Kieran O'BRien

Naturalistic fMRI offers a patient‐friendly alternative to task‐based language mapping but presents analytical challenges due to complex brain responses. This study presents an automated framework for language mapping using naturalistic fMRI analysed with a Rotation Forest (RotF) classifier. Participants were presented a televised quiz show during fMRI acquisition and completed seven task‐based language paradigms, which were used as reference maps. RotF models were trained on voxel‐wise time series labelled using task‐based activation maps and evaluated on independent participants. Performance was assessed using the structural similarity index measure (SSIM) and the language lateralisation indices (LI). RotF generated activation maps demonstrated strong spatial and lateralisation agreement with task‐based activation, capturing both receptive and expressive language regions. Data quality, which was assessed through head motion and post‐scan questionnaire score, influenced model performance and provided a practical quality control measure. Feature selection identified a short, informative stimulus segment, enabling reliable language mapping with a scan duration of approximately 5 min. These findings demonstrate the feasibility of accurate, efficient and patient‐friendly language mapping using naturalistic fMRI combined with machine learning.

Quantitative Susceptibility Mapping (QSM) relies on a tissue’s underlying macroscopic geometry to lead to measurable orientation-dependent field perturbations. To understand and assess QSM error in vivo, anthropomorphic phantoms provide a useful model that mimic the electromagnetic properties and morphology of underlying tissue. Herein, we designed and manufactured an MRI compatible anthropomorphic phantom with cylindrical and spheroid compartments containing realistic susceptibilities to mimic hemorrhages, calcifications, and blood vessels. We estimated accuracy (ε, bias, RMSE) and repeatability (RC) of MEDI-susceptibility measurements within ROIs. We evaluated voxel-based agreement to validate susceptibility mapping under different acquisition conditions (3T versus 7T) and reconstruction algorithms (COSMOS versus MEDI). Reliable MEDI-based susceptibility measurements were obtained from ellipsoids but not from straws. The ellipsoids (|ε| = 0.007 to 0.083 ppm at 3T; 0.050 to 0.118 ppm at 7T) were more accurate than the straws (|ε| = 0.084 to 0.190 ppm at 3T; 0.105 to 0.160 ppm at 7T). The repeatability coefficient across all 6 ROIs (RC = 0.652 ppm at 3T; 0.459 ppm at 7T) was substantially larger than across the 4 ellipsoid ROIs only (RC’ = 0.168 ppm at 3T; 0.141 ppm at 7T). The accuracy at 3T (bias = -0.002 ppm, RMSE = 0.082 ppm) was better than the accuracy at 7T (bias = -0.056 ppm, RMSE = 0.092 ppm). Using voxels from the 4 ellipsoid ROIs, we observed excellent agreement between COSMOS and MEDI susceptibility maps at 3T, with linear regression of y=1.00x-0.01 (r=0.99). We observed some underestimation of MEDI susceptibility maps relative to COSMOS at 7T, with linear regression and y=0.93x-0.04 (r=0.99). The results imply that QSM reconstructions are reliable with 3T scanners but can be challenging with 7T scanners at high magnetic susceptibilities.

Background: Joint line obliquity (JLO) often increases after medial opening wedge high tibial osteotomy (MOWHTO), and excessive JLO may adversely affect the lateral compartment. However, its effects on cartilage tissue and postoperative outcomes remain unclear. This study investigated whether increased postoperative JLO is associated with early cartilage changes on quantitative MRI T 2 mapping and differences in clinical outcomes. Methods: A total of 33 knees undergoing MOWHTO were retrospectively reviewed. Radiographic parameters—hip-knee-ankle angle (HKA), medial proximal tibial angle (MPTA), and JLO—were assessed preoperatively and at 6 months. Based on postoperative radiographic measurements, patients were classified as increased JLO/MPTA (Group I: JLO > 5° and/or MPTA > 94°) or acceptable range JLO/MPTA (Group A: JLO < 5° and MPTA < 94°). Quantitative MRI T 2 mapping was performed preoperatively and at 12 months to measure T 2 relaxation times ( T 2 values) and Δ T 2 values (postoperative–preoperative) in the patella, trochlea, lateral femoral condyle (LFC), and lateral tibial plateau (LTP), analyzing superficial, deep, and total cartilage layers. Knee Injury and Osteoarthritis Outcome Score (KOOS) was assessed at both time points. Results: Group I included 13 patients and group A included 20 patients. Postoperative HKA was similar between groups ( p = 0.326). No significant intergroup differences were observed in preoperative or postoperative T 2 values in the patellofemoral or lateral compartments. However, Δ T 2 values in the LTP were significantly higher in group I than in group A in the superficial (3.3 ± 6.5 versus −1.7 ± 5.9 ms, p = 0.022) and total (4.5 ± 5.7 versus − 0.6 ± 5.0 ms, p = 0.012) layers. Group I also had worse KOOS–Symptoms ( p = 0.016), QOL ( p = 0.022), and Sport/Rec ( p = 0.040). Conclusions: Even with similar postoperative limb alignment, increased JLO/MPTA was associated with early degenerative changes in lateral tibial cartilage and worse clinical outcomes. Maintaining postoperative JLO within the acceptable range during MOWHTO may help maintain lateral compartment cartilage health and optimize clinical outcomes. Level of Evidence : Level IV.

