BIDS Derivatives: Standardization of Processing Results in Brain Imaging
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
<strong>Introduction</strong> We present BIDS-Derivatives, a set of principles for organizing and describing outputs of computations performed on brain imaging data, enabling researchers and tools to understand and reuse those outputs in subsequent processing.<br> <br> BIDS-Derivatives is an extension to the Brain Imaging Data Structure (BIDS), which is a standard for organizing magnetic resonance imaging (MRI) [2], electrophysiological [6, 7, 8] and behavioral data generated by a broad range of neuroscientific experiments. BIDS has facilitated the generation of tools (BIDS-Apps) [3] that may run with minimal intervention on BIDS datasets, adapting to the details of the available data. BIDS also provides a common structure for archiving data, both within labs and in large-scale databases such as OpenNeuro [4] and the NIMH Data Archive [10]. <strong>Methods</strong> The BIDS specification is hosted on GitHub and published on ReadTheDocs [9]. Significant modifications to BIDS are formulated as BIDS Extension Proposals (BEPs), which may be developed as separate documents or as "forks" of the document source.<br> <br> Derivatives were conceived during early BIDS discussions as a category distinct from raw experimental data, ranging from preprocessed data to publishable results. A BEP was initially drafted in February 2016. Further work defining the scope of derivatives at an August 2017 meeting led to the division of the effort into fine-grained proposals [5].<br> <br> In July 2018, a survey of the neuroimaging community was taken to establish priorities (essential, desirable or inessential) for structural, functional and diffusion MRI derivatives. The results of the survey were posted [1] in advance of an August 2018 workshop of 31 participants, where sub-proposals were pushed toward completion and common principles were established. In December 2018, Release Candidate 1 was published, including all imaging modalities, for implementation and feedback.<br> <br> In July 2019, a "Common Derivatives" proposal was re-introduced establishing more general principles, to be followed by subsequent modality-specific and non-imaging proposals. <strong>Results</strong> BIDS-Derivatives are specified in version 1.3.0 of the BIDS standard. This initial release specifies common derivatives, including dataset-level metadata, naming rules for preprocessed data of any modality, and generic imaging derivatives.<br> <br> Dataset metadata and organization follow BIDS conventions, and have been extended to allow the source dataset(s) to be linked and provenance information recorded of software used to generate the dataset.<br> <br> File-level naming rules permit space and desc keywords, allowing pipelines to distinguish files by a reference space or a generic description field. Custom references spaces may also be specified with the SpatialReference metadata field. All derived files must distinguish themselves from original (e.g., raw) data files by some component in the filename, permitting the inclusion of original and derived data in the same dataset, if necessary.<br> <br> Imaging-specific derivatives specified in this initial release include naming conventions for resampling parameters (e.g., resolution and surface mesh density) and specifications of regions of interest as masks or deterministic and probabilistic segmentations. <strong>Conclusions</strong> A standard for specifying derivatives will simplify the sharing and archiving of preprocessed data and the results of analyses. It will permit data repositories to provide canonical, preprocessed versions of datasets, simplify further automated processing, and facilitate collaboration between researchers and replication of analyses of published datasets.<br> <br> This initial release establishes common principles that guide future derivative specifications. Additional specifications of anatomical, functional and diffusion derivatives are planned within the next year, and electrophysiological, positron emission tomography, and connectomic derivatives are in progress.<br> <br> BIDS is an open effort, and everyone is encouraged to contribute, regardless of level of expertise.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.103 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it