The rise of large-scale imaging studies in psychiatry

GigaScience

Table 1 Comparison of study categories

Category	Comparability across datasets or sources	Control over analyses¹	Depth of phenotyping available	Ease of collecting a large sample²
Planned studies	Highest possible similarities in data	Highest	Can be very deep but specific to the study hypotheses	Most difficult
Aggregated mega-analyses	Moderate; Some filtering of available datasets for improves comparability	High	Low to Moderate	Moderately easy; it requires existing datasets and investigators willing to share them
Opportunistic studies	Moderate, but can be high if scanning centers agree to use minimal standard protocols	High	Moderate	Moderately difficult; it often requires institutional arrangements to get started
Meta-analyses (post-hoc)	Lowest	Lowest	Lowest	Easiest; all published results that fit the meta-analytic question are available
Meta-analyses (prospective)	Lowest	High	Lowest	Moderate; it requires existing datasets and teams to analyze them in collaboration

¹: Control over analyses covers whether planned analyses across each subset of the dataset are done the same way, either by a central authority or planned and agreed upon across groups performing the analyses.
²: Ease refers in a general way to the efforts involved in design and calibration, data aggregation and analysis, apart from study specific issues of the particular clinical population or obtaining any necessary funding, for example.

ISSN: 2047-217X