The importance of accurate structural discrimination of the human grey matter regions has motivated the development of observer-independent reproducible methods that account for inter-individual architectonic variations. We introduce a non-invasive statistical residual analysis framework, employing unique tissue-specific magnetic resonance fingerprinting (MRF) signals after adjusting for the effect of T₁ and T₂^* MR relaxometry parameters (here termed MRF residuals). A 7 T Siemens MR scanner was used to acquire MRF signals, quantitative transmit magnetic field (B₁⁺) maps and T₁-weighted anatomical images of eleven cortical areas (5L, 5M, 5Ci, 7A, 7P, 7PC, hIP3, BA2, BA4a, BA4p and BA6) from six female participants. MRF residual signal for each voxel was calculated as the difference between the actual and best matching MRF signal evolutions from a precomputed MRF dictionary covering a range of T₁, T₂^* and B₁⁺ values. To compare MRF residuals between regions of interest, normalised autocorrelation was used as a shape-based statistical signal characterisation method and the Euclidean distance between autocorrelation profiles of residuals was used to measure the interareal dissimilarity. In the eleven cortical areas in both cerebral hemispheres of six participants, the proposed MRF residual analysis consistently showed interareal dissimilarity profiles that concorded with histological studies, indicating that MRF residuals potentially contain tissue microstructural information. MRF residual signals provide additional area-specific information that is complementary to the MR relaxometry-based (T₁, T₂^*) information used previously for distinguishing microstructural differences between human cerebral cortex regions in vivo. The proposed approach led to more accurate identification of structural variations across cortical areas of interest.

Objective: Ultra-high-field B0 ≥ 7 tesla (7T) cardiovascular magnetic resonance (CMR) offers increased resolution. However, electrocardiogram (ECG) gating is impacted by the magneto-hydrodynamic effect distorting the ECG trace. We explored the technical feasibility of a 7T magnetic resonance scanner using an ECG trigger learning algorithm to quantitatively assess cardiac volumes and vascular flow. Methods: 7T scans were performed on 10 healthy volunteers on a whole-body research MRI MR scanner (Siemens Healthineers, Erlangen, Germany) with 8 channel Tx/32 channels Rx cardiac coils (MRI Tools GmbH, Berlin, Germany). Vectorcardiogram ECG was performed using a learning phase outside of the magnetic field, with a trigger algorithm overcoming severe ECG signal distortions. Vectorcardiograms were quantitatively analyzed for false negative and false positive events. Cine CMR was performed after 3rd-order B₀ shimming using a high-resolution breath-held ECG-retro-gated segmented spoiled gradient echo, and 2D phase contrast flow imaging. Artefacts were assessed using a semi-quantitative scale. Results: 7T CMR scans were acquired in all patients (100%) using the vectorcardiogram learning method. 3,142 R-waves were quantitatively analyzed, yielding sensitivity of 97.6% and specificity of 98.7%. Mean image quality score was 0.9, sufficient to quantitate both cardiac volumes, ejection fraction, and aortic and pulmonary blood flow. Mean left ventricular ejection fraction was 56.4%, right ventricular ejection fraction was 51.4%. Conclusion: Reliable cardiac ECG triggering is feasible in healthy volunteers at 7T utilizing a state-of-the-art three-lead trigger device despite signal distortion from the magnetohydrodynamic effect. This provides sufficient image quality for quantitative analysis. Other ultra-high-field imaging applications such as human brain functional MRI with physiologic noise correction may benefit from this method of ECG triggering.

