“…The landmarks in our dataset comprised 309 pairs of points, including points defining the apex and base of the prostate, and various patient-specific landmarks including zonal structure boundaries, water-filled cysts, and calcifications ( Hamid et al., 2019 ). It should be noted that the overall spatial distribution of these landmarks may be representative of the full TRE distribution in this application ( Hu et al., 2008 ; Hahn et al., 2010 ; Karnik et al., 2010 ; Hu et al., 2011 ; Heinrich et al., 2012 ; Hu et al., 2012 ; Mitra et al, 2012a ; Mitra et al., 2012b ; De Silva et al., 2013 ; Sun et al., 2014 ; Fedorov et al., 2015 ; Hu et al., 2015 ; Sun et al., 2015 ; Yang et al., 2015 ; Zettinig et al., 2015 ; Wang et al., 2016 Onofrey et al., 2017 ; Hu et al., 2018 ; Sun et al., 2018 ; Yan et al., 2018 ; Sultana et al., 2019 ; Xu et al., 2020 ; Fu et al., 2021 ; Hu et al., 2021 ), but landmark-based TREs nevertheless provide a useful estimate of the errors associated with localising tumours within the prostate. The computational time was also recorded for each registration experiment.…”