Table 2.

Reference-standard report labels across all abnormality categories. We refer to the ‘presence or absence of any abnormality’ dataset as the ‘binary’ dataset and the ‘specialised categories of abnormality’ dataset as the ‘granular’ dataset. Granular definitions are provided in the supplemental material. Briefly, ‘small vessel disease’ refers to the presence of moderate or severe small vessel disease [26]; ‘vascular’ includes abnormalities such as aneurysms; ‘atrophy’ refers to volume loss in excess of age; ‘encephalomalacia’ refers to any cause of permanent tissue damage including previous surgery or the chronic sequelae of infarcts or haemorrhages

Dataset	Binary label dataset King’s College Hospital NHS Foundation Trust (n = 3000)	Binary label dataset Guy’s and St Thomas’ NHS Foundation Trust (n = 500)	Granular label dataset Guy’s and St Thomas’ NHS Foundation Trust (n = 2000)
Category	Abnormal	Abnormal	Small vessel disease	Acute stroke	Mass	Vascular	White matter inflammation	Atrophy	Encephalomalacia
Number of examinations	1152	215	266	251	351	287	257	264	384
Interrater agreement (Fleiss kappa)	0.87	0.89	0.85	0.84	0.92	0.91	0.94	0.79	0.83