Whole-body attenuation correction (AC) is still challenging in combined PET/MR scanners. We describe Dixon-VIBE Deep Learning (DIVIDE), a deep-learning network that allows synthesizing pelvis pseudo-CT maps based only on the standard Dixon volumetric interpolated breath-hold examination (Dixon-VIBE) images currently acquired for AC in some commercial scanners. Methods: We propose a network that maps between the four 2-dimensional (2D) Dixon MR images (water, fat, in-phase, and out-of-phase) and their corresponding 2D CT image. In contrast to previous methods, we used transposed convolutions to learn the up-sampling parameters, we used whole 2D slices to provide context information, and we pretrained the network with brain images. Twenty-eight datasets obtained from 19 patients who underwent PET/CT and PET/MR examinations were used to evaluate the proposed method. We assessed the accuracy of the μ-maps and reconstructed PET images by performing voxel- and region-based analysis comparing the SUVs (in g/mL) obtained after AC using the Dixon-VIBE (PET_Dixon), DIVIDE (PET_DIVIDE), and CT-based (PET_CT) methods. Additionally, the bias in quantification was estimated in synthetic lesions defined in the prostate, rectum, pelvis, and spine. Results: Absolute mean relative change values relative to CT AC were lower than 2% on average for the DIVIDE method in every region of interest except for bone tissue, where it was lower than 4% and 6.75 times smaller than the relative change of the Dixon method. There was an excellent voxel-by-voxel correlation between PET_CT and PET_DIVIDE (R² = 0.9998, P < 0.01). The Bland–Altman plot between PET_CT and PET_DIVIDE showed that the average of the differences and the variability were lower (mean PET_CT–PET_DIVIDE SUV, 0.0003; PET_CT–PET_DIVIDE SD, 0.0094; 95% confidence interval, [−0.0180,0.0188]) than the average of differences between PET_CT and PET_Dixon (mean PET_CT–PET_Dixon SUV, 0.0006; PET_CT–PET_Dixon SD, 0.0264; 95% confidence interval, [−0.0510,0.0524]). Statistically significant changes in PET data quantification were observed between the 2 methods in the synthetic lesions, with the largest improvement in femur and spine lesions. Conclusion: The DIVIDE method can accurately synthesize a pelvis pseudo-CT scan from standard Dixon-VIBE images, allowing for accurate AC in combined PET/MR scanners. Additionally, our implementation allows rapid pseudo-CT synthesis, making it suitable for routine applications and even allowing retrospective processing of Dixon-VIBE data.

We recently showed that patients with different chronic pain conditions (such as chronic low back pain, fibromyalgia, migraine and Gulf War illness) demonstrated elevated brain and/or spinal cord levels of the glial marker 18-kDa translocator protein (TSPO), which suggests that neuroinflammation might be a pervasive phenomenon observable across multiple aetiologically heterogeneous pain disorders. Interestingly, the spatial distribution of this neuroinflammatory signal appears to exhibit a degree of disease specificity (e.g. with respect to the involvement of the primary somatosensory cortex), suggesting that different pain conditions may exhibit distinct ‘neuroinflammatory signatures’. To explore this hypothesis further, we tested whether neuroinflammatory signal can characterize putative aetiological subtypes of chronic low back pain patients based on clinical presentation. Specifically, we explored neuroinflammation in patients whose chronic low back pain either did or did not radiate to the leg (i.e. ‘radicular’ versus ‘axial’ back pain). Fifty-four patients with chronic low back pain, 26 with axial back pain [43.7 ± 16.6 years old (mean ± SD)] and 28 with radicular back pain (48.3 ± 13.2 years old), underwent PET/MRI with ¹¹C-PBR28, a second-generation radioligand for TSPO. ¹¹C-PBR28 signal was quantified using standardized uptake values ratio (validated against volume of distribution ratio; n = 23). Functional MRI data were collected simultaneously to the ¹¹C-PBR28 data (i) to functionally localize the primary somatosensory cortex back and leg subregions; and (ii) to perform functional connectivity analyses (in order to investigate possible neurophysiological correlations of the neuroinflammatory signal). PET and functional MRI measures were compared across groups, cross-correlated with one another and with the severity of ‘fibromyalgianess’ (i.e. the degree of pain centralization, or ‘nociplastic pain’). Furthermore, statistical mediation models were used to explore possible causal relationships between these three variables. For the primary somatosensory cortex representation of back/leg, ¹¹C-PBR28 PET signal and functional connectivity to the thalamus were: (i) higher in radicular compared to axial back pain patients; (ii) positively correlated with each other; (iii) positively correlated with fibromyalgianess scores, across groups; and finally (iv) fibromyalgianess mediated the association between ¹¹C-PBR28 PET signal and primary somatosensory cortex–thalamus connectivity across groups. Our findings support the existence of ‘neuroinflammatory signatures’ that are accompanied by neurophysiological changes and correlate with clinical presentation (in particular, with the degree of nociplastic pain) in chronic pain patients. These signatures may contribute to the subtyping of distinct pain syndromes and also provide information about interindividual variability in neuroimmune brain signals, within diagnostic groups, that could eventually serve as targets for mechanism-based precision medicine approaches.

