Table 4.

The lesion-based diagnostic performance of AI model in testing cohort and the comparison of the AI performance among few, medium and extensive lesion groups

	Group for number of lesions			χ2	P value
	Few	Medium	Extensive
Sensitivity	58.63	64.34	89.56	163.41	< 0.001
Specificity	89.41	85.24	62.85	108.69	< 0.001
Accuracy	81.78	78.03	82.79	8.06	0.018
PPV	64.89	69.67	87.37	83.70	< 0.001
NPV	86.76	82.04	67.93	49.24	< 0.001

Chi-square test was performed to compare the performance of AI model among different groups of the number of lesions. Few lesions group: 1–3 lesions per image; Medium lesions group: 4–6 lesions per image; Extensive lesions group: > 6 lesions per image