[MINC-users] NaN handling in RMINC

[RW]

The image is unrelated - my apologies.

Rebekah

------- Original Message -------
On Wednesday, June 7th, 2023 at 11:31 PM, RW <rwickens at protonmail.com> wrote:

> Hello,
>
> I’m using minc toolkit 1.9.17 on the mincvm-packer ([Update for Virtualbox 6.1.12 and packer 1.6.0](https://github.com/CoBrALab/MINC-VM/commit/15ca2dd0ee9287136232da4fe8e3ac9104c3437a)) via Windows Oracle VirtualBox 6.1, with lubuntu 18.04, and R version 3.6.3 and RMINC version 1.5.2.3.
>
> When performing simple linear regressions using mincLm on RMINC, I notice that if one of the scans contains NaN values (via minccalc), the resulting Betas (regression coefficients) appear to have implausible ranges (e.g., “Range: 3230528607327814463144632678524911616.000000 -0.061514).
>
> Is there a way I can ensure that NaNs of a given subject are ignored in the Beta calculations?
>
> Thank you for your assistance,
>
> Rebekah