Abstract

Deep learning neural network techniques have been widely applied to medical radiological imaging diagnostics and have been proven to exceed the accuracy rates of highly trained radiologists. Numerous deep learning architectures have been applied to this task. Specifically, Convolutional Neural Networks (CNNs) have been widely applied and have demonstrated excellent performance characteristics. More recently, Capsule Networks (Capsnet) have managed to improve on the already stellar performance of CNNs by overcoming some of their weaknesses, such as pose and image transformation sensitivity. However, modifications to the Capsule Network can improve performance characteristics even further. The originally proposed Capsule Network utilizes convolutional techniques typical of CNNs in the initial layers of the network. By replacing the convolutional layers with dense layers, the Dense Capsule Network (DCNET) preserves more of the spatial image information which improves the accuracy of the network. In this paper the DCNET is applied to the problem of diagnosing pneumonia in ChestX-ray14, which is a publicly available chest X-ray dataset, consisting of over 100,000 radiological chest images capturing 14 distinct diseases. Diagnostic performance of DCNET is compared to that of CNNs, Capsnet, as well as that of expert radiologists.

References

1. G. Marques, D. Agarwal, and I. de la Torre Díez, “Automated medical diagnosis of COVID-19 through EfficientNet convolutional neural network,” Appl. Soft Comput., vol. 96, p. 106691, Nov. 2020, doi: 10.1016/J.ASOC.2020.106691.
2. K. R. Kruthika, Rajeswari, and H. D. Maheshappa, “Alzheimer’s Disease Neuroimaging Initiative. CBIR system using capsule networks and 3D CNN for Alzheimer’s disease diagnosis,” Inform. Med. Unlocked., vol. 14, pp. 59–68, Jan. 2019, doi: 10.1016/j.imu.2018.12.001.
3. K. J. Cios, K. Chen, and R. A. Langenderfer, “Use of Neural Networks in Detecting Cardiac Diseases from Echocardiographic Images,” IEEE Eng. Med. Biol. Mag., vol. 9, no. 3, pp. 58–60, 1990, doi: 10.1109/51.59215.
4. T. Neela and S. Namburu, “ECG signal classification using capsule neural networks,” IET Networks, vol. 10, no. 3, pp. 103–109, May 2021, doi: 10.1049/NTW2.12018.
5. C. L. Fischer Walker, I. Rudan, L. Liu,, “Global burden of childhood pneumonia and diarrhoea,” Lancet, vol. 381, no. 9875, pp. 1405–1416, 2013, doi: 10.1016/S0140-6736(13)60222-6.
6. A. T. Society, “Top 20 Pneumonia Facts — 201 9,” 2016. .
7. [7] H. F. H. System, “Pneumonia often misdiagnosed on patient readmissions, studies find -- ScienceDaily.” https://www.sciencedaily.com/releases/2010/10/101022123749.htm (accessed Apr. 22, 2022).
8. P. Rajpurkar et al., “CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning,” Nov. 2017, Accessed: May 20, 2022. [Online]. Available: http://arxiv.org/abs/1711.05225.
9. “CVPR 2021 Open Access Repository.” https://openaccess.thecvf.com/content/CVPR2021/html/Gu_Capsule_Network_Is_Not_More_Robust_Than_Convolutional_Network_CVPR_2021_paper.html (accessed Mar. 27, 2022).
10. S. Sabour, N. Frosst, and G. E. Hinton, “Dynamic Routing Between Capsules,” Adv. Neural Inf. Process. Syst., vol. 2017-Decem, pp. 3857–3867, Oct. 2017, doi: 10.48550/arxiv.1710.09829.
11. S. Samarth, R. Phaye, A. Sikka, A. Dhall, and D. Bathula, “Dense and Diverse Capsule Networks: Making the Capsules Learn Better,” May 2018, doi: 10.48550/arxiv.1805.04001.
12. G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely Connected Convolutional Networks,” Proc. - 30th IEEE Conf. Comput. Vis. Pattern Recognition, CVPR 2017, vol. 2017-January, pp. 2261–2269, Aug. 2016, doi: 10.48550/arxiv.1608.06993.
13. J. Yang, “MedMNIST v2: A Large-Scale Lightweight Benchmark for 2D and 3D Biomedical Image Classification,” Oct. 2021, doi: 10.48550/arxiv.2110.14795.

Download PDF
Back