A new automatic algorithm for quantification of myocardial infarction imaged by late gadolinium enhancement cardiovascular magnetic resonance: experimental validation and comparison to expert delineations in multi-center, multi-vendor patient data

Table 1 Bias and regional agreement in paired IR and PSIR LGE images from multi-center patient studies

	Magnitude IR LGE				Phase sensitive IR LGE				PSIR vs IR
	Bias to expert delineation [%LVM]	R-value	DSC full extent	DSC core extent	Bias to expert delineation [%LVM]	R-value	DSC full extent	DSC core extent	Bias [%LVM]	R-value
EWA algorithm	0 ± 5	0.89	0.82 ± 0.14	0.81 ± 0.15	−1 ± 5	0.88	0.82 ± 0.17	0.79 ± 0.15	−1 ± 4	0.91
Original weighted algorithm	−7 ± 8	0.68	0.70 ± 0.32	0.67 ± 0.32	*	*	*	*	*	*
EM threshold	6 ± 7	0.88	-	0.67 ± 0.14	6 ± 8	0.86	-	0.68 ± 0.14	0 ± 6	0.91
2SD threshold	7 ± 7	0.85	-	0.69 ± 0.15	8 ± 6	0.86	-	0.70 ± 0.13	1 ± 5	0.94
3SD threshold	0 ± 7	0.81	-	0.70 ± 0.21	−2 ± 7	0.79	-	0.70 ± 0.19	−2 ± 4	0.94
5SD threshold	−8 ± 8	0.68	-	0.50 ± 0.33	−13 ± 10	0.38	-	0.36 ± 0.31	−4 ± 6	0.81
FWHM (min) threshold	−8 ± 9	0.54	-	0.58 ± 0.20	9 ± 12	0.47	-	0.69 ± 0.17	18 ± 12	0.44
FWHM (remote) threshold	**	**	-	**	−8 ± 7	0.74	-	0.66 ± 0.19	**	**
Otsu threshold	−8 ± 11	0.50	-	0.50 ± 0.32	10 ± 15	0.46	-	0.64 ± 0.20	18 ± 17	0.35

Bias as % of left ventricular mass (%LVM), regression R-value and regional agreement by DSC to expert delineation for the EWA algorithm, the original weighted algorithm [14] and the threshold method of EM, 2SD, 3SD and 5SD from remote, and FWHM from minimum intensity [8], FWHM from mean intensity in remote [12] and Otsu's threshold [26] in paired magnitude inversion recovery (IR) and phase sensitive inversion recovery (PSIR) images (n = 49) and bias and regression R-value for PSIR vs IR LGE images. * the original weighted algorithm by Heiberg et al. [14] was developed for IR images and therefore only applied in IR images. ** the FWHM remote threshold was developed for PSIR images as part of the FACT algorithm by Hsu et al. [12] and therefore only applied in PSIR images

ISSN: 1532-429X