Fast Prospective Detection of Contrast Inflow in X-ray Angiograms with Convolutional Neural Network and Recurrent Neural Network

Hua Ma, Pierre Ambrosini, Theo van Walsum

ABSTRACT:

Automatic detection of contrast inflow in X-ray angiographic sequences can facilitate image guidance in computer-assisted cardiac interventions. In this paper, we propose two different approaches for prospective contrast inflow detection. The methods were developed and evaluated to detect contrast frames from X-ray sequences. The first approach trains a convolutional neural network (CNN) to distinguish whether a frame has contrast agent or not. The second method extracts contrast features from images with enhanced vessel structures; the contrast frames are then detected based on changes in the feature curve using long short-term memory (LSTM), a recurrent neural network architecture. Our experiments show that both approaches achieve good performance on detection of the beginning contrast frame from X-ray sequences and are more robust than a state-of-the-art method. As the proposed methods work in prospective settings and run fast, they have the potential of being used in clinical practice.

Download paper here

Recommended citation

@inproceedings{ma2017fast,
  title={Fast prospective detection of contrast inflow in X-ray angiograms with convolutional neural network and recurrent neural network},
  author={Ma, Hua and Ambrosini, Pierre and van Walsum, Theo},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={453--461},
  year={2017},
  organization={Springer}
}