Artificial Intelligence to derive aligned strain in cine CMR to detect patients with myocardial fibrosis: an open and scrutinizable approach.

Cine Cardiac Magnetic Resonance (CMR) is the gold standard for cardiac function evaluation, incorporating ejection fraction (EF) and strain as vital indicators of abnormal deformation. Rare pathologies like Duchenne muscular dystrophies (DMD) are monitored with repeated late gadolinium-enhanced (LGE) CMR for identification of myocardial fibrosis. However, it is judicious to reduce repeated gadolinium exposure and rather employ strain analysis from cine CMR. This solution is limited so far since full strain curves are not comparable between individual cardiac cycles and current practice mainly neglects diastolic deformation patterns. Our novel Deep Learning-based approach derives strain values aligned by key frames throughout the cardiac cycle. In a reproducibility scenario (57+82 patients), our results reveal five times more significant differences (22 vs. 4) between patients with scar and without, enhancing scar detection by +30%, improving detection of patients with preserved EF by +61%, with an overall sensitivity/specificity of 82/81%.

Velocity-Selected Rotational State Distributions of Nitric Oxide Scattered off Graphene Revealed by Surface-Velocity Map Imaging.

We report velocity-dependent internal energy distributions of nitric oxide molecules, NO, scattered off graphene supported on gold to further explore the dynamics of the collision process between NO radicals and graphene. These experiments were performed by directing a molecular beam of NO onto graphene in a surface-velocity map imaging setup, which allowed us to record internal energy distributions of the NO radicals as a function of their velocity. We do not observe bond formation but (1) major contributions from direct inelastic scattering and (2) a smaller trapping-desorption component where some physisorbed NO molecules have residence times on the order of microseconds. This is in agreement with our classical molecular dynamics simulations which also observe a small proportion of two- and multi-bounce collisions events but likewise a small proportion of NO radicals trapped at the surface for the entire length of the molecular dynamics simulations (a few picoseconds). Despite a collision energy of 0.31 eV, which would be sufficient to populate NO(v = 1), we do not detect vibrationally excited nitric oxide.

Le Cœur en Sabot: shape associations with adverse events in repaired tetralogy of Fallot.

BACKGROUND: Maladaptive remodelling mechanisms occur in patients with repaired tetralogy of Fallot (rToF) resulting in a cycle of metabolic and structural changes. Biventricular shape analysis may indicate mechanisms associated with adverse events independent of pulmonary regurgitant volume index (PRVI). We aimed to determine novel remodelling patterns associated with adverse events in patients with rToF using shape and function analysis. METHODS: Biventricular shape and function were studied in 192 patients with rToF (median time from TOF repair to baseline evaluation 13.5 years). Linear discriminant analysis (LDA) and principal component analysis (PCA) were used to identify shape differences between patients with and without adverse events. Adverse events included death, arrhythmias, and cardiac arrest with median follow-up of 10 years. RESULTS: LDA and PCA showed that shape characteristics pertaining to adverse events included a more circular left ventricle (LV) (decreased eccentricity), dilated (increased sphericity) LV base, increased right ventricular (RV) apical sphericity, and decreased RV basal sphericity. Multivariate LDA showed that the optimal discriminative model included only RV apical ejection fraction and one PCA mode associated with a more circular and dilated LV base (AUC = 0.77). PRVI did not add value, and shape changes associated with increased PRVI were not predictive of adverse outcomes. CONCLUSION: Pathological remodelling patterns in patients with rToF are significantly associated with adverse events, independent of PRVI. Mechanisms related to incident events include LV basal dilation with a reduced RV apical ejection fraction.

Molecular Dynamics Simulations of Nitric Oxide Scattering Off Graphene.

We performed classical molecular dynamics simulations to model the scattering process of nitric oxide, NO, off graphene supported on gold. This is motivated by our desire to probe the energy transfer in collisions with graphene. Since many of these collision systems comprising of graphene and small molecules have been shown to scatter non-reactively, classical molecular dynamics appear to describe such systems sufficiently. We directed thousands of trajectories of NO molecules onto graphene along the surface normal, while varying impact position, but also speed, orientation, and rotational excitation of the nitric oxide, and compare the results with experimental data. While experiment and theory do not match quantitatively, we observe agreement that the relative amount of kinetic energy lost during the collision increases with increasing initial kinetic energy of the NO. Furthermore, while at higher collision energies, all NO molecules lose some energy, and the vast majority of NO is scattered back, in contrast at low impact energies, the fraction of those nitric oxide molecules that are trapped at the surface increases, and some NO molecules even gain some kinetic energy during the collision process. The collision energy seems to preferentially go into the collective motion of the carbon atoms in the graphene sheet.

