On 2015 Apr 27, Torgil Vangberg commented:

We are working on making the script compatible with SPM12, and it will certainly be of interest to see if T1- and T2-weighted inputs will offer any further improvement in accuracy. Actually we did not expect much improvement using multi-spectral inputs since the T2 scans had 4 mm slices, but still it improved the agreement with manual tracing. The algorithm has also been tested with good results on a separate datasets that had 1mm isotropic T1- and T2-weighted images.

On 2015 Apr 15, Gerard Ridgway commented:

It's interesting to see that multi-spectral data improves agreement with manual tracings (even though the tracing used only T1-weighted images). I think this is complementary to our work showing improvements in SPM12 over older versions (Malone IB, 2015), since SPM12 can also handle multi-spectral data. For example, considering the standard deviation of the differences with respect to manual, this paper's best (lowest) single-modality result is 41.64 mL for a method based on SPM8's New Segment toolbox. We obtained a value of 35.4 mL with SPM12. With T1- and T2-weighted data, this new paper obtains an impressively low 26.7 mL, also with lower mean difference (bias) than we found. It would be interesting to know whether SPM12 can further improve results using T1-weighted and T2-weighted data.

On 2015 Apr 15, Gerard Ridgway commented:

It's interesting to see that multi-spectral data improves agreement with manual tracings (even though the tracing used only T1-weighted images). I think this is complementary to our work showing improvements in SPM12 over older versions (Malone IB, 2015), since SPM12 can also handle multi-spectral data. For example, considering the standard deviation of the differences with respect to manual, this paper's best (lowest) single-modality result is 41.64 mL for a method based on SPM8's New Segment toolbox. We obtained a value of 35.4 mL with SPM12. With T1- and T2-weighted data, this new paper obtains an impressively low 26.7 mL, also with lower mean difference (bias) than we found. It would be interesting to know whether SPM12 can further improve results using T1-weighted and T2-weighted data.

On 2015 Apr 27, Torgil Vangberg commented:

We are working on making the script compatible with SPM12, and it will certainly be of interest to see if T1- and T2-weighted inputs will offer any further improvement in accuracy. Actually we did not expect much improvement using multi-spectral inputs since the T2 scans had 4 mm slices, but still it improved the agreement with manual tracing. The algorithm has also been tested with good results on a separate datasets that had 1mm isotropic T1- and T2-weighted images.