A transformer-based multi-task deep learning model for simultaneous infiltrated brain area identification and segmentation of gliomas

Table 1 Demographic and clinical characteristics of patient cohort

	Training and validation sets		Independent test set
	N	%	N	%
No. of patients	300		54	100
Age
Range	16–82		11–71
Average	47.38 ± 12.75		48.89 ± 14.28
Gender
Male	168	56	27	50
Female	132	44	27	50
Grade range	II-IV		II-IV
II	149	49.67	21	37.04
III	59	19.67	14	22.22
IV	92	30.67	19	40.47
Infiltrated brain areas^a
frontal lobe (F)	172	57.33	29	53.7
parietal lobe (P)	65	21.67	11	20.37
occipital lobe (O)	36	12	4	7.41
temporal lobe (T)	115	38.33	27	50
insula lobe (I)	25	8.33	2	3.7
Area counts
> = 1	300	100	54	100
> = 2	98	32.67	17	31.48
> = 3	15	5	2	3.7

^aglioma-infiltrated brain areas, including frontal, parietal, occipital, temporal and insula lobe

ISSN: 1470-7330