Mutually Improved Endoscopic Image Synthesis and Landmark Detection in Unpaired Image-to-Image Translation.

The CycleGAN framework allows for unsupervised image-to-image translation of unpaired data. In a scenario of surgical training on a physical surgical simulator, this method can be used to transform endoscopic images of phantoms into images which more closely resemble the intra-operative appearance of the same surgical target structure. This can be viewed as a novel augmented reality approach, which we coined Hyperrealism in previous work. In this use case, it is of paramount importance to display objects like needles, sutures or instruments consistent in both domains while altering the style to a more tissue-like appearance. Segmentation of these objects would allow for a direct transfer, however, contouring of these, partly tiny and thin foreground objects is cumbersome and perhaps inaccurate. Instead, we propose to use landmark detection on the points when sutures pass into the tissue. This objective is directly incorporated into a CycleGAN framework by treating the performance of pre-trained detector models as an additional optimization goal. We show that a task defined on these sparse landmark labels improves consistency of synthesis by the generator network in both domains. Comparing a baseline CycleGAN architecture to our proposed extension (DetCycleGAN), mean precision (PPV) improved by +61.32, mean sensitivity (TPR) by +37.91, and mean F1 score by +0.4743. Furthermore, it could be shown that by dataset fusion, generated intra-operative images can be leveraged as additional training data for the detection network itself.

Correction: Imaging the reactivity and width of graphene's boundary region.

Correction for 'Imaging the reactivity and width of graphene's boundary region' by Huda S. AlSalem et al., Chem. Commun., 2020, 56, 9612-9615, DOI: 10.1039/D0CC02675A.

Unsupervised Domain Adaptation From Axial to Short-Axis Multi-Slice Cardiac MR Images by Incorporating Pretrained Task Networks.

Anisotropic multi-slice Cardiac Magnetic Resonance (CMR) Images are conventionally acquired in patient-specific short-axis (SAX) orientation. In specific cardiovascular diseases that affect right ventricular (RV) morphology, acquisitions in standard axial (AX) orientation are preferred by some investigators, due to potential superiority in RV volume measurement for treatment planning. Unfortunately, due to the rare occurrence of these diseases, data in this domain is scarce. Recent research in deep learning-based methods mainly focused on SAX CMR images and they had proven to be very successful. In this work, we show that there is a considerable domain shift between AX and SAX images, and therefore, direct application of existing models yield sub-optimal results on AX samples. We propose a novel unsupervised domain adaptation approach, which uses task-related probabilities in an attention mechanism. Beyond that, cycle consistency is imposed on the learned patient-individual 3D rigid transformation to improve stability when automatically re-sampling the AX images to SAX orientations. The network was trained on 122 registered 3D AX-SAX CMR volume pairs from a multi-centric patient cohort. A mean 3D Dice of 0.86 ± 0.06 for the left ventricle, 0.65 ± 0.08 for the myocardium, and 0.77 ± 0.10 for the right ventricle could be achieved. This is an improvement of 25% in Dice for RV in comparison to direct application on axial slices. To conclude, our pre-trained task module has neither seen CMR images nor labels from the target domain, but is able to segment them after the domain gap is reduced. Code: https://github.com/Cardio-AI/3d-mri-domain-adaptation.

Energetics of hydrogen adsorption and diffusion for the main surface planes and all magnetic structures of γ-iron using density functional theory.

In this study, we calculated the energetics of hydrogen atoms adsorbing on and diffusing into the first few layers of γ-Fe for the (100), (110) and (111) surfaces and for the non-magnetic (NM), ferromagnetic (FM), and antiferromagnetic single (AFM1) and double layer (AFMD) structures. These studies are relevant as they atomistically simulate the early stages of hydrogen embrittlement in steels. We employed density functional theory to establish adsorption sites and energies for each plane and the minimum energy pathways for diffusion through the first few layers with associated activation barriers. Adsorption energies for all cases vary between ∼3.7 and 4.4 eV, and the energy barriers to diffusion in the bulk region vary between ∼0.2 and 1.2 eV for the twelve cases, with the highest and lowest bulk diffusion barriers occurring in the NM(111) and the FM(100) case, respectively. We conclude that the texturing of steels in order to expose certain cleavage planes or magnetic structures can decrease the likelihood of hydrogen embrittlement.

