Automated inversion time selection for late gadolinium-enhanced cardiac magnetic resonance imaging.

OBJECTIVES: To develop and share a deep learning method that can accurately identify optimal inversion time (TI) from multi-vendor, multi-institutional and multi-field strength inversion scout (TI scout) sequences for late gadolinium enhancement cardiac MRI. MATERIALS AND METHODS: Retrospective multicentre study conducted on 1136 1.5-T and 3-T cardiac MRI examinations from four centres and three scanner vendors. Deep learning models, comprising a convolutional neural network (CNN) that provides input to a long short-term memory (LSTM) network, were trained on TI scout pixel data from centres 1 to 3 to identify optimal TI, using ground truth annotations by two readers. Accuracy within 50 ms, mean absolute error (MAE), Lin's concordance coefficient (LCCC) and reduced major axis regression (RMAR) were used to select the best model from validation results, and applied to holdout test data. Robustness of the best-performing model was also tested on imaging data from centre 4. RESULTS: The best model (SE-ResNet18-LSTM) produced accuracy of 96.1%, MAE 22.9 ms and LCCC 0.47 compared to ground truth on the holdout test set and accuracy of 97.3%, MAE 15.2 ms and LCCC 0.64 when tested on unseen external (centre 4) data. Differences in vendor performance were observed, with greatest accuracy for the most commonly represented vendor in the training data. CONCLUSION: A deep learning model was developed that can identify optimal inversion time from TI scout images on multi-vendor data with high accuracy, including on previously unseen external data. We make this model available to the scientific community for further assessment or development. CLINICAL RELEVANCE STATEMENT: A robust automated inversion time selection tool for late gadolinium-enhanced imaging allows for reproducible and efficient cross-vendor inversion time selection. KEY POINTS: • A model comprising convolutional and recurrent neural networks was developed to extract optimal TI from TI scout images. • Model accuracy within 50 ms of ground truth on multi-vendor holdout and external data of 96.1% and 97.3% respectively was achieved. • This model could improve workflow efficiency and standardise optimal TI selection for consistent LGE imaging.

Subsegmental pulmonary embolism and anticoagulant therapy: the impact of clinical context.

BACKGROUND: Anticoagulation for subsegmental pulmonary embolism (SSPE) is controversial. AIM: To assess the impact of clinical context on anticoagulation and outcomes of SSPE. METHODS: We electronically searched computed tomography pulmonary angiogram reports to identify SSPE. We extracted demographic, risk factor, investigations and outcome data from the electronic medical record. We stratified patients according to anticoagulation and no anticoagulation. RESULTS: From 1 January 2017 to 31 December 2019, we identified 166 patients with SSPE in 5827 pulmonary angiogram reports. Of these, 123 (74%) received anticoagulation. Compared with non-anticoagulated patients, such patients had a different clinical context: higher rates of previous venous thromboembolism (11% vs 0%; P = 0.019), more recent surgery (26% vs 9%; P = 0.015), more elevated serum D-dimer (22% vs 5%; P = 0.004), more lung parenchymal abnormalities (76% vs 61%; P = 0.037) and were almost twice as likely to require inpatient care (76% vs 42%; P < 0.001). Such patients also had twice the all-cause mortality at 1 year (32% vs 16%). CONCLUSIONS: SSPE is diagnosed in almost 3% of pulmonary angiograms and is associated with high mortality, regardless of anticoagulation, due to coexistent disease processes rather than SSPE. Anticoagulation appears dominant but markedly affected by the clinical context of risk factors, alternative indications and illness severity. Thus, the controversy is partly artificial because anticoagulation after SSPE is clinically contextual with SSPE as only one of several factors.

Topological Stone-Wales Defects Enhance Bonding and Electronic Coupling at the Graphene/Metal Interface.

