Table 2.

Summary of performance metrics for anatomical structure segmentation using the DeepLabv3-based (A) and SegFormer-based (B) combined models (common encoder with structure-specific decoders) on the test dataset (for each metric, mean and standard deviation are displayed).

	Anatomical structure	F1 score	IoU	Precision	Recall	Specificity
A. DeepLabv3	Abdominal wall	0.86±0.11	0.77±0.15	0.81±0.15	0.95±0.09	0.95±0.04
	Colon	0.75±0.19	0.63±0.21	0.71±0.18	0.84±0.23	0.95±0.04
	Inferior mesenteric artery	0.53±0.25	0.40±0.21	0.52±0.22	0.68±0.32	0.99±0.01
	Intestinal veins	0.46±0.32	0.35±0.27	0.70±0.23	0.48±0.36	1.00±0.00
	Liver	0.65±0.34	0.57±0.33	0.76±0.23	0.69±0.38	0.98±0.03
	Pancreas	0.32±0.30	0.23±0.24	0.61±0.33	0.32±0.35	0.99±0.01
	Small intestine	0.81±0.19	0.72±0.21	0.81±0.17	0.87±0.23	0.96±0.03
	Spleen	0.78±0.24	0.69±0.24	0.76±0.18	0.89±0.26	0.99±0.01
	Stomach	0.63±0.32	0.53±0.29	0.68±0.23	0.74±0.37	0.98±0.02
	Ureter	0.40±0.28	0.29±0.22	0.44±0.27	0.56±0.40	0.99±0.01
	Vesicular glands	0.42±0.27	0.30±0.23	0.41±0.30	0.56±0.36	0.98±0.02
B. SegFormer	Abdominal wall	0.76±0.24	0.66±0.24	0.78±0.15	0.86±0.28	0.92±0.07
	Colon	0.64±0.27	0.52±0.24	0.66±0.19	0.78±0.34	0.93±0.06
	Inferior mesenteric artery	0.40±0.24	0.28±0.18	0.37±0.21	0.68±0.38	0.97±0.02
	Intestinal veins	0.43±0.33	0.33±0.27	0.63±0.24	0.52±0.41	0.99±0.01
	Liver	0.62±0.35	0.53±0.32	0.76±0.19	0.71±0.40	0.95±0.07
	Pancreas	0.35±0.32	0.26±0.25	0.56±0.28	0.43±0.41	0.99±0.01
	Small intestine	0.78±0.19	0.67±0.20	0.74±0.14	0.90±0.23	0.94±0.06
	Spleen	0.71±0.24	0.59±0.23	0.65±0.21	0.89±0.25	0.99±0.01
	Stomach	0.65±0.32	0.55±0.29	0.71±0.22	0.75±0.36	0.98±0.02
	Ureter	0.38±0.29	0.27±0.23	0.43±0.27	0.55±0.40	0.99±0.01
	Vesicular glands	0.38±0.25	0.26±0.20	0.32±0.25	0.66±0.35	0.96±0.03

AI, artificial intelligence; IoU, Intersection-over-Union; SD, standard deviation.