Convolutional Recurrent Neural Network for Bubble Detection in a Portable Continuous Bladder Irrigation Monitor
Continuous bladder irrigation (CBI) is commonly used to prevent urinary problems after prostate or bladder surgery. Nowadays, the irrigation flow rate is regulated manually based on the color (qualitative estimation of the blood concentration) of the drainage fluid. To monitor the blood concentration quantitatively and continuously, we have developed a portable CBI monitor based on the Lambert-Beer law. It measures transmitted light intensity via a camera sensor and deduces the blood concentration. To achieve high reliability, we need to guarantee that the measurement is conducted when there is no air bubble passing through the view of the camera. To detect bubble occurrences, we propose a convolutional recurrent neural network with a sequence of images as input: the convolutional layers extract spatial features from 2D images; the recurrent layers capture temporal features in the image sequence. Our experimental results show that our network has smaller scale and higher accuracy compared with conventional convolutional and recurrent neural networks.
KeywordsContinuous bladder irrigation Blood concentration measurement Bubble detection Convolutional neural network Recurrent neural network
This work is partially supported by the BMBF project VisIMON under grant No. 16SV7861R.
- 1.Nojiri, Y., Okamura, K., Kinukawa, T., et al.: Continuous bladder irrigation following transurethral resection of the prostate (TURP). Nihon Hinyokika Gakkai Zasshi 98(6), 770–775 (2007)Google Scholar
- 5.Steve, S.: Management of clot retention following urological surgery. Nurs. Times 98(28), 48–50 (2002)Google Scholar
- 7.Rai, A., Yadav, N.: Autonomous continuous bladder irrigation devices, colorimeters, and methods of monitoring bladder pressure (2017)Google Scholar
- 8.Alex, K., Ilya, S., Geoffrey, E.H.: ImageNet classification with deep convolutional neural networks. Adv. Neural Inf. Process. Syst. 25(2), 1915–1929 (2012)Google Scholar
- 9.Ali, S.R., Hossein, A., Josephine, S., et al.: CNN features off-the-shelf: an astounding baseline for recognition. In: Conference on Computer Vision and Pattern Recognition Workshops. IEEE, USA (2014)Google Scholar
- 11.Tomas, M., Martin, K., Lukas, B.: Recurrent neural network based language model. In: 11th Annual Conference of the International Speech Communication Association, Japan (2010)Google Scholar
- 12.Michael, A., Michel, G., Chris, Q., et al.: Joint language and translation modeling with recurrent neural networks. In: Conference on Empirical Methods in Natural Language Processing, pp. 1044–1054. Association for Computational Linguistics, USA (2013)Google Scholar
- 13.Alex, G., Navdeep, J.: Towards end-to-end speech recognition with recurrent neural networks. In: 31st International Conference on Machine Learning, China (2014)Google Scholar
- 15.Zhiwei, D., Arash, V., Hexiang, H., et al.: Structure inference machines: recurrent neural networks for analyzing relations in group activity recognition. In: Conference on Computer Vision and Pattern Recognition. IEEE, USA (2016)Google Scholar
- 16.Ming, L., Xiaolin, H.: Recurrent convolutional neural network for object recognition. In: Conference on Computer Vision and Pattern Recognition. IEEE, USA (2015)Google Scholar
- 17.Russell S., Mykhaylo A., Andrew Y.N.: End-to-end people detection in crowded scenes. In: Conference on Computer Vision and Pattern Recognition. IEEE, USA (2016)Google Scholar
- 18.Sutskever I.: Training recurrent neural networks. Ph.D. thesis, University of Toronto (2012)Google Scholar