arXiv subjects

Guilherme J. M. Garcia

Publications and source records attributed to Guilherme J. M. Garcia.

6 recordsLinked to original sources

arXiv2026

Airway cross-sectional area measurement using quantitative endoscopy with AI-enabled depth estimation: A proof-of-concept study

Endoscopy is widely used in medicine to visualize anatomical structures for diagnosis and treatment evaluation. However, it does not currently allow rapid, quantitative measurement of airway cross-sectional area (CSA), a key metric for diagnosing and managing respiratory pathologies. There is growing interest in AI-enabled quantitative endoscopy using depth estimation, but a fundamental challenge is the need for ground-truth depth data to train algorithms capable of quantifying depth in physical units (e.g., meters). In this study, 60,000 paired virtual endoscopy images and depth maps were created in Blender from 3D airway models built from computed tomography. A U-Net convolutional neural network was trained to predict the depth maps from the endoscopy images. Point clouds were reconstructed from the AI-generated depth maps using the Equisolid fisheye lens model, and airway perimeters were identified by intersecting a plane with the point cloud to estimate CSA. Nasopharyngeal CSA, a key metric in obstructive sleep apnea, was quantified using this AI-enabled approach with a median error of 5.3 \% relative to the 3D model ground-truth. This pilot study using synthetic data provides strong evidence for the feasibility of developing AI-enabled quantitative endoscopy for airway CSA measurement.

eess.IV↗

arXiv2004

Absorbing-state phase transitions with extremal dynamics

Extremal dynamics represents a path to self-organized criticality in which the order parameter is tuned to a value of zero. The order parameter is associated with a phase transition to an absorbing state. Given a process that exhibits a phase transition to an absorbing state, we define an ``extremal absorbing" process, providing the link to the associated extremal (nonabsorbing) process. Stationary properties of the latter correspond to those at the absorbing-state phase transition in the former. Studying the absorbing version of an extremal dynamics model allows to determine certain critical exponents that are not otherwise accessible. In the case of the Bak-Sneppen (BS) model, the absorbing version is closely related to the "$f$-avalanche" introduced by Paczuski, Maslov and Bak [Phys. Rev. E {\bf 53}, 414 (1996)], or, in spreading simulations to the "BS branching process" also studied by these authors. The corresponding nonextremal process belongs to the directed percolation universality class. We revisit the absorbing BS model, obtaining refined estimates for the threshold and critical exponents in one dimension. We also study an extremal version of the usual contact process, using mean-field theory and simulation. The extremal condition slows the spread of activity and modifies the critical behavior radically, defining an ``extremal directed percolation" universality class of absorbing-state phase transitions. Asymmetric updating is a relevant perturbation for this class, even though it is irrelevant for the corresponding nonextremal class.

cond-mat.stat-mech↗

arXiv2004

The Scaling of Maximum and Basal Metabolic Rates of Mammals and Birds

Allometric scaling is one of the most pervasive laws in biology. Its origin, however, is still a matter of dispute. Recent studies have established that maximum metabolic rate scales with an exponent larger than that found for basal metabolism. This unpredicted result sets a challenge that can decide which of the concurrent hypotheses is the correct theory. Here we show that both scaling laws can be deduced from a single network model. Besides the 3/4-law for basal metabolism, the model predicts that maximum metabolic rate scales as $M^{6/7}$, maximum heart rate as $M^{-1/7}$, and muscular capillary density as $M^{-1/7}$, in agreement with data.

physics.bio-ph↗

arXiv2004

Asymmetric dynamics and critical behavior in the Bak-Sneppen model

We investigate, using mean-field theory and simulation, the effect of asymmetry on the critical behavior and probability density of Bak-Sneppen models. Two kinds of anisotropy are investigated: (i) different numbers of sites to the left and right of the central (minimum) site are updated and (ii) sites to the left and right of the central site are renewed in different ways. Of particular interest is the crossover from symmetric to asymmetric scaling for weakly asymmetric dynamics, and the collapse of data with different numbers of updated sites but the same degree of asymmetry. All non-symmetric rules studied fall, independent of the degree of asymmetry, in the same universality class. Conversely, symmetric variants reproduce the exponents of the original model. Our results confirm the existence of two symmetry-based universality classes for extremal dynamics.

cond-mat.stat-mech↗

arXiv2004

On singular probability densities generated by extremal dynamics

Extremal dynamics is the mechanism that drives the Bak-Sneppen model into a (self-organized) critical state, marked by a singular stationary probability density $p(x)$. With the aim of understanding this phenomenon, we study the BS model and several variants via mean-field theory and simulation. In all cases, we find that $p(x)$ is singular at one or more points, as a consequence of extremal dynamics. Furthermore we show that the extremal barrier $x_i$ always belongs to the `prohibited' interval, in which $p(x)=0$. Our simulations indicate that the Bak-Sneppen universality class is robust with regard to changes in the updating rule: we find the same value for the exponent $π$ for all variants. Mean-field theory, which furnishes an exact description for the model on a complete graph, reproduces the character of the probability distribution found in simulations. For the modified processes mean-field theory takes the form of a functional equation for $p(x)$.

cond-mat.stat-mech↗

arXiv2004

On the thresholds, probability densities, and critical exponents of Bak-Sneppen-like models

We report a simple method to accurately determine the threshold and the exponent $ν$ of the Bak-Sneppen model and also investigate the BS universality class. For the random-neighbor version of the BS model, we find the threshold $x^*=0.33332(3)$, in agreement with the exact result $x^*=1/3$ given by mean-field theory. For the one-dimensional original model, we find $x^*=0.6672(2)$ in good agreement with the results reported in the literature; for the anisotropic BS model we obtain $x^*=0.7240(1)$. We study the finite size effect $x^*(L)-x^*(L \to \infty) \propto L^{-ν}$, observed in a system with $L$ sites, and find $ν= 1.00(1)$ for the random-neighbor version, $ν= 1.40(1)$ for the original model, and $ν=1.58(1)$ for the anisotropic case. Finally, we discuss the effect of defining the extremal site as the one which minimizes a general function $f(x)$, instead of simply $f(x)=x$ as in the original updating rule. We emphasize that models with extremal dynamics have singular stationary probability distributions $p(x)$. Our simulations indicate the existence of two symmetry-based universality classes.

cond-mat.stat-mech↗