Unclear outlier detection strategy in ``detect_outliers_mad_median_absolute_deviation_mask``

[oesteban]

I would need some help understanding the strategy behind this method. Are you first finding outlying voxels in each volume and then of those, deciding which outliers are outlying in time?

[jagruti8]

Hi @oesteban,

I have taken this detector idea from the paper "Cox, R.W. Outlier Detection in FMRl Time Series. ISMRM(2002)."
First, it counts the number of outlying voxels for the entire image, by finding the median and median absolute deviation for each voxel and then it sums these outlying voxels for each time-point. The time-points that have outlying voxels beyond a certain threshold are considered as outliers.

Also, I have explained the entire strategy in algorithm.txt file.

Sorry for the late reply, I didn't get any notification in my gmail. Also, I will address to the other issues as soon as possible.

Hope this answers your question.

Thanks