In the past ten years, artificial intelligence has encountered such dramatic progress that it is now seen as a tool of choice to solve environmental issues and, in the first place, greenhouse gas emissions (GHG). At the same time, the deep learning community began to realize that training models with more and more parameters require a lot of energy and, as a consequence, GHG emissions. To our knowledge, questioning the complete net environmental impacts of AI solutions for the environment (AI for Green) and not only GHG, has never been addressed directly. In this article, we propose to study the possible negative impacts of AI for Green. First, we review the different types of AI impacts; then, we present the different methodologies used to assess those impacts and show how to apply life cycle assessment to AI services. Finally, we discuss how to assess the environmental usefulness of a general AI service and point out the limitations of existing work in AI for Green.

Magnetic resonance imaging has proved to be suitable and efficient for in vivo investigation of the early process of brain gyrification in fetuses and preterm newborns but the question remains as to whether cortical-related measurements derived from both cases are comparable or not. Indeed, the developmental folding trajectories drawn up from both populations have not been compared so far, neither from cross-sectional nor from longitudinal datasets. The present study aimed to compare features of cortical folding between healthy fetuses and early imaged preterm newborns on a cross-sectional basis, over a developmental period critical for the folding process (21–36 weeks of gestational age [GA]). A particular attention was carried out to reduce the methodological biases between the 2 populations. To provide an accurate group comparison, several global parameters characterizing the cortical morphometry were derived. In both groups, those metrics provided good proxies for the dramatic brain growth and cortical folding over this developmental period. Except for the cortical volume and the rate of sulci appearance, they depicted different trajectories in both groups suggesting that the transition from into ex utero has a visible impact on cortical morphology that is at least dependent on the GA at birth in preterm newborns.

Cortical folding exhibits both reproducibility and variability in the geometry and topology of its patterns. These two properties are obviously the result of the brain development that goes through local cellular and molecular interactions which have important consequences on the global shape of the cortex. Hypotheses to explain the convoluted aspect of the brain are still intensively debated and do not focus necessarily on the variability of folds. Here we propose a phenomenological model based on reaction-diffusion mechanisms involving Turing morphogens that are responsible for the differential growth of two types of areas, sulci (bottom of folds) and gyri (top of folds). We use a finite element approach of our model that is able to compute the evolution of morphogens on any kind of surface and to deform it through an iterative process. Our model mimics the progressive folding of the cortical surface along foetal development. Moreover it reveals patterns of reproducibility when we look at several realizations of the model from a noisy initial condition. However this reproducibility must be tempered by the fact that a same fold engendered by the model can have different topological properties, in one or several parts. These two results on the reproducibility and variability of the model echo the sulcal roots theory that postulates the existence of anatomical entities around which the folding organizes itself. These sulcal roots would correspond to initial conditions in our model. Last but not least, the parameters of our model are able to produce different kinds of patterns that can be linked to developmental pathologies such as polymicrogyria and lissencephaly. The main significance of our model is that it proposes a first approach to the issue of reproducibility and variability of the cortical folding. The anatomical variability of the human brain folds remains an unclear and challenging issue. However it is clear that this variability is the product of the brain development. Several hypotheses coexist for explaining the rapid development of cortical sulci and it is of the highest interest that understanding their variability would improve the comparison of anatomical and functional data across cohorts of subjects. In this article we propose to extend a model of cortical folding based on interactions between growth factors that shape the cortical surface. First the originality of our approach lies in the fact that the surface on which these mechanisms take place is deformed iteratively and engenders geometric patterns that can be linked to cortical sulci. Secondly we show that some statistical properties of our model can reflect the reproducibility and the variability of sulcal structures. At the end we compare different patterns produced by the model to different pathologies of brain development.

In the context of inter subject brain surface matching, we present a parameterization of the cortical surface constrained by a model of cortical organization. The parameterization is defined via an harmonic mapping of each hemisphere surface to a rectangular planar domain that integrates a representation of the model. As opposed to previous landmark-based registration methods we do not match folds between individuals but instead optimize the fit between cortical sulci and specific iso-coordinate axis in the model. This strategy overcomes some limitation to sulcus-based registration techniques such as topological variability in sulcal landmarks across subjects. Experiments on 62 subjects with manually traced sulci are presented and compared with the result of the Freesurfer software. The evaluation involves a measure of dispersion of sulci with both angular and area distortions. We show that the model-based strategy can lead to a natural, efficient and very fast (less than 5 min per hemisphere) method for defining inter subjects correspondences. We discuss how this approach also reduces the problems inherent to anatomically defined landmarks and open the way to the investigation of cortical organization through the notion of orientation and alignment of structures across the cortex.