Evaluation of functional MRI-based human brain parcellation: a review

Abstract

Brain parcellations play a crucial role in the analysis of brain imaging data sets, as they can significantly affect the outcome of the analysis. In recent years, several novel approaches for constructing MRI-based brain parcellations have been developed with promising results. In the absence of ground truth, several evaluation approaches have been used to evaluate currently available brain parcellations. In this article, we review and critique methods used for evaluating functional brain parcellations constructed using fMRI data sets. We also describe how some of these evaluation methods have been used to estimate the optimal parcellation granularity. We provide a critical discussion of the current approach to the problem of identifying the optimal brain parcellation that is suited for a given neuroimaging study. We argue that the criteria for an optimal brain parcellation must depend on the application the parcellation is intended for. We describe a teleological approach to the evaluation of brain parcellations, where brain parcellations are evaluated in different contexts and optimal brain parcellations for each context are identified separately. We conclude by discussing several directions for further research that would result in improved evaluation strategies.

Lack of reproducibility of resting-state functional MRI findings in migraine with aura

“…However, we also investigated parcellations with substantially fewer parcels post-hoc to ensure that negative results were not a result of noise and overfitting due to over-parametrization. These parcellation atlases included the 17-network functional parcellation described in a previous publication (30) and the 90-region cortical anatomical parcellation presented previously (31) as these are widely used in earlier studies (32,33). Additionally, we tested several soft parcellation schemes including the Allen and MIALAB ICA probabilistic atlas (34), the 64-region version of the DIFUMO atlas (35), and the 122-region deterministic multiscale functional brain parcellation.…”

Application of deep learning in fMRI-based human brain parcellation: a review

“…Additionally, some alternative strategies have been adopted in parcellation evaluation, including comparison with classical parcellation (e.g. cytoarchitectonic atlas), clustering validity, spatial continuity and performance of downstream tasks [16,29,45], see a comprehensive review by Moghimi [47]. It is worth noting that these evaluation measurements can be employed to search the optimal parameters of parcellation such as parcel number [14,16].…”

“…To evaluate the generated parcellations, researchers have developed a series of metrics such as the accuracy/overlap with target parcellations, reproducibility, homogeneity, spatial continuity/symmetry, and downstream application in brain cognition/disease [40,47], which facilitate a wide range of mining space for deep learning models. However, these numerous metrics also make comparing and choosing between various models extremely difficult.…”

Functional-based parcellation of the mouse prefrontal cortex for network perturbation analysis

“…Second, the generated correlational matrix is further clustered using, for instance, methods from machine learning or graph theory (Arslan et al, 2018). Over the last years FC based parcellations using both resting-state and task-based fMRI were applied in humans to build whole-brain and regional parcellations (Arslan et al, 2018; James, Hazaroglu, & Bush, 2016; Moghimi et al, 2022). Importantly, the interspecies comparison of FC-based parcellations allowed to refine the concept of homologous brain areas.…”