Detecting Pulmonary Embolism using Deep Neural Networks

doi:10.23940/ijpe.21.03.p8.322332

Abstract

Abstract:

Medical image processing is a method to create visual representations of the internal parts of the human body such as organs or tissues which helps diagnose and monitor diseases. Pulmonary Embolism (PE) is a medical issue where there is an artery obstruction in the lungs. PE is the third most common cause of cardiovascular death and is related to multiple inherited and acquired risk factors. The earlier diagnosis of PE detection helps to increase the patient's survival. With the advancement of Artificial Intelligence (AI), deep learning has become the leading technique, as it established significant capabilities in medical image processing tasks. In this research, a popular deep learning technique called Convolution Neural Network (CNN) is used to detect the pulmonary embolism in lung CT scan images. Four different types of predefined CNN architectures such as the Inception, VGG-16, ResNet50, and Mobilenet are used to compare the performance of the CNN model. In this experiment, RSNA STR Pulmonary Embolism Chest CT scan image dataset is analyzed. The empirical results show that the Inception-based CNN model provides better results when compared to other CNN architectures.

Key words: Pulmonary Embolism, deep learning, convolution neural network, VGG-16, ResNet50, Inception, Mobilenet

J Akilandeswaria, G. Jothib, A Naveenkumara, R. S. Sabeenianc, P. Iyyanara, and M. E Paramasivamc . Detecting Pulmonary Embolism using Deep Neural Networks [J]. Int J Performability Eng, 2021, 17(3): 322-332.

Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks

References 16.

[1].	RobbR.A. Biomedical imaging, visualization, and analysis. John Wiley and Sons, Inc., 1999.
[2].	BĕlohlávekJ., DytrychV. and LinhartA. Pulmonary embolism, part I: Epidemiology, risk factors and risk stratification, pathophysiology, clinical presentation, diagnosis and nonthrombotic pulmonary embolism. Experimental and Clinical Cardiology, 18(2), p.129, 2013.
[3].	MaX., NiuY., GuL., WangY., ZhaoY., BaileyJ. and LuF. Understanding adversarial attacks on deep learning based medical image analysis systems. Pattern Recognition, 110, p. 107332, 2021.
[4].	YangX., LinY., SuJ., WangX., LiX., LinJ. and ChengK.T. A two-stage convolutional neural network for pulmonary embolism detection from CTPA images. IEEE Access, 7, pp.84849-84857, 2019.
[5].	ShiL, RajanD, AbedinS, YellapragadaM.S, BeymerD, and DehghanE. Automatic Diagnosis of Pulmonary Embolism Using an Attention-guided Framework: A Large-scale Study. In Proceedings of the Third Conference on Medical Imaging with Deep Learning, 121, pp.743-754, 2020.
[6].	RajanD., BeymerD., AbedinS. and DehghanE. Pi-PE: A Pipeline for Pulmonary Embolism Detection using Sparsely Annotated 3D CT Images. In Machine Learning for Health Workshop, pp.220-232, 2020.
[7].	HuangS.C., KothariT., BanerjeeI., ChuteC., BallR.L., BorusN., HuangA., PatelB.N., RajpurkarP., IrvinJ. and DunnmonJ. PENet—A scalable deep-learning model for automated diagnosis of pulmonary embolism using volumetric CT imaging. NPJ digital medicine, 3(1), pp.1-9, 2020.
[8].	LongK., TangL., PuX., RenY., ZhengM., GaoL., SongC., HanS., ZhouM. and DengF. Probability-based Mask R-CNN for pulmonary embolism detection. Neurocomputing, 422, pp.345-353, 2021.
[9].	MuenzelD., FingerleA.A., ZahelT., SauterA., VlassenbroekA., DobritzM., RummenyE.J.and NoëlP.B. CT angiography: post-processed contrast enhancement for improved detection of pulmonary embolism. Academic radiology, 24(2), pp.131-136, 2017.
[10].	LeCunY., BottouL., BengioY. and HaffnerP. Gradient-based learning applied to document recognition. In Proceedings of the IEEE, 86(11), pp.2278-2324, 1998.
[11].	SimonyanK, and ZissermanA. Very deep convolutional networks for large-scale image recognition. ICLR 2015, 2015.
[12].	HeK., ZhangX., RenS. and SunJ. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.770-778, 2016.
[13].	SzegedyC., LiuW., JiaY., SermanetP., ReedS., AnguelovD., ErhanD., VanhouckeV. and RabinovichA. Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.1-9, 2015.
[14].	HowardA. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M. and Adam, H. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv: 1704. 04861, 2017.
[15].	https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/data (last accessed March 2021)
[16].	TajbakhshN., ShinJ.Y., GotwayM.B.and LiangJ. Computer-aided detection and visualization of pulmonary embolism using a novel, compact, and discriminative image representation. Medical image analysis, 58, p. 101541, 2019.