Imaging the reactivity and width of graphene's boundary region.

The reactivity of graphene at its boundary region has been imaged using non-linear spectroscopy to address the controversy whether the terraces of graphene or its edges are more reactive. Graphene was functionalised with phenyl groups, and we subsequently scanned our vibrational sum-frequency generation setup from the functionalised graphene terraces across the edges. A greater phenyl signal is clearly observed at the edges, showing evidence of increased reactivity in the boundary region. We estimate an upper limit of 1 mm for the width of the CVD graphene boundary region.

Adsorption site, orientation and alignment of NO adsorbed on Au(100) using 3D-velocity map imaging, X-ray photoelectron spectroscopy and density functional theory.

Nitric oxide adsorption on a Au(100) single crystal has been investigated to identify the type of adsorption, the adsorption site, and the orientation and alignment of the adsorbed NO relative to the surface. This was done using a combination of 3D-surface velocity map imaging, near-ambient pressure X-ray photoelectron spectroscopy, and density functional theory. NO was observed to be molecularly adsorbed on gold at ∼200 K. Very narrow angular distributions and cold rotational distributions of photodesorbed NO indicate that NO adsorbs on high-symmetry sites on the Au crystal, with the N-O bond axis close to the surface normal. Our density functional theory calculations show that NO preferentially adsorbs on the symmetric bridge (2f) site, which ensures efficient overlap of the NO π* orbital with the orbitals on the two neighbouring Au atoms, and with the N-O bond axis aligned along the surface normal, in agreement with our conclusions from the rotational state distributions. The combination of XPS, which reveals the orientation of NO on gold, with 3D-surface velocity map imaging and density functional theory thus allowed us to determine the adsorption site, orientation and alignment of nitric oxide adsorbed on Au(100).

Characterisation, coverage, and orientation of functionalised graphene using sum-frequency generation spectroscopy.

We report the unambiguous detection of phenyl groups covalently attached to functionalised graphene using non-linear spectroscopy. Sum-frequency generation was employed to probe graphene on a gold surface after chemical functionalisation using a benzene diazonium salt. We observe a distinct resonance at 3064 cm-1 which can clearly be assigned to an aromatic C-H stretch by comparison with a self-assembled monolayer on a gold substrate formed from benzenethiol. Not only does sum-frequency generation spectroscopy allow one to characterise functionalised graphene with higher sensitivity and much better specificity than many other spectroscopic techniques, but it also opens up the possibility to assess the coverage of graphene with functional groups, and to determine their orientation relative to the graphene surface.

Molecular hydrogen production from amorphous solid water during low energy electron irradiation.

This work investigates the production of molecular hydrogen isotopologues (H2, HD, and D2) during low energy electron irradiation of layered and isotopically labelled thin films of amorphous solid water (ASW) in ultrahigh vacuum. Experimentally, the production of these molecules with both irradiation time and incident electron energy in the range 400 to 500 eV is reported as a function of the depth of a buried D2O layer in an H2O film. H2 is produced consistently in all measurements, reflecting the H2O component of the film, though it does exhibit a modest reduction in intensity at the time corresponding to product escape from the buried D2O layer. In contrast, HD and D2 production exhibit peaks at times corresponding to product escape from the buried D2O layer in the composite film. These features broaden the deeper the HD or D2 is formed due to diffusion. A simple random-walk model is presented that can qualitatively explain the appearance profile of these peaks as a function of the incident electron penetration.

Kinetic Energy and Angular Distributions of He and Ar Atoms Evaporating from Liquid Dodecane.

