Skip to main content
. 2019 May 13;1(1):20180031. doi: 10.1259/bjro.20180031

Table 1.

Summary of the role of AI in radiation oncology

Step of the workflow Present AI role Present and future implications Reference no.
Image acquisition Development of sCT scan from MRI images.
  1. No requirement of separate Planning CT

  2. Better for image registration

26, 27 and 28.
Tumour segmentation Deep learning methods in contouring OAR and target tissue.
  1. Faster, more consistent contouring

  2. Helpful in adaptive planning

31, 32, 33 and 34.
Image registration Deep learning approaches. Faster and more precise image registration than intensity-based methods 36, 37.
Radiation planning Voxel based dose prediction and dose monitoring. Faster and more precise planning process 38 and 39.
Using historical patients’ data and present patient’s characteristics. Individualisation of dose constraints 40 41, 42 and 43.
Radiation delivery Using soft resort activator controlling flexion of neck. Decreased intra fraction motion. 44 and 45.
Using deep learning for estimating breathing pattern. Accurate tumour tracking with less errors of lag and predictive measures 48 and 49.