RESUMEN
Generative adversarial networks (GANs) and diffusion models (DMs) have revolutionized the creation of synthetically generated but realistic-looking images. Distinguishing such generated images from real camera captures is one of the key tasks in current multimedia forensics research. One particular challenge is the generalization to unseen generators or post-processing. This can be viewed as an issue of handling out-of-distribution inputs. Forensic detectors can be hardened by the extensive augmentation of the training data or specifically tailored networks. Nevertheless, such precautions only manage but do not remove the risk of prediction failures on inputs that look reasonable to an analyst but in fact are out of the training distribution of the network. With this work, we aim to close this gap with a Bayesian Neural Network (BNN) that provides an additional uncertainty measure to warn an analyst of difficult decisions. More specifically, the BNN learns the task at hand and also detects potential confusion between post-processing and image generator artifacts. Our experiments show that the BNN achieves on-par performance with the state-of-the-art detectors while producing more reliable predictions on out-of-distribution examples.
RESUMEN
Background: We report a case of a 72-year-old patient developing a significant tricuspid regurgitation (TR) 6 years after a left ventricular assist device (LVAD) implantation. The aim of this case is to demonstrate the feasibility of transcatheter edge-to-edge repair (TEER) of the tricuspid valve and the excellent clinical benefit in long-term follow-up in an LVAD patient. Case summary: Our patient presented with recurrent acute heart failure syndrome. After a fulminant myocarditis in 2005, his previous treatment consisted of cardiac resynchronization therapy, TEER of the mitral valve, and LVAD (HeartMate III) implantation. At that point in time, his TR was only mild to moderate. Due to recurrent hospitalization despite optimized heart failure medication, we decided to treat the patient with a tricuspid TEER. His immediate post-interventional result and 1-year follow-up showed an excellent outcome with only minimal TR on transoesophageal echocardiogram. Discussion: In general, TR improves after LVAD implantation. However, there are two possible pathophysiological mechanisms, which result in an increasing TR: firstly, supporting LV dysfunction may lead to a leftward shift of the interventricular septum with restriction of the tricuspid leaflets. Secondly, the increase of venous preload with LVAD support may result in an annular dilatation with secondary TR, particularly in patients with pre-existing right ventricular dysfunction. According to the data currently available, the unpredictable course of developing TR necessitates regular clinical examination and echocardiographic investigation. Treatment with TEER appears to be feasible and safe, with excellent 1-year results in patients with previously implanted LVADs.