Zavaglia, M. | bioRxiv Science

bioRxiv2023

Ground-truth validation of uni- and multivariate lesion inference approaches

Lesion analysis aims to reveal causal contributions of brain regions to brain functions. Various strategies have been used for such lesion inferences. These approaches can be broadly categorized as univariate or multivariate methods. Here we analysed data from 581 patients with acute ischemic injury, parcellated into 41 Brodmann areas, and systematically investigated the inferences made by two univariate and two multivariate lesion analysis methods via ground-truth simulations, in which we defined a priori contributions of brain areas to assumed brain function. Particularly, we analysed single-region models, with only single areas presumed to contribute functionally, and multiple-region models, with two contributing regions that interacted in a synergistic, redundant or mutually inhibitory mode. The functional contributions could vary in proportion to the lesion damage or in a binary way. The analyses showed a considerably better performance of the tested multivariate than univariate methods in terms of accuracy and mis-inference error. Specifically, the univariate approaches of Lesion Symptom Mapping (LSM) as well as Lesion Symptom Correlation (LSC) mis-inferred substantial contributions from several areas even in the single-region models, and also after accounting for lesion size. By contrast, the multivariate approaches of Multi- Area Pattern Prediction (MAPP), which is based on machine learning, and Multi-perturbation Shapley value Analysis (MSA), based on coalitional game theory, delivered consistently higher accuracy and specificity. Our findings suggest that the tested multivariate approaches produce largely reliable lesion inferences, without requiring lesion size consideration, while the application of the univariate methods may yield substantial mis-localizations that limit the reliability of functional attributions.

neuroscience↗

bioRxiv2018

Discrimination of the hierarchical structure of cortical layers in 2-photon microscopy data by combined unsupervised and supervised machine learning

The laminar organization of the cerebral cortex is a fundamental characteristic of the brain, with essential implications for cortical function. Due to the rapidly growing amount of high-resolution brain imaging data, a great demand arises for automated and flexible methods for discriminating the laminar texture of the cortex. Here, we propose a combined approach of unsupervised and supervised machine learning to discriminate the hierarchical cortical laminar organization in high-resolution 2-photon microscopic neural image data without observer bias, that is, without the prerequisite of manually labeled training data. For local cortical foci, we modify an unsupervised clustering approach to identify and represent the laminar cortical structure. Subsequently, supervised machine learning is applied to transfer the resulting layer labels across different locations and image data, to ensure the existence of a consistent layer label system. By using neurobiologically meaningful features, the discrimination results are shown to be consistent with the layer classification of the classical Brodmann scheme, and provide additional insight into the structure of the cerebral cortex and its hierarchical organization. Thus, our work paves a new way for studying the anatomical organization of the cerebral cortex, and potentially its functional organization.

neuroscience↗