AI-based X-ray fracture analysis of the distal radius: accuracy between representative classification, detection and segmentation deep learning models for clinical practice.

OBJECTIVES: To aid in selecting the optimal artificial intelligence (AI) solution for clinical application, we directly compared performances of selected representative custom-trained or commercial classification, detection and segmentation models for fracture detection on musculoskeletal radiographs of the distal radius by aligning their outputs. DESIGN AND SETTING: This single-centre retrospective study was conducted on a random subset of emergency department radiographs from 2008 to 2018 of the distal radius in Germany. MATERIALS AND METHODS: An image set was created to be compatible with training and testing classification and segmentation models by annotating examinations for fractures and overlaying fracture masks, if applicable. Representative classification and segmentation models were trained on 80% of the data. After output binarisation, their derived fracture detection performances as well as that of a standard commercially available solution were compared on the remaining X-rays (20%) using mainly accuracy and area under the receiver operating characteristic (AUROC). RESULTS: A total of 2856 examinations with 712 (24.9%) fractures were included in the analysis. Accuracies reached up to 0.97 for the classification model, 0.94 for the segmentation model and 0.95 for BoneView. Cohen's kappa was at least 0.80 in pairwise comparisons, while Fleiss' kappa was 0.83 for all models. Fracture predictions were visualised with all three methods at different levels of detail, ranking from downsampled image region for classification over bounding box for detection to single pixel-level delineation for segmentation. CONCLUSIONS: All three investigated approaches reached high performances for detection of distal radius fractures with simple preprocessing and postprocessing protocols on the custom-trained models. Despite their underlying structural differences, selection of one's fracture analysis AI tool in the frame of this study reduces to the desired flavour of automation: automated classification, AI-assisted manual fracture reading or minimised false negatives.

Multiclass datasets expand neural network utility: an example on ankle radiographs.

PURPOSE: Artificial intelligence in computer vision has been increasingly adapted in clinical application since the implementation of neural networks, potentially providing incremental information beyond the mere detection of pathology. As its algorithmic approach propagates input variation, neural networks could be used to identify and evaluate relevant image features. In this study, we introduce a basic dataset structure and demonstrate a pertaining use case. METHODS: A multidimensional classification of ankle x-rays (n = 1493) rating a variety of features including fracture certainty was used to confirm its usability for separating input variations. We trained a customized neural network on the task of fracture detection using a state-of-the-art preprocessing and training protocol. By grouping the radiographs into subsets according to their image features, the influence of selected features on model performance was evaluated via selective training. RESULTS: The models trained on our dataset outperformed most comparable models of current literature with an ROC AUC of 0.943. Excluding ankle x-rays with signs of surgery improved fracture classification performance (AUC 0.955), while limiting the training set to only healthy ankles with and without fracture had no consistent effect. CONCLUSION: Using multiclass datasets and comparing model performance, we were able to demonstrate signs of surgery as a confounding factor, which, following elimination, improved our model. Also eliminating pathologies other than fracture in contrast had no effect on model performance, suggesting a beneficial influence of feature variability for robust model training. Thus, multiclass datasets allow for evaluation of distinct image features, deepening our understanding of pathology imaging.

Advanced Molecular Tweezers with Lipid Anchors against SARS-CoV-2 and Other Respiratory Viruses.

The COVID-19 pandemic caused by SARS-CoV-2 presents a global health emergency. Therapeutic options against SARS-CoV-2 are still very limited but urgently required. Molecular tweezers are supramolecular agents that destabilize the envelope of viruses resulting in a loss of viral infectivity. Here, we show that first-generation tweezers, CLR01 and CLR05, disrupt the SARS-CoV-2 envelope and abrogate viral infectivity. To increase the antiviral activity, a series of 34 advanced molecular tweezers were synthesized by insertion of aliphatic or aromatic ester groups on the phosphate moieties of the parent molecule CLR01. A structure-activity relationship study enabled the identification of tweezers with a markedly enhanced ability to destroy lipid bilayers and to suppress SARS-CoV-2 infection. Selected tweezer derivatives retain activity in airway mucus and inactivate the SARS-CoV-2 wildtype and variants of concern as well as respiratory syncytial, influenza, and measles viruses. Moreover, inhibitory activity of advanced tweezers against respiratory syncytial virus and SARS-CoV-2 was confirmed in mice. Thus, potentiated tweezers are broad-spectrum antiviral agents with great prospects for clinical development to combat highly pathogenic viruses.

Chirality sensing of terpenes, steroids, amino acids, peptides and drugs with acyclic cucurbit[n]urils and molecular tweezers.

Achiral chromophoric hosts, i.e. acyclic cucurbit[n]urils and molecular tweezers, were found to respond with characteristic Circular Dichroism (CD) spectra to the presence of micromolar concentrations of chiral hydrocarbons, terpenes, steroids, amino acids and their derivates, and drugs in water. In favourable cases, this allows for analyte identification or for reaction monitoring.

Molecular Lysine Tweezers Counteract Aberrant Protein Aggregation.

Molecular tweezers (MTs) are supramolecular host molecules equipped with two aromatic pincers linked together by a spacer (Gakh, 2018). They are endowed with fascinating properties originating from their ability to hold guests between their aromatic pincers (Chen and Whitlock, 1978; Zimmerman, 1991; Harmata, 2004). MTs are finding an increasing number of medicinal applications, e.g., as bis-intercalators for DNA such as the anticancer drug Ditercalinium (Gao et al., 1991), drug activity reverters such as the bisglycoluril tweezers Calabadion 1 (Ma et al., 2012) as well as radioimmuno detectors such as Venus flytrap clusters (Paxton et al., 1991). We recently embarked on a program to create water-soluble tweezers which selectively bind the side chains of lysine and arginine inside their cavity. This unique recognition mode is enabled by a torus-shaped, polycyclic framework, which is equipped with two hydrophilic phosphate groups. Cationic amino acid residues are bound by the synergistic effect of disperse, hydrophobic, and electrostatic interactions in a kinetically fast reversible process. Interactions of the same kind play a key role in numerous protein-protein interactions, as well as in pathologic protein aggregation. Therefore, these particular MTs show a high potential to disrupt such events, and indeed inhibit misfolding and self-assembly of amyloidogenic polypeptides without toxic side effects. The mini-review provides insight into the unique binding mode of MTs both toward peptides and aggregating proteins. It presents the synthesis of the lead compound CLR01 and its control, CLR03. Different biophysical experiments are explained which elucidate and help to better understand their mechanism of action. Specifically, we show how toxic aggregates of oligomeric and fibrillar protein species are dissolved and redirected to form amorphous, benign assemblies. Importantly, these new chemical tools are shown to be essentially non-toxic in vivo. Due to their reversible moderately tight binding, these agents are not protein-, but rather process-specific, which suggests a broad range of applications in protein misfolding events. Thus, MTs are highly promising candidates for disease-modifying therapy in early stages of neurodegenerative diseases. This is an outstanding example in the evolution of supramolecular concepts toward biological application.

Two-color emissive probes for click reactions.

Cu(I)-catalyzed azide-alkyne cyclization (CuAAC) is the paradigmatic click reaction of continuous interest. Especially fluorogenic and FRET probes have become indispensable tools for life sciences. Here, we present a fluorescent alkyne for monitoring CuAAC, which undergoes a bathochromic shift upon reaction. Application in single-molecule and catalysis research is foreseen.