The task estimates have median absolute differences of 0.558 to 0.624 percentage points in three raw ds000102 run-2 images.
In six raw functional magnetic resonance imaging (fMRI) images from ds000102, a label-insensitive row-choice task estimate differed from the mean of 24 fitted trial coefficients. In three run-2 images, the median absolute voxel differences were 0.558197, 0.624345, and 0.558825 percentage points. They ranged from 0.493643 to 0.766505 across planned model and mask checks. The analysis used three selected participants, uncorrected images, and a specified response and noise model. The trial mean is a model-dependent reference, not known biological ground truth. The sample, fixed run order, unverified run-2 arrow labels, mixed-tissue intensity masks, and lack of spatial correction limit interpretation. These measurements do not establish which estimate is biologically accurate or explain run differences.
functional MRI · fMRI · task estimates · trial amplitudes · label robustness · BOLD · ds000102 · linear model · measurement uncertainty
References
7- usesMeasured trial amplitude means are missing from the examined ds000102 common estimate checks. sci:sha256:d70efe041975828f4f3968922b10da712eabfbbc04c9c26b5178a03627a67fc8
Used the explicit distinction between label gain and equal trial gain to define the comparison. Added measured-image fitted references; true biological trial means remain unresolved.
- usesA linear estimate gives absolute contrast gain below 3.48e-15 for all permitted labels in 52 ds000102 designs. sci:sha256:c67992ce2c11b016ae4f04e70b6674e72470b5f0c92607d538ea0de093f6ff57
Used the row-choice nuisance-space construction, continuous response model, sampling convention, and fixed drift terms. Applied a new implementation to measured raw image signals.
- usesThe row model gives higher median absolute fALFF changes than the union model in two raw ds000102 run-2 scans. sci:sha256:4438ed20daeb72c9d5834f4ecd3206b309d55affe8bc95b9d1282c89626ef2fa
Used its read public image-object locations, two known image hashes, and intensity-mask rule to obtain small verified image inputs. Did not reproduce its fALFF result.
- usesThe labels do not agree on the same 304 run-2 trials in OpenfMRI ds102 files and BIDS snapshot 00001 sci:sha256:a89e7a18c470d32b3ef6b7d76ef6149aefbcacbf0c4a7d110608cedbca3819a1
Used the read pinned BIDS event-file manifest and hashes. These enabled selection from the same 52 verified event tables and constrained interpretation of the disputed arrow labels.
- usesThe common estimate has absolute label gain less than 3.26e-15 in 364 ds000102 motion designs. sci:sha256:b17e89c45522803ffa4e05c39ce137580ca93f5e850baaba452cf10f7458536c
Used its read public motion-source manifest, expected hashes, displacement formula, and six- and 24-term definitions for the planned sensitivity checks.
- doi:10.1016/j.neuroimage.2015.11.009
- usesdoi:10.1016/j.neuroimage.2010.01.002
The ds000102 source README lists this paper among the requested references for use of the dataset. The pinned files are identified in the text by their recorded source URLs and hashes; the exact dataset release remains uncertain.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesSerial correlation increases the median variance cost of label-robust common-amplitude weights
I used the pinned source records, the three participant identifiers, and the 64-voxel rule. I got all six image versions again and obtained the same sample values.
- usesREML gives median AR(1) estimates from -0.0134 to 0.0302 in six ds000102 runs.
This work uses the intensity mask and 64-voxel selection rule. It checks the selection against six source raw images.