Defects play a critical role for the functionality and performance of materials, but the understanding of the related effects is often lacking, because the typically low concentrations of defects make them difficult to study. A prominent case is the topological defects in two-dimensional materials such as graphene. The performance of graphene-based (opto-)electronic devices depends critically on the properties of the graphene/metal interfaces at the contacting electrodes. The question of how these interface properties depend on the ubiquitous topological defects in graphene is of high practical relevance, but could not be answered so far. Here, we focus on the prototypical Stone-Wales (S-W) topological defect and combine theoretical analysis with experimental investigations of molecular model systems. We show that the embedded defects undergo enhanced bonding and electron transfer with a copper surface, compared to regular graphene. These findings are experimentally corroborated using molecular models, where azupyrene mimics the S-W defect, while its isomer pyrene represents the ideal graphene structure. Experimental interaction energies, electronic-structure analysis, and adsorption distance differences confirm the defect-controlled bonding quantitatively. Our study reveals the important role of defects for the electronic coupling at graphene/metal interfaces and suggests that topological defect engineering can be used for performance control.

CardiSort: a convolutional neural network for cross vendor automated sorting of cardiac MR images.

OBJECTIVES: To develop an image-based automatic deep learning method to classify cardiac MR images by sequence type and imaging plane for improved clinical post-processing efficiency. METHODS: Multivendor cardiac MRI studies were retrospectively collected from 4 centres and 3 vendors. A two-head convolutional neural network ('CardiSort') was trained to classify 35 sequences by imaging sequence (n = 17) and plane (n = 10). Single vendor training (SVT) on single-centre images (n = 234 patients) and multivendor training (MVT) with multicentre images (n = 434 patients, 3 centres) were performed. Model accuracy and F1 scores on a hold-out test set were calculated, with ground truth labels by an expert radiologist. External validation of MVT (MVTexternal) was performed on data from 3 previously unseen magnet systems from 2 vendors (n = 80 patients). RESULTS: Model sequence/plane/overall accuracy and F1-scores were 85.2%/93.2%/81.8% and 0.82 for SVT and 96.1%/97.9%/94.3% and 0.94 MVT on the hold-out test set. MVTexternal yielded sequence/plane/combined accuracy and F1-scores of 92.7%/93.0%/86.6% and 0.86. There was high accuracy for common sequences and conventional cardiac planes. Poor accuracy was observed for underrepresented classes and sequences where there was greater variability in acquisition parameters across centres, such as perfusion imaging. CONCLUSIONS: A deep learning network was developed on multivendor data to classify MRI studies into component sequences and planes, with external validation. With refinement, it has potential to improve workflow by enabling automated sequence selection, an important first step in completely automated post-processing pipelines. KEY POINTS: • Deep learning can be applied for consistent and efficient classification of cardiac MR image types. • A multicentre, multivendor study using a deep learning algorithm (CardiSort) showed high classification accuracy on a hold-out test set with good generalisation to images from previously unseen magnet systems. • CardiSort has potential to improve clinical workflows, as a vital first step in developing fully automated post-processing pipelines.

Temperature-programmed desorption of large molecules: influence of thin film structure and origin of intermolecular repulsion.

Although the exact knowledge of the binding energy of organic adsorbates on solid surfaces is of vital importance for the realization of molecular nanostructures and the theoretical modelling of molecule-substrate interactions, an experimental determination is by no means trivial. Temperature-programmed desorption (TPD) is a widely used technique that can provide such information, but a quantitative analysis requires detailed knowledge of the pre-exponential factor of desorption and is therefore rarely performed on a quantitative level for larger molecules that often exhibit notable mutual intermolecular interactions. Here, we provide a thorough anlysis of TPD data of monolayers of pentacene and perfluoropentacene adsorbed on Au(111) that serve as a model system for polycyclic aromatic hydrocarbons adsorbed on noble metal surfaces. We show that the pre-exponential factor varies by several orders of magnitude with the surface coverage and evolves in a step-like fashion due to the sudden activation of a rotational degree of freedom during thermally controlled monolayer desorption. Using complementary coverage-dependent work function measurements, the interface dipole moments were determined. This allows to identify the origin and quantify the relative contributions of the lateral intermolecular interactions, which we modelled by force field calculations. This analysis clearly shows that the main cause for intermolecular repulsion are electrostatic interactions between the intramolecular charge distributions, while interface dipoles play only a minor role.

