Skip to main content
. Author manuscript; available in PMC: 2017 Nov 14.
Published in final edited form as: Brainlesion. 2016 Oct;10154:32–42. doi: 10.1007/978-3-319-55524-9_4

Table 1.

Comparison of SVM outputs from the basic, intermediate, and multi-modal registration protocols. Bolded entries are significantly improved over the previous step, italicized entries for the multi-modal step are significantly improved over the initial step, as shown in t-tests of the 10 CV repeats.

Single channel registration/Standard template
BAC Accuracy Sensitivity Specificity F1
FA 0.7995 0.8039 0.8640 0.7350 0.8222
MD 0.7621 0.7692 0.8525 0.6717 0.7992
RD 0.7823 0.7892 0.8670 0.6975 0.8143
AD 0.7475 0.7580 0.8525 0.6425 0.7920
Single channel registration/Study specific template
BAC Accuracy Sensitivity Specificity F1
FA 0.8012 0.8118 0.8850 0.7175 0.8414
MD 0.8004 0.8043 0.8350 0.7658 0.8228
RD 0.8042 0.8132 0.8775 0.7308 0.8405
AD 0.7325 0.7520 0.8775 0.5875 0.7978
Multi-modal registration/Study specific template
BAC Accuracy Sensitivity Specificity F1
FA 0.8038 0.8146 0.8900 0.7175 0.8437
MD 0.8104 0.8143 0.8450 0.7758 0.8328
RD 0.8029 0.8134 0.8875 0.7183 0.8432
AD 0.7379 0.7751 0.8850 0.5908 0.8031