Fully Convolutional Architectures for Multi-Class Segmentation in Chest Radiographs

The success of deep convolutional neural networks on image classification and recognition tasks has led to new applications in very diversified contexts, including the field of medical imaging. In this paper we investigate and propose neural network architectures within the context of automated segmentation of anatomical organs in chest radiographs, namely for lungs, clavicles and heart. The problem of relative intrinsic imbalances in the dataspace is solved by relating prior class data distributions to the loss function. We investigate three different models and propose the best performing one based on the evaluation results. The models are trained and tested on the publicly available JSRT database, consisting of 247 X-ray images the ground truth masks for which are available in the SCR database. The networks have been trained in a multi-class setup with three target classes. Our best performing model, trained with the loss function based on the Dice coefficient, reached mean Jaccard overlap scores of 95.0% for lungs, 86.8% for clavicles and 88.2% for heart in the multi-class configuration. Though we proceed in a multi-class configuration, our best network reached competitive results on lung segmentation and outperformed single-class stateof-the-art methods on clavicles and heart segmentation tasks.