Table 3. Quantification of delineation errors on CT planning for thoracic organ at risk (OAR).

Group	Number of delineators and cases	OAR	Magnitude of observer delineation discrepancy
Collier et al. 2003 (36)	6 dosimetrists; 6 cases	Heart	Intra-observer variation (cm): average 0.5; maximum 7.6
			Inter-observer variation (cm): average 0.7; maximum 8.1
		Oesophagus	Intra-observer variation (cm): average 0.3; maximum 2.9
			Inter-observer variation (cm): average 0.4; maximum 3.1
		Spinal cord	Intra-observer variation (cm): average 0.1; maximum 0.7
			Inter-observer variation (cm): average 0.2; maximum 0.9
McCall et al. 2016 (37)	13 dosimetrists; 3 cases (gold standard defined by 2 radiation oncologists)	Heart	Inter-observer: mean DICE coefficient (SD) 0.91 (0.03)
		Oesophagus	Inter-observer: mean DICE coefficient (SD) 0.74 (0.03)
		Spinal cord	Inter-observer: mean DICE coefficient (SD) 0.86 (0.02)
		Lungs	Inter-observer: mean DICE coefficient (SD) 0.96 (0.01)
Cui et al. 2015 (38)	12 clinicians; 3 cases	Heart	Inter-observer
			Case 1 Mean DICE coefficient 0.86
			Case 2 Mean DICE coefficient 0.87
			*Case 3 Mean DICE coefficient 0.93
		Oesophagus	Inter-observer
			Case 1 Mean DICE coefficient 0.77
			Case 2 Mean DICE coefficient 0.76
			*Case 3 Mean DICE coefficient 0.84
		Brachial plexus	Inter-observer
			Case 1 Mean DICE coefficient 0.41
			Case 2 Mean DICE coefficient 0.40
			*†Case 3 Mean DICE coefficient 0.43

^†, Use of an atlas to aid delineation (http://www.rtog.org/CoreLab/ContouringAtlases/LungAtlas.aspx). *, Dice’s coefficients significantly improved (t-test, P<0.05) from those in case 1 and 2. SD, standard deviation; cm, centimeters.