Simultaneous [18 F]-fluorodeoxyglucose positron emission tomography and functional magnetic resonance imaging (FDG-PET/fMRI) provides the capability to image two sources of energetic dynamics in the brain – cerebral glucose uptake and the cerebrovascular haemodynamic response. Resting-state fMRI connectivity has been enormously useful for characterising interactions between distributed brain regions in humans. Metabolic connectivity has recently emerged as a complementary measure to investigate brain network dynamics. Functional PET (fPET) is a new approach for measuring FDG uptake with high temporal resolution and has recently shown promise for assessing the dynamics of neural metabolism. Simultaneous fMRI/fPET is a relatively new hybrid imaging modality, with only a few biomedical imaging research facilities able to acquire FDG PET and BOLD fMRI data simultaneously. We present data for n = 27 healthy young adults (18–20 yrs) who underwent a 95-min simultaneous fMRI/fPET scan while resting with their eyes open. This dataset provides significant re-use value to understand the neural dynamics of glucose metabolism and the haemodynamic response, the synchrony, and interaction between these measures, and the development of new single- and multi-modality image preparation and analysis procedures.

Quantitative susceptibility mapping (QSM) reveals pathological changes in widespread diseases such as Parkinson’s disease, Multiple Sclerosis, or hepatic iron overload. QSM requires multiple processing steps after the acquisition of magnetic resonance imaging (MRI) phase measurements such as unwrapping, background field removal and the solution of an ill-posed field-to-source-inversion. Current techniques utilize iterative optimization procedures to solve the inversion and background field correction, which are computationally expensive and lead to suboptimal or over-regularized solutions requiring a careful choice of parameters that make a clinical application of QSM challenging. We have previously demonstrated that a deep convolutional neural network can invert the magnetic dipole kernel with a very efficient feed forward multiplication not requiring iterative optimization or the choice of regularization parameters. In this work, we extended this approach to remove background fields in QSM. The prototype method, called SHARQnet, was trained on simulated background fields and tested on 3T and 7T brain datasets. We show that SHARQnet outperforms current background field removal procedures and generalizes to a wide range of input data without requiring any parameter adjustments. In summary, we demonstrate that the solution of ill-posed problems in QSM can be achieved by learning the underlying physics causing the artifacts and removing them in an efficient and reliable manner and thereby will help to bring QSM towards clinical applications.

Quantitative susceptibility mapping (QSM) aims to extract the magnetic susceptibility of tissue from magnetic resonance imaging (MRI) phase measurements. The mapping of magnetic susceptibility in vivo has gained broad interest in several fields of science and medicine because it yields relevant information on biological tissue properties, predominantly myelin, iron and calcium. Thereby, QSM can also reveal pathological changes of these key components in devastating diseases such as Parkinson’s disease, Multiple Sclerosis, or hepatic iron overload. As QSM requires the solution of an ill-posed field-to-source-inversion, current techniques utilize manual optimization of regularization parameters to balance between smoothing, artifacts and quantification accuracy. We trained a fully convolutional deep neural network - DeepQSM - to invert the magnetic dipole kernel convolution. This network is capable of solving the ill-posed field-to-source inversion on real-world in vivo MRI phase data without the need for manual parameter tuning, which proves that this network has generalized the underlying mathematical principle of the dipole inversion. We demonstrate that DeepQSM’s susceptibility maps enable identification of deep brain substructures that are not visible in MRI phase data and provide information on their respective magnetic tissue properties. We illustrate DeepQSM’s clinical relevance in a patient with multiple sclerosis showing its sensitivity to white matter lesions. In summary, DeepQSM can be used to determine the composition of myelin sheets of nerve fibers in the brain, and to assess quantitative information on iron homeostasis and its dysregulation, and will subsequently contribute to a better understanding of these biological processes in health and disease.

This dataset contains a GRE and a QSM model of the human brain