On-surface porphyrin transmetalation with Pb/Cu redox exchange.

Metal complexes at surfaces and interfaces play an important role in many areas of modern technology, including catalysis, sensors, and organic electronics. An important aspect of these interfaces is the possible exchange of the metal center, because this reaction can drastically alter the properties of the metal complex and thus of the interface. Here, we demonstrate that such metal exchange reactions are indeed possible and can proceed already at moderate temperatures even in the absence of solvents. Specifically, we studied the redox transmetalation of a monolayer of lead(ii)-tetraphenylporphyrin (PbTPP) with copper from a Cu(111) surface under ultrahigh-vacuum (UHV) conditions using multiple surface-sensitive techniques. Temperature-dependent X-ray photoelectron spectroscopy (XPS) reveals that the Pb/Cu exchange starts already below 380 K and is complete at 600 K. The identity of the reaction product, CuTPP, is confirmed by mass spectrometric detection in a temperature-programmed reaction (TPR) experiment. Scanning tunneling microscopy (STM) sheds light on the adsorbate structure of PbTPP at 300 K and uncovers the structural changes accompanying the transmetalation and side-reactions of the phenyl substituents. Moreover, individual free Pb atoms are observed as a product of the metal exchange.

Biphenylene network: A nonbenzenoid carbon allotrope.

The quest for planar sp2-hybridized carbon allotropes other than graphene, such as graphenylene and biphenylene networks, has stimulated substantial research efforts because of the materials' predicted mechanical, electronic, and transport properties. However, their syntheses remain challenging given the lack of reliable protocols for generating nonhexagonal rings during the in-plane tiling of carbon atoms. We report the bottom-up growth of an ultraflat biphenylene network with periodically arranged four-, six-, and eight-membered rings of sp2-hybridized carbon atoms through an on-surface interpolymer dehydrofluorination (HF-zipping) reaction. The characterization of this biphenylene network by scanning probe methods reveals that it is metallic rather than a dielectric. We expect the interpolymer HF-zipping method to complement the toolbox for the synthesis of other nonbenzenoid carbon allotropes.

Engineering of TMDC-OSC hybrid interfaces: the thermodynamics of unitary and mixed acene monolayers on MoS2.

Hybrid systems of two-dimensional (2D) materials such as transition metal dichalcogenides (TMDCs) and organic semiconductors (OSCs) have become subject of great interest for future device architectures. Although OSC-TMDC hybrid systems have been used in first device demonstrations, the precise preparation of ultra-thin OSC films on TMDCs has not been addressed. Due to the weak van der Waals interaction between TMDCs and OSCs, this requires precise knowledge of the thermodynamics at hand. Here, we use temperature-programmed desorption (TPD) and Monte Carlo (MC) simulations of TPD traces to characterize the desorption kinetics of pentacene (PEN) and perfluoropentacene (PFP) on MoS2 as a model system for OSCs on TMDCs. We show that the monolayers of PEN and PFP are thermally stabilized compared to their multilayers, which allows preparation of nominal monolayers by selective desorption of multilayers. This stabilization is, however, caused by entropy due to a high molecular mobility rather than an enhanced molecule-substrate bond. Consequently, the nominal monolayers are not densely packed films. Molecular mobility can be suppressed in mixed monolayers of PEN and PFP that, due to intermolecular attraction, form highly ordered films as shown by scanning tunneling microscopy. Although this reduces the entropic stabilization, the intermolecular attraction further stabilizes mixed films.

Direct synthesis of dilithium tetraphenylporphyrin: facile reaction of a free-base porphyrin with vapor-deposited lithium.

