Modelling Adenocarcinomas In Situ with 3D Cellular Automaton: A Parallel Approach

Abstract

 Adenocarcinomas are solid tumors that begins in the duct architecture of the endocrine glands in human body, constituting some of the most frequent tumor (breast or prostate), whith high morbidity and mortality, and treatment costs in constant growth for public health systems. It is therefore of great importance to detect the disease when it has not yet acquired an invasive character and remains confined (in situ) to the glandular ducts. For this, a deep knowledge of this phase of the neoplasm is important, and computational simulation is another tool that contributes to this knowledge, since it allows the analysis of a multitude of therapeutical strategies in silico, leading to test in vitro and in vivo only the most promising, and reducing laboratory costs. This paper presents a generalized parallel model of adenocarcinomas using three-dimensional cellular automata as a mathematical model of simulation. The model is applied to breast in situ adenocarcinoma (DCIS) to model the evolution of the disease when it is encapsulated in the glandular duct, parallelised using a parallelism scheme by dividing the data domain, identifying the need for synchronization between tasks and solving it properly; a strategy to accelerate computations is proposed by means of an incremental management of the data domain, and finally the results of a pool of experiments are offered, from an implementation developed using the C++ programming language, with good results in terms of computational performance.