A Correspondence-Based Network Approach for Groupwise Analysis of Patient-Specific Spatiotemporal Data.

Autor: Atkins PR; Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT, USA.; Department of Orthopaedics, University of Utah, Salt Lake City, UT, USA., Morris A; Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT, USA., Elhabian SY; Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT, USA.; School of Computing, University of Utah, Salt Lake City, UT, USA., Anderson AE; Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT, USA. aea4@utah.edu.; Department of Orthopaedics, University of Utah, Salt Lake City, UT, USA. aea4@utah.edu.; Department of Biomedical Engineering, University of Utah, Salt Lake City, UT, USA. aea4@utah.edu.; Department of Physical Therapy, University of Utah, Salt Lake City, UT, USA. aea4@utah.edu.
Jazyk: angličtina
Zdroj: Annals of biomedical engineering [Ann Biomed Eng] 2023 Oct; Vol. 51 (10), pp. 2289-2300. Date of Electronic Publication: 2023 Jun 25.
DOI: 10.1007/s10439-023-03270-6
Abstrakt: Methods for statistically analyzing patient-specific data that vary both spatially and over time are currently either limited to summary statistics or require elaborate surface registration. We propose a new method, called correspondence-based network analysis, which leverages particle-based shape modeling to establish correspondence across a population and preserve patient-specific measurements and predictions through statistical analysis. Herein, we evaluated this method using three published datasets of the hip describing cortical bone thickness of the proximal femur, cartilage contact stress, and dynamic joint space between control and patient cohorts to evaluate activity- and group-based differences, as applicable, using traditional statistical parametric mapping (SPM) and our proposed spatially considerate correspondence-based network analysis approach. The network approach was insensitive to correspondence density, while the traditional application of SPM showed decreasing area of the region of significance with increasing correspondence density. In comparison to SPM, the network approach identified broader and more connected regions of significance for all three datasets. The correspondence-based network analysis approach identified differences between groups and activities without loss of subject and spatial specificity which could improve clinical interpretation of results.
(© 2023. The Author(s) under exclusive licence to Biomedical Engineering Society.)
Databáze: MEDLINE
Externí odkaz: