratio in a mice model of myocardial infarction [17]. Buonincontri et al. evaluated heart function, late gadolinium enhancement and glucose metabolism by means of ¹⁸F-FDG PET/MR during the same imaging session in a mice model of myocardial infarction. They showed that PET/MR was an accurate and efficient tool for global assessment of myocardial infarction [18]. Pedersen et al. also demonstrated the feasibility of the PET/MR imaging approach to assess atherosclerosis in a minipig model using a hybrid PET/MR scanner [19]. Glucose metabolism and sodium content are both impaired in heart failure and myocardial infarction [20–22]. In the present study, we demonstrated the feasibility of assessing glucose metabolism in the presence of a ¹H/²³Na surface coil, which can be later used to assess tissue sodium content.

Surface coils have several advantages over volume coils. They are effective for RF excitation in deep regions like the heart while limiting power deposition to deeper anatomical structures. In contrast, volume coils induce power deposition in a larger volume, potentially exceeding local SAR limits and RF-driven heating [23]. In addition, surface coils are usually preferred to body coils for X-nuclei imaging as they offer an increased signal [24].

In preclinical PET/MR imaging, MR hardware like coils and animal bed are not visible in MR images but contribute significantly to photon attenuation in PET images. Thus, they might be accurately added to the attenuation map at their correct spatial location. However, although attenuation correction methods have been extensively evaluated in clinical PET/MR, these methods remain poorly investigated in preclinical systems. Carney et al. [25] used a method that adjusts for CT tube voltage (kVp) when converting Hounsfield units (HU) to linear attenuation coefficients at 511 keV. This method ensures consistent HU-to-µ (511 keV) conversions across different scanners and tube voltages. However, this approach was developed for normal human tissues and is not applicable to the highly attenuating hardware materials found in PET/MR systems. In a phantom study, McDonald et al. investigated the effect of surface coils on attenuation correction using a clinical PET/MR, after measuring coils attenuation by CT scanning [26]. They demonstrated that a bilinear method for converting CT values overestimated the µ (511 keV) attenuation coefficients of the metal components, leading to significant artifacts and to a 28% overestimation of PET quantification. According to further studies, we used a bilinear transformation from HU to LACs, with two different slopes and a break point referring to the slopes intersection, improves the accuracy of AC maps [11, 27]. In order to further improve the accuracy of AC map, the ATS adapter and the metallic components of the surface coil were replaced with material-specific values as previously suggested [28]. However, when using a CT-derived AC map, the effect of the surface coil is location-dependent, the impact of measured activity being higher close to the coil, emphasizing the importance of the determination of the coil position which may be improved using external landmarks [29]. In the present study, a dedicated cradle allowed the standardization of coil positioning, further preventing the use of external landmarks.

As a result, we found only negligible differences between the two cradles compared to previous studies [7, 30]. In addition, these differences were observed in phantom but not in vivo. Using a similar surface coil and a 4.7-T PET/MR system, Evans et al. found that the SUV was underestimated by 40% in MRAC compared to gold standard transmissions scans AC, but the surface coil casing and animal bed were not included in the PET/MR attenuation correction [30]. In addition, although transmission AC was considered the gold standard method, it led to low spatial resolution µ maps and introduced noise into the final images, as previously reported [31, 32]. Bini et al. [7] used a 3T PET/MR system with a quadrature body coil and found that the MRAC method underestimates PET values by less than 10% in the aorta, liver, kidney and by − 26.1%, − 16.8% in the spine and back muscles, respectively, compared to CTAC SUV values in a rabbit model. In this latter study, the attenuation of the animal bed was not considered and the MR coil was removed before PET acquisition, which could have induced animal displacement and prevented validation of PET/MR imaging.

Our differences in the SUVmean results were slightly lower compared to a recent study by Balber et al. who investigated the influence of the different sources of attenuation including the animal bed and three different cylindrical MR coils in an in ovo study using a 9.4-T PET/MRI. They revealed that without AC of MR coils and animal bed, activity was underestimated by 60% at the center and 50% at the egg’s edge. Including the animal bed in the AC reduced this underestimation of activity and when all hardware components were included, the underestimation were decreased to 2% with a slight bias remaining between the center and the edges [33].

A possible limitation of our study is the fixed order of image acquisition. All animals were scanned using the CC before the DC, resulting in a longer administration of anesthesia. This potentially contributed to the slight increase in cardiac ¹⁸F-FDG uptake as isoflurane may increase cardiac glucose metabolism [34]. However, the difference was not significant, likely due to the close monitoring of anesthesia using a target breath rate of 50–70 which insures a stable and reproducible anesthesia [35].

Conclusion

In phantom, the attenuation correction of the dedicated cradle with surface coil through a pre-calculated attenuation maps led to an increase in the RC and uniformity compared to a conventional cradle with a resonator. On the other hand, the uniformity remained below 5%. However, in vivo this DC did not impact the SUV values. Therefore, it has no impact on quantitative PET imaging when using a surface coil for ¹H MR, ²³Na MR and PET within the same imaging session.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PET

Positron emission tomography

MR

Magnetic resonance

AC

Attenuation correction

DC

Dedicated cradle

CT

Computed tomography

CC

Conventional cradle

LACs

Linear attenuation coefficients

NEMA

National Electrical Manufacturers Association

RC

Recovery coefficients

%STD

Uniformity

SOR

Spillover ratio

¹⁸F-FDG

¹⁸F-fluorodeoxyglucose

SUV

Standardized uptake value

FOV

Field of view

HU

Hounsfield units

PLA

Polylactic acid

OSEM

Ordered subset expectation maximization

MAP

Maximum a posteriori

MLEM

Maximum likelihood expectation maximization

RC

Recovery coefficient

PVC

Partial volume correction

PSF

Point spread function

ROI

Region of interest

IU_i

Uniformity within a single slice

IU_AXIAL

Axial uniformity between slices

CNR

Contrast-to-noise ratio

Author Contribution

All authors made substantial contributions to the data acquisition, revised the article critically and gave final approval of the manuscript. Study conception and design: AM, HR, MJ. Acquisition of data: HR, MG, MN, MH, AL, AM. Analysis and interpretation of data: HR, MG, AL, MJ, AM. Manuscript draft: HR, AM.

Funding

Hanan Rida is supported in part by a grant from the GCS G4 as part of the FHU-CARNAVAL project labeled by AVIESAN. This work was conducted as part of a project supported by the Region Normandie (grant RIN SoGlu-PET-MR).

Data availability

The data will be made available on request from the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.