Repeated/missing contrast info when converting sleuth file to dict

[yifan0330]

Summary

When using nimare.io.convert_sleuth_to_dataset, I found that if there're repeated contrast info (or missing contrast info which is represented as \\ Auther here) for multiple contrasts, it's recognised as one contrast and repeatedly overwritten by the contrast appearing later. Personally, I think maybe appending data from repeated contrasts as one contrast might be a more reasonable option.

[tsalo]

@yifan0330 can you share an example text snippet? Also, is this file directly from Sleuth, or curated manually in the Sleuth style?

[yifan0330]

@tsalo I came across this issue with a dataset that Angie shared with me, the dataset doesn't contain any specific contrast info and refer it as \\Auther here, here's an example text snippet,

//Reference=Talairach
//Author here
//Subjects=9
-14.832100 -28.597500 -9.493200
-23.061100 -16.852500 -13.924800
10.146500 -31.521900 -9.347700
29.665700 -11.797800 -9.851400

//Author here
//Subjects=9
-28.634500 -11.495400 -10.809000
29.644100 -20.182200 -10.646100

//Author here
//Subjects=42
-25.801300 -2.864100 -12.645900
26.875700 -3.660300 -6.425100

//Author here
//Subjects=24
-23.060500 -3.139800 -9.923100
24.170300 -3.123300 -11.825100

//Author here
//Subjects=42
-25.801300 -2.864100 -12.645900
24.177500 -0.328500 -11.560200 ``

when converted into dataset, it was repeatedly overwritten and only data from the last contrast was kept.

[tsalo]

TBH that seems more like a noncompliant Sleuth file, rather than an issue with NiMARE's ingestion, but I'll defer to others' opinions on this.

[yifan0330]

I see! Just curious, is it possible to find repeated contrast info in a sleuth file?

[tsalo]

In an official Sleuth file, I believe that different contrasts from the same study would have different metadata. Something like this:

//Reference=Talairach
//Author1 2023; Contrast 1 Name
//Subjects=9
-14.832100 -28.597500 -9.493200
-23.061100 -16.852500 -13.924800
10.146500 -31.521900 -9.347700
29.665700 -11.797800 -9.851400

//Author1 2023; Contrast 2 Name
//Subjects=9
-28.634500 -11.495400 -10.809000
29.644100 -20.182200 -10.646100