Desarrollo de un prototipo de aplicativo web para el análisis de lesiones mamarias a partir de mamografías mediante simulación mesoscópica

Abstract

The analysis of diffusion in breast lesions requires specialized imaging techniques and infrastructure, which may have limited availability in certain research environments. This constraint motivates computational alternatives based on standard mammograms. To address this need, a web prototype was developed to analyze regions of interest (ROIs) using mesoscopic simulation. The model simulates the motion of particles representing water within breast tissue, whose heterogeneity is derived from the ROI intensity levels. The velocity autocorrelation function (Cv) measures the degree to which this motion retains its initial direction. The characteristic time (tauC) estimates when this directional memory is lost. The mesoscopic diffusion coefficient (MDC) summarizes the ease of particle movement within the simulated domain, while the normalized coefficient (MDC*) related the MDC to a free-diffusion reference for inter-simulation comparison. The proposed solution utilizes a distributed architecture that separates the web application from the computationally intensive processing executed on a university server. The system provides users with an accessible environment to manage cases, check their status, and review results, ensuring full traceability among images, parameters, simulations, and generated reports. Furthermore, it privately stores information and allows simulations to run independently of the user’s computer. For the same ROI and equivalent parameters, processing time decreased from 18 hours and 40 minutes in the local environment to 2 hours and 55 minutes on the server, representing a 84.4% reduction in execution time. Additionally, load testing with k6 yielded zero HTTP errors. These results demonstrate the viability of the prototype as a usable, accessible, traceable, and reproducible academic tool.