Skip to main content

Table 1 Characteristics of gold standard manually abstracted radiology reportsα

From: Performance of a rule-based semi-automated method to optimize chart abstraction for surveillance imaging among patients treated for non-small cell lung cancer

Characteristics of radiology reports

Annotated reports (N (%))

Image types

n = 3011

 Bone scan

23 (0.8)

 Chest X-ray

1320 (43.8)

 CT abdomen/pelvis

127 (4.2)

 CT chest

902 (30.0)

 CT chest/abdomen/pelvis

123 (4.1)

 CT head

149 (4.9)

 MRI body

58 (1.9)

 MRI brain

42 (1.4)

 PET scan

267 (8.9)

Image indication

n = 3009

 Surveillance

954 (31.7)

 Symptomatic

649 (21.6)

 Follow up from prior abnormal chest imaging

244 (8.1)

 Follow up from prior abnormal other imaging

30 (1.0)

 Other

480 (16.0)

 Unknown

652 (21.7)

Image findings

n = 3008

 Suspicious

331 (11.0)

 Recurrence

110 (3.7)

 Benign

959 (31.9)

 Nonspecific

1019 (33.9)

 Second primary lung cancer

14 (0.5)

 Second primary cancer, other

11 (0.4)

 Other (unrelated to cancer)

564 (18.8)

  1. αThe total number of reportss with image type annotated = 3011, the total number of reports with image indication annotated = 3009, the total number of reports with image findings annotated = 3008. 180 reports were found in our manually abstracted dataset that were coded as null, indicating that they were not relevant images. Thus the total number of reports representing 361 patients was 3191