Skip to main content

Advertisement

Table 2 Ten most frequently co-occurred “Subject/Object” relationships identified from the corpus of 23,410 image reports

From: Using data-driven sublanguage pattern mining to induce knowledge models: application in medical image reports knowledge representation

Co-occurrence Pair Example Count
Location:Body Part frontal view:of(Situated_at):vertebral body; lower outer quadrant:of(Modifies):right breast 19,625
Observation:Body Part erythema:of(Occurs_in):left breast; mass lesion(Occurs_in):in:left breast 15,219
Pathology:Body Part B-cell lymphoma:of(Occurs_in):breast; fibroadenoma:with(Modifies):tissue 14,904
Medical Activity:Body Part ultrasound: in(Acts_on):left breast; CT scans: of(Acts_on):skull 13,479
Observation:Pathology x-ray findings:as(Indicative_of):pleural effusions; all features:of(Indicative_of):fibroadenoma 13,439
Functional Concept:Pathology outcome:of(Describes):breast cancer; case:of(Related_to): previous DCIS 12,394
Pathology:Pathology Haemangiomas:are(Be): benign vascular tumors; complications:include(Has):vessel thrombosis 12,119
Medical Activity:Pathology drainage:confirmed(Shows):breast abscess; mastectomy:for(Acts_on):breast malignancy 11,924
Medical Activity:Observation Chest x-ray:performed for(Deals_with):chest pain; biopsy:of(Acts_on):small lesion 11,890
Observation:Observation breast lump:with(Shows):occasional pain; features:of(Shows):benign lesion 11,882