Difficulty Reproducing Anomalous Bundle Labeling from Fig. 4-5

[MorganJHLee]

Hello,
First, thank you for the excellent work on this tool.
I am attempting to reproduce the results presented in your paper, specifically the demonstration of anomalous fiber bundle labeling shown in Figures 4-5.
Steps to Reproduce

• I loaded the featuresB dataset, which I believe is the correct data used for Figures 4-5.
• I selected the same tracts of interest that were demonstrated in the paper's example.
• I then ran the anomaly detection process using inspect.

Expected Behavior

I expected to see a significant number of anomalous fibers flagged along the bundles, similar to the visualization in Figures 4-5 of the paper.

Actual Behavior

When I run the analysis, very few anomalies are detected and flagged. The result is not visually comparable to the published figures, showing a much lower anomaly rate.
Question
I have followed the paper's description, but I suspect I might be missing a step or a specific configuration.
Could you please clarify if there are other parameters, settings (e.g., anomaly thresholds), or preprocessing steps that need to be adjusted from the default values to successfully reproduce the results shown in Figures 4-5?
Any guidance you could provide would be greatly appreciated.

[chamberm]

Hi,

It looks like you are doing the right steps. My only assumption is that featuresB and the associated demog file do not contain age and sex (as used in the paper) - and the regression step is therefore omitted in your case?

Max

[MorganJHLee]

Thanks! But if so, is there any way to reproduce the results from Fig. 4-5 ?