Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.
Repetitive and alternating lower limb movements are a specific component of human gait. Due to technical challenges, the neural mechanisms underlying such movements have not been previously studied with functional magnetic resonance imaging. In this study, we present a novel treadmill device employed to investigate the kinematics and the brain activation patterns involved in alternating and repetitive movements of the lower limbs. Once inside the scanner, 19 healthy subjects were guided by two visual cues and instructed to perform a motor task which involved repetitive and alternating movements of both lower limbs while selecting their individual comfortable amplitude on the treadmill. The device facilitated the performance of coordinated stepping while registering the concurrent lower-limb displacements, which allowed us to quantify some movement primary kinematic features such as amplitude and frequency. During stepping, significant blood oxygen level dependent signal increases were observed bilaterally in primary and secondary sensorimotor cortex, the supplementary motor area, premotor cortex, prefrontal cortex, superior and inferior parietal lobules, putamen and cerebellum, regions that are known to be involved in lower limb motor control. Brain activations related to individual adjustments during motor performance were identified in a right lateralized network including striatal, extrastriatal, and fronto-parietal areas.
BackgroundImaging studies help to understand the evolution of key cognitive processes related to aging, such as working memory (WM). This study aimed to test three hypotheses in older adults. First, that the brain activation pattern associated to WM processes in elderly during successful low load tasks is located in posterior sensory and associative areas; second, that the prefrontal and parietal cortex and basal ganglia should be more active during high-demand tasks; third, that cerebellar activations are related to high-demand cognitive tasks and have a specific lateralization depending on the condition.MethodsWe used a neuropsychological assessment with functional magnetic resonance imaging and a core N-back paradigm design that was maintained across the combination of four conditions of stimuli and two memory loads in a sample of twenty elderly subjects.ResultsDuring low-loads, activations were located in the visual ventral network. In high loads, there was an involvement of the basal ganglia and cerebellum in addition to the frontal and parietal cortices. Moreover, we detected an executive control role of the cerebellum in a relatively symmetric fronto-parietal network. Nevertheless, this network showed a predominantly left lateralization in parietal regions associated presumably with an overuse of verbal storage strategies. The differential activations between conditions were stimuli-dependent and were located in sensory areas.ConclusionSuccessful WM processes in the elderly population are accompanied by an activation pattern that involves cerebellar regions working together with a fronto-parietal network.
Decision making can be regarded as the outcome of cognitive processes leading to the selection of a course of action among several alternatives. Borrowing a central measurement from information theory, Shannon entropy, we quantified the uncertainties produced by decisions of participants within an economic decision task under different configurations of reward probability and time. These descriptors were used to obtain blood oxygen level-dependent (BOLD) signal correlates of uncertainty and two clusters codifying the Shannon entropy of task configurations were identified: a large cluster including parts of the right middle cingulate cortex (MCC) and left and right pre-supplementary motor areas (pre-SMA) and a small cluster at the left anterior thalamus. Subsequent functional connectivity analyses using the psycho-physiological interactions model identified areas involved in the functional integration of uncertainty. Results indicate that clusters mostly located at frontal and temporal cortices experienced an increased connectivity with the right MCC and left and right pre-SMA as the uncertainty was higher. Furthermore, pre-SMA was also functionally connected to a rich set of areas, most of them associative areas located at occipital and parietal lobes. This study provides a map of the human brain segregation and integration (i.e., neural substrate and functional connectivity respectively) of the uncertainty associated to an economic decision making paradigm.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.