TABLE 2.

Sensitivity (TPR) and precision (PPV) of the automatic segmentation approach. Sensitivity and precision of the proposed methodology with respect to manual segmentation. For each validation set, the results are presented as mean ± SD, with N the number of samples per data set.

	D1	D7	D14	D21	Total
Roundness <0.8	n = 62	n = 126	n = 196	n = 98	n = 482
Sensitivity	0.9683 ± 0.0178	0.9768 ± 0.0146	0.9786 ± 0.0153	0.9791 ± 0.0096	0.9769
					± 0.0149
Precision	0.9880 ± 0.0067	0.9824 ± 0.0152	0.9769 ± 0.0099	0.9771 ± 0.0141	0.9798
					± 0.0127
Roundness >0.8	n = 138	n = 83	n = 75	n = 74	n = 370
Sensitivity	0.9727 ± 0.0139	0.9811 ± 0.0129	0.9792 ± 0.0195	0.9789 ± 0.0106	0.9771 ± 0.0148
Precision	0.9853 ± 0.0088	0.9808 ± 0.0092	0.9755 ± 0.0159	0.9802 ± 0.0084	0.9813 ± 0.0112