Go Figure: Transparency in neuroscience images preserves context and clarifies interpretation

Abstract: Visualizations are vital for communicating scientific results. Historically, neuroimaging figures have only depicted regions that surpass a given statistical threshold. This practice substantially biases interpretation of the results and subsequent meta-analyses, particularly towards non-reproducibility. Here we advocate for a "transparent thresholding" approach that not only highlights statistically significant regions but also includes subthreshold locations, which provide key experimental context. This balances the dual needs of distilling modeling results and enabling informed interpretations for modern neuroimaging. We present four examples that demonstrate the many benefits of transparent thresholding, including: removing ambiguity, decreasing hypersensitivity to non-physiological features, catching potential artifacts, improving cross-study comparisons, reducing non-reproducibility biases, and clarifying interpretations. We also demonstrate the many software packages that implement transparent thresholding, several of which were added or streamlined recently as part of this work. A point-counterpoint discussion addresses issues with thresholding raised in real conversations with researchers in the field. We hope that by showing how transparent thresholding can drastically improve the interpretation (and reproducibility) of neuroimaging findings, more researchers will adopt this method.

Insights into Transparent Thresholding in Neuroimaging

The paper "Go Figure: Transparency in neuroscience images preserves context and clarifies interpretation" presents a methodological innovation addressing a persistent challenge in neuroimaging: the interpretation and presentation of thresholded brain data. Historically, neuroimaging results have employed "opaque thresholding," highlighting only statistically significant regions. This approach, while simplifying visual representation, obscures important contextual information by hiding subthreshold data, potentially leading to biases and misinterpretations.

Transparency in Neuroimaging Visualization

The authors propose an alternative "transparent thresholding" technique that retains both suprathreshold and subthreshold data in neuroimaging visualizations. By encompassing subthreshold information, this approach offers several advantages over traditional thresholding methods. Notably, transparent thresholding reduces ambiguity, allows for the detection of artifacts, improves cross-study comparisons, and addresses issues related to non-reproducibility. The enhanced visual context provides researchers with a more comprehensive understanding of neuroimaging data, facilitating more nuanced interpretations of brain activity.

Implications of Subthreshold Visibility

By preserving subthreshold data, transparent thresholding challenges traditional interpretations of neuroimaging findings and aligns better with the non-binary nature of brain function. The paper discusses the implications of this visualization strategy using examples from functional magnetic resonance imaging (FMRI) data, among other modalities. By comparing opaque and transparent methodologies, the authors demonstrate the reduction in hypersensitivity to arbitrary parameters like sample size, which can otherwise lead to instability in reported findings.

Practical Applications and Implementations

The paper highlights its practical implementation in a variety of neuroimaging software packages. This broad accessibility makes it feasible for researchers to adopt transparent thresholding, thus enhancing reproducibility and reducing biases such as p-hacking and publication bias. By allowing a more faithful representation of data, this method supports the iterative scientific method, reinforcing the cumulative understanding of brain processes through more reliable meta-analyses.

Future Directions

The adoption of transparent thresholding is poised to impact both theoretical and practical domains in neuroscience. Theoretically, it promotes an anti-localizationist view of brain networks, recognizing the significance of subthreshold activation in broader neural networks. Practically, the use of this method in quality control of FMRI data analysis pipelines demonstrates its utility in improving data evaluation and interpretation. Future developments may see this approach integrated with AI-enhanced image interpretation tools, enabling more sophisticated analyses and potentially uncovering new insights into brain function.

In conclusion, this paper offers a compelling argument for the adoption of transparent thresholding in neuroimaging. Its implementation across a wide range of software and the demonstrated advantages in terms of interpretive depth and reproducibility suggest that this approach could become a new standard in the field, enhancing the clarity and utility of neuroimaging data.