Logo image
Back
Optical Biomarkers for Cerebral Injury Detection and Classification Derived From Diffuse Optical Spectroscopy
Journal article   Peer reviewed

Optical Biomarkers for Cerebral Injury Detection and Classification Derived From Diffuse Optical Spectroscopy

Luis Miguel Gomero, Meltem Izzetoglu, Randolph Seno Sinahon, Shadi Malaeb and Kurtulus Izzetoglu
IEEE transactions on instrumentation and measurement, v 74, pp 1-1
07 Oct 2025
DOI: https://doi.org/10.1109/TIM.2025.3619245
Featured in Collection :   UN Sustainable Development Goals @ Drexel

Abstract

Accuracy Animal model (piglet) Biomedical monitoring Biomedical optical imaging Brain injuries cerebral blood flow & oxygenation diffuse correlation spectroscopy (DCS) Electric shock hemorrhagic shock Hemorrhaging injury classification machine learning (ML) Monitoring near-infrared spectroscopy (NIRS) Optical imaging Support vector machines Blood Flow Hypoxia
Shock is a severe condition resulting from substantial fluid loss, leading to significant reductions in blood volume, inadequate tissue perfusion, and cellular hypoxia. Identification and detection of its type (i.e. hemorrhagic or hypoxic) and prediction and monitoring of its progression from pre-shock to shock can help in the selection of appropriate treatment and management procedures which in turn can help in the improvement of patient outcomes and hence reduce morbidity and mortality rates. Various brain imaging modalities employing noninvasive, safe, and portable techniques such as optics-based diffuse correlation spectroscopy (DCS) and near-infrared spectroscopy (NIRS) can help in the continuous monitoring of changes in cerebral blood flow (CBF) and oxygenation in relation to shock and its progression in a direct and timely manner and even at the site of injury. In this study we aimed to apply various machine learning models and data augmentation techniques to reliably classify stages and types of shock using our previously collected simultaneous DCS and NIRS data set and time and frequency domain CBF and oxygenation markers from piglet models of hemorrhagic and hypoxic shock in comparison to each other and to sham controls. Our results demonstrate that integrating optical technologies with powerful machine learning algorithms can accurately classify presence/absence of shock (100% accuracy for hemorrhagic and up to 100% for hypoxic shock), stages of hemorrhagic shock (up to 87.50% accuracy), hypoxic shock (up to 78.33%) and shock type (hemorrhagic vs hypoxic with up to 96.19% accuracy) providing a rapid, non-invasive, and reliable tool for the clinical assessment of shock and its progression. Overall, the Radial basis function support vector machine (RBF SVM) consistently achieved strong results and robust performance across all cases considering accuracy, Matthews correlation coefficients (MCC), and F1-scores.

Metrics

7 Record Views

Details

UN Sustainable Development Goals (SDGs)

This publication has contributed to the advancement of the following goals:

#3 Good Health and Well-Being

InCites Highlights

Data related to this publication, from InCites Benchmarking & Analytics tool:

Collaboration types
Domestic collaboration
Web of Science research areas
Engineering, Electrical & Electronic
Instruments & Instrumentation
Logo image