ABSTRACT
Introduction: In the field of pediatric trauma computer-aided detection (CADe) and computer-aided diagnosis (CADx) systems have emerged offering a promising avenue for improved patient care. Especially children with wrist fractures may benefit from machine learning (ML) solutions, since some of these lesions may be overlooked on conventional X-ray due to minimal compression without dislocation or mistaken for cartilaginous growth plates. In this article, we describe the development and optimization of AI algorithms for wrist fracture detection in children. Methods: A team of IT-specialists, pediatric radiologists and pediatric surgeons used the freely available GRAZPEDWRI-DX dataset containing annotated pediatric trauma wrist radiographs of 6,091 patients, a total number of 10,643 studies (20,327 images). First, a basic object detection model, a You Only Look Once object detector of the seventh generation (YOLOv7) was trained and tested on these data. Then, team decisions were taken to adjust data preparation, image sizes used for training and testing, and configuration of the detection model. Furthermore, we investigated each of these models using an Explainable Artificial Intelligence (XAI) method called Gradient Class Activation Mapping (Grad-CAM). This method visualizes where a model directs its attention to before classifying and regressing a certain class through saliency maps. Results: Mean average precision (mAP) improved when applying optimizations pre-processing the dataset images (maximum increases of +25.51% mAP@0.5 and +39.78% mAP@[0.5:0.95]), as well as the object detection model itself (maximum increases of +13.36% mAP@0.5 and +27.01% mAP@[0.5:0.95]). Generally, when analyzing the resulting models using XAI methods, higher scoring model variations in terms of mAP paid more attention to broader regions of the image, prioritizing detection accuracy over precision compared to the less accurate models. Discussion: This paper supports the implementation of ML solutions for pediatric trauma care. Optimization of a large X-ray dataset and the YOLOv7 model improve the model's ability to detect objects and provide valid diagnostic support to health care specialists. Such optimization protocols must be understood and advocated, before comparing ML performances against health care specialists.
ABSTRACT
Wood pellets have been reported to emit toxic gaseous emissions during transport and storage. Carbon monoxide (CO) emission, due to the high toxicity of the gas and the possibility of it being present at high levels, is the most imminent threat to be considered before entering a pellet storage facility. For small-scale (<30 tons storage capacity) residential pellet storage facilities, ventilation, preferably natural ventilation utilizing already existing openings, has become the most favored solution to overcome the problem of high CO concentrations. However, there is little knowledge on the ventilation rates that can be reached and thus on the effectiveness of such measures. The aim of the study was to investigate ventilation rates for a specific small-scale pellet storage system depending on characteristic temperature differences. Furthermore, the influence of the implementation of a chimney and the influence of cross-ventilation on the ventilation rates were investigated. The air exchange rates observed in the experiments ranged between close to zero and up to 8 m(3) h(-1), depending largely on the existing temperature differences and the existence of cross-ventilation. The results demonstrate that implementing natural ventilation is a possible measure to enhance safety from CO emissions, but not one without limitations.
