As noted from the article where this image originated, the authors used SPM's segmentation-normalization
  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3358461/
As noted when the MNI template was created, the algorithm used led to the template being larger than average. In contrast, SPM normalizes to an average sized brain. This can be seen nicely in Figure 1 of of Horn et al. (2017)
  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6050588/
Therefore, if you used an MNI template for your normalization (FSL, ANTS) you would probably want to take this into account when using this template, as it is based on SPM normalization.

While I am happy that my MRIcron software has proved popular and robust, if your hardware can support it I would encourage you to try out MRIcroGL.
https://www.nitrc.org/projects/mricrogl
With MRIcroGL, if you choose the File/OpenStandard menu, you will see two templates: mni152 and spm152. The difference between these images is whether they use the larger MNI size or the more accurate SPM size.

You really want to choose your atlas to match your normalization procedures. In my own work with older stroke patients, I use templates that have larger ventricles to match the typical normalization seen for these people who are older than the sample used for the MNI dataset.