Fig. 5
From: Deep learning for semi-automated unidirectional measurement of lung tumor size in CT
Example of outlier measurement differences from Bland-Altman plot. a The lesion underestimated by DL algorithm was subject to controversy on whether its spiculations should be included in the measurement or not. b The lesion was underestimated by DL algorithm with no clear clinical reasoning behind the measurement. c The lesion was overestimated by DL algorithm as the central density was included in the measurement d The lesion was overestimated by DL algorithm as the algorithm was interfered by two separate lesions sharing the same field of view