We report both kinetic energy and angular distributions for He and Ar atoms evaporating from C12H26. All results were obtained by performing molecular dynamics simulations of liquid C12H26 with around 10-20 noble gas atoms dissolved in the liquid and by subsequently following the trajectories of the noble gas atoms after evaporation from the liquid. Whereas He evaporates with a kinetic energy distribution of (1.05 ± 0.03) × 2RT (corrected for the geometry used in experiments: (1.08 ± 0.03) × 2RT, experimentally obtained value: (1.14 ± 0.01) × 2RT), Ar displays a kinetic energy distribution that better matches a Maxwell-Boltzmann distribution at the temperature of the liquid ((0.99 ± 0.04) × 2RT). This behavior is also reflected in the angular distributions, which are close to a cosine distribution for Ar but slightly narrower, especially for faster atoms, in the case of He. This behavior of He is most likely due to the weak interaction potential between He and the liquid hydrocarbon.

Photodesorption of NO from Au(100) using 3D surface-velocity map imaging.

We measured the fully resolved 3-dimensional velocity distributions of nitric oxide photodesorbed from a gold single crystal. These experiments combine time-of-flight measurements and the velocity map imaging technique to yield velocity distributions resolved in three dimensions for a prototypical surface-adsorbate system. Nitric oxide adsorbed on Au(100) was photodesorbed using a 355 nm laser beam. The desorbed NO molecules were ionised in the gas-phase by resonance-enhanced multi-photon ionisation within a set of velocity map imaging optics. The NO molecules preferentially leave the surface along the surface normal with a very narrow angular distribution, indicating a non-thermal desorption process.

Electron-stimulated reactions in layered CO/H2O films: hydrogen atom diffusion and the sequential hydrogenation of CO to methanol.

Low-energy (100 eV) electron-stimulated reactions in layered H2O/CO/H2O ices are investigated. For CO layers buried in amorphous solid water (ASW) films at depths of 50 monolayers (ML) or less from the vacuum interface, both oxidation and reduction reactions are observed. However, for CO buried more deeply in ASW films, only the reduction of CO to methanol is observed. Experiments with layered films of H2O and D2O show that the hydrogen atoms participating in the reduction of the buried CO originate in the region that is 10-50 ML below the surface of the ASW films and subsequently diffuse through the film. For deeply buried CO layers, the CO reduction reactions quickly increase with temperature above ∼60 K. We present a simple chemical kinetic model that treats the diffusion of hydrogen atoms in the ASW and sequential hydrogenation of the CO to methanol to account for the observations.

Near-threshold photodissociation dynamics of CHCl3.

Energy- and angle-resolved photofragment distributions for ground-state Cl ((2)P3/2) and spin-orbit excited Cl* ((2)P1/2) have been recorded using the velocity map imaging technique after photodissociation of chloroform at wavelengths of 193 and ∼235 nm. Translational energy distributions are rather broad and peak between 0.6 and 1.0 eV. The spin-orbit branching ratios [Cl*]/[Cl] are 1 and 0.3 at 193 and 235 nm, respectively, indicating the involvement of two or more excited state surfaces. Considering the anisotropy parameters and branching ratios collectively, we conclude that the reaction at 193 nm takes place predominantly on the (1)Q1 surface, while the (3)Q1 surface gains importance at lower dissociation energies around 235 nm.

Validation of velocity map imaging conditions over larger areas.

We have established through simulations and experiments the area over which Velocity Map Imaging (VMI) conditions prevail. We designed a VMI setup in which we can vary the ionization position perpendicular to the center axis of the time-of-flight spectrometer. We show that weak extraction conditions are far superior over standard three-plate setups if the aim is to increase the ionization volume without distorting VMI conditions. This is important for a number of crossed molecular beam experiments that already utilize weak extraction conditions, but to a greater extent for surface studies where fragments are desorbed or scattered off a surface in all directions. Our results on the dissociation of NO2 at 226 nm show that ionization of the fragments can occur up to ±5.5 mm away from the center axis of the time-of-flight spectrometer without affecting resolution or arrival position.

Three-dimensional velocity map imaging of KBr surface photochemistry.

We have combined the velocity map imaging technique with time-of-flight measurements to study the surface photochemistry of KBr single crystals. This approach yields 3-dimensional velocity distributions of Br atoms resulting from 193 nm photodesorption. The velocity distributions indicate that at least two non-thermal mechanisms contribute to the photodesorption dynamics. Our experimental geometry also allows us to measure the Br(2P3/2):Br(2P1/2) branching ratio, which is found to be 24:1.