A solvent-free dilithium porphyrin was synthesized by direct reaction of free-base meso-tetraphenylporphyrin with elemental lithium in ultra-high vacuum. The reaction product dilithium tetraphenylporphyrin was studied by temperature-programmed desorption mass spectrometry (TPD-MS) and hard X-ray photoelectron spectroscopy (HAXPES). The solid-state reaction is thermodynamically favored, according to density functional theory (DFT) calculations.

Binary Lead Fluoride Pb3 F8.

The binary lead fluoride Pb3 F8 was synthesized by the reaction of anhydrous HF with Pb3 O4 or by the reaction of BrF3 with PbF2 . The compound was characterized by single-crystal and powder X-ray diffraction, IR, Raman, and solid-state MAS 19 F NMR spectroscopy, as well as thermogravimetric analysis, XP and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy. Solid-state quantum-chemical calculations are provided for the vibrational analyses and band assignments. The electronic band structure offers an inside view of the mixed valence compound.

Reactive metal-organic interfaces studied with hard x-ray photoelectron spectroscopy: controlled formation of metalloporphyrin interphase layers during metal vapor deposition onto porphyrin films.

Interfaces between organic semiconductors and metallic layers are ubiquitous in organic (opto-) electronic devices and can significantly influence their functionality. Here, we studied in situ prepared metal-organic interfaces, which were obtained by vapor deposition of metals (Co, Fe) onto organic semiconductor films (2H-tetraphenylporphyrin), with hard x-ray photoelectron spectroscopy. In these systems, the interphase zones, which are formed by diffusion and reaction of the metal in the organic material, can be clearly distinguished spectroscopically from the unreacted organic bulk, since they comprise the corresponding metalloporphyrins, CoTPP and FeTPP. In order to gain control over the thickness of the interphase layers, we varied process parameters such as sample temperature and metal-atom flux during interface preparation. We found that the temperature of the organic film during metal deposition was the only parameter that significantly influenced the formation of the interphase layers: their thicknesses were typically ~0.5 nm for deposition at 90 K, compared to ~1 nm at 300 K, irrespective of metal atom flux and chemical nature of the metal atom (Fe versus Co). Notably, these values are significantly smaller than the thicknesses of other metal/organics interphase regions reported in the literature.

LiNi(0.5)Mn(1.5)O4 high-voltage cathode coated with Li4Ti5O12: a hard X-ray photoelectron spectroscopy (HAXPES) study.

A Li4Ti5O12 (LTO) film was coated as buffer layer onto a LiNi0.5Mn1.5O4 (LNMO) high-voltage cathode, and after cycling of the cathode in a battery electrolyte, the LTO film was investigated by means of synchrotron radiation based hard X-ray photoelectron spectroscopy (HAXPES). By tuning the photon energy between 2 keV and 6 keV, we obtained non-destructive depth profiles of the coating material with probing depths ranging from 6 nm to 20 nm. The coating was found to be covered by a few nanometers thin surface layer resulting from electrolyte decomposition. This layer consisted predominantly of organic polymers as well as metal fluorides and fluorophosphates. A positive influence of the Li4Ti5O12 coating with regard to the size and stability of the surface layer was found. The coating itself consisted of a uniform mixture of Li(I), Ti(IV), Ni(II) and Mn(IV) oxides that most likely adopted a spinel structure by forming a solid solution of the two spinels LiNi0.5Mn1.5O4 and Li4Ti5O12 with Li, Mn, Ni and Ti cations mixing on the spinel octahedral sites. The diffusion of Ni and Mn ions into the Li4Ti5O12 lattice occurred during the heat treatment when preparing the cathode. The doping of Li4Ti5O12 with the open d-shell ions Ni(2+) (d(8)) and Mn(4+) (d(3)) should increase the electronic conductivity of the coating significantly, as was found in previous studies. The complex signal structure of the Ti 2p, Ni 2p and Mn 2p core levels provides insight into the chemical nature of the transition metal ions.