Explainable localization of premature ventricular contraction using deep learning-based semantic segmentation of 12-lead electrocardiogram.

Background: Predicting the origin of premature ventricular contraction (PVC) from the preoperative electrocardiogram (ECG) is important for catheter ablation therapies. We propose an explainable method that localizes PVC origin based on the semantic segmentation result of a 12-lead ECG using a deep neural network, considering suitable diagnosis support for clinical application.

Methods: The deep learning-based semantic segmentation model was trained using 265 12-lead ECG recordings from 84 patients with frequent PVCs.

[Routine written intervention asserting the advantages of multiple blood cultures increases their order rate among doctors].

Purpose: Obtaining two or more blood culture sets is important for achieving good sensitivity and for detecting contamination. However, many doctors still only order one set for their laboratory testing. We wished to determine if routine written intervention to these doctors could increase the number of multiple blood cultures they ordered.

Microscopic polyangiitis accompanied by pleuritis as the only pulmonary manifestation of occupational silica exposure.

A 68-year-old man, who had worked for processing quartz-containing stones for more than 50 years, complained of low-grade fever and arthralgia. Mediastinal lymph nodes were markedly swollen on chest computed tomography. Pathological findings of the lymph node were compatible with silicosis, with a high titer of myeloperoxidase anti-neutrophil cytoplasmic antibody (MPO-ANCA).

An adult case of multiple squamous papillomas of the trachea associated with human papilloma virus type 6.

A 72-year-old woman with primary biliary cirrhosis complained of dry cough and wheezing. Chest computed tomography showed a tumor arising from the posterior wall of the trachea. Bronchoscopic examination revealed that the tumor was cauliflower-like, with two small polypoid tumors.