Attenuation correction in hybrid PET/MR scanners is still a challenging task. This paper describes a methodology for synthesizing a pseudo-CT volume from a single T1 weighted volume, thus allowing us to create correct AC maps. Methods: We propose a fast pseudo-CT generation from a patient-specific MRI T1-weighted image using a group-wise patch-based approach and a MRI-CT atlas dictionary. For every voxel in the input MR image, we compute the similarity of the patch containing that voxel with the patches of all MR images in the database, that lie in a certain anatomical neighborhood. The pseudo-CT is obtained as a local weighted linear combination of the CT values of the corresponding patches. The algorithm was implemented in a Graphical Processing Unit. Results: We evaluated our method both qualitatively and quantitatively for PET/MR correction. The approach performed successfully in all cases considered. We compared the SUVs of the PET image obtained after attenuation correction using the patient specific CT volume, and using the corresponding computed pseudo-CT volume. The voxel by voxel correlation between SUV obtained with both methods was very high (R2=0.9980, p<0.0001), while the Bland-Altman test showed that the average of the differences was very low (0.0006±0.0594). A region of interest analysis was also performed. The correlation between mean SUV value and maximum SUV value of every region was very high (R2=0.9989, p<0.0001 and R2=0.9904, p<0.0001, respectively). Conclusion: Results indicate how our method can accurately approximate the patient-specific CT and may serve as a potential solution for accurate attenuation correction in hybrid PET/MR systems. The quality of the corrected PET using our pseudo-CT is comparable to having acquired a patient-specific CT, thus, improving the results obtained with the UTE-based attenuation correction maps currently used in the mMR scanner. The GPU implementation led to a substantial decrease in computational time making the approach suitable for real applications.

Purpose Specific absorption rate (SAR) amplification around active implantable medical devices during diagnostic MRI procedures poses a potential risk for patient safety. In this study, we present a parallel transmit (pTx) strategy that can be used to safely scan patients with deep brain stimulation (DBS) implants. Methods We performed electromagnetic simulations at 3T using a uniform phantom and a multitissue realistic head model with a generic DBS implant. Our strategy is based on using implant‐friendly modes, which are defined as the modes of an array that reduce the local SAR around the DBS lead tip. These modes are used in a spokes pulse design algorithm in order to produce highly uniform magnitude least‐squares flip angle excitations. Results Local SAR (1 g) at the lead tip is reduced below 0.1 W/kg compared with 31.2 W/kg, which is obtained by a simple quadrature birdcage excitation without any sort of SAR mitigation. For the multitissue realistic head model, peak 10 g local SAR and global SAR are obtained as 4.52 W/kg and 0.48 W/kg, respectively. A uniform axial flip angle is also obtained (NRMSE <3%). Conclusion Parallel transmit arrays can be used to generate implant‐friendly modes and to reduce SAR around DBS implants while constraining peak local SAR and global SAR and maximizing flip angle homogeneity. Magn Reson Med 73:1896–1903, 2015.