A practical step-by step guide for radiologists on machine learning.

Congress:

2019 ASM

Poster Number:

R-0052

Type:

Educational Exhibit

Keywords:

Education and training, Computer Applications-General, Neural networks, Artificial Intelligence

Authors:

D. T. Wang¹, S. S. Wang², L. L. Wang³; ¹Ohio/US , ²Utah/US, ³Cincinnati/US

DOI:

10.26044/ranzcr2019/R-0052

DOI-Link:

https://dx.doi.org/10.26044/ranzcr2019/R-0052

Fig. 1: Plot of features (x) versus output (y). The red line represents the line of...

Fig. 3: Plot of features (x) against output (y). The data points are the same as figure...

Fig. 10: Checking the tensorflow version.

Fig. 11: Installing required packages.

Fig. 12: Training over 10 epochs, with accuracy shown.

Learning objectives

Understand basic premise of machine learning Understand terminology specific for machine learning Apply basic programming skills to create their owndeep learning algorithm

Background

Machine learning is not a new concept and has been present since the 1950s. It has, however, seen an exponential increase in interest and application in the past 10 years partly due to the increased computer processing power available. This increase in popularity is not just in the computer science field, but has spilled over into the radiology realm, with the recent creation of the Radiology: Artificial Intelligence journal. While there are many reference articles available, a step-by-step how-to guide is lacking. The premise of...

Imaging findings OR Procedure details

Tensorflow1 is an open source platform that allows ease of creating your own machine learning algorithm. Instead of dealing with linear algebra and writing each layer by individually, you can simply specify the number of nodes, layers and activation function. In the following section, we will demonstrate how to retrain a machine learning algorithm for a new classification task. Getting started The following instructions are for Ubuntu and a Nvidia graphics card. Install Nvidia driver, Docker (https://docs.docker.com/install/) and Nvidia docker (https://github.com/NVIDIA/nvidia-docker). Pull the tensorflow docker...

Conclusion

Machine learning in radiology is probably at it peak at on the Gartner hype cycle, with numerous research articles written and commercial software being developed at this time. In order to allow everyone some understanding behind this process, we have introduced the basic concept underlying deep learning and illustrated a brief way to get started on any deep learning project.

Personal information

David Wang is currently a nuclear medicine fellow at the University of Cincinnati Medical Center (UCMC), Cincinnati, OH after having completed a neuroradiology fellowship at UCMC. He trainedat the Royal Melbourne Hospitaland was admitted as a fellow of RANZCR in 2018.

References

Abadi, M., Agarwal, A., Barham, P., Brevdo, E.,Chen,Z.,Citro,C., et al. (2015). TensorFlow: Large-scale machine learning on heterogeneous systems. Software available from tensorflow.org.