Fig. 1
From: Deep learning for semi-automated unidirectional measurement of lung tumor size in CT
Challenges associated with lesion segmentation and its conversion to unidirectional RECIST measurement. a Automated lesion measurement is challenged by the absence of visual distinction between beginning and ending points and their surrounding areas. b Lesion boundaries are visually distinct and well-defined; both segmentation and conversion of segmentation into measurement can easily be automated using existing techniques and tools. c Both segmentation and conversion of segmentation into measurement require a significant amount of clinical reasoning, which poses challenges to the idea of automated measurement through segmentation