Automated Data Transformation and Feature Extraction for Oxygenation-Sensitive Cardiovascular Magnetic Resonance Images.

Oxygenation-sensitive cardiovascular magnetic resonance (OS-CMR) is a novel, powerful tool for assessing coronary function in vivo. The data extraction and analysis however are labor-intensive. The objective of this study was to provide an automated approach for the extraction, visualization, and biomarker selection of OS-CMR images. We created a Python-based tool to automate extraction and export of raw patient data, featuring 3336 attributes per participant, into a template compatible with common data analytics frameworks, including the functionality to select predictive features for the given disease state. Each analysis was completed in about 2 min. The features selected by both ANOVA and MIC significantly outperformed (p < 0.001) the null set and complete set of features in two datasets, with mean AUROC scores of 0.89eatures f 0.94lete set of features in two datasets, with mean AUROC scores that our tool is suitable for automated data extraction and analysis of OS-CMR images.

A Metaheuristic for Amortized Search in High-Dimensional Parameter Spaces.

Parameter inference for dynamical models of (bio)physical systems remains a challenging problem. Intractable gradients, high-dimensional spaces, and non-linear model functions are typically problematic without large computational budgets. A recent body of work in that area has focused on Bayesian inference methods, which consider parameters under their statistical distributions and therefore, do not derive point estimates of optimal parameter values. Here we propose a new metaheuristic that drives dimensionality reductions from feature-informed transformations (DR-FFIT) to address these bottlenecks. DR-FFIT implements an efficient sampling strategy that facilitates a gradient-free parameter search in high-dimensional spaces. We use artificial neural networks to obtain differentiable proxies for the model's features of interest. The resulting gradients enable the estimation of a local active subspace of the model within a defined sampling region. This approach enables efficient dimensionality reductions of highly non-linear search spaces at a low computational cost. Our test data show that DR-FFIT boosts the performances of random-search and simulated-annealing against well-established metaheuristics, and improves the goodness-of-fit of the model, all within contained run-time costs.