Multiple Sclerosis (MS) is considered a neurodegenerative disease that can cause multiple injuries within the Central Nervous System (CNS). MS can be diagnosed and qualitatively investigated with nuclear Magnetic Resonance (MR). In this study, a radiomics analysis is carried out considering 30 patients who underwent through MR studies consisting of Tl-weighted (T1W), T-2-weighted (T2W), Fluid Attenuated Inversion Recovery (FLAIR), and post-contrast administration T1W (T1WKS) images. Since radiomics features can vary greatly in relation to experimental extraction conditions, a first analysis is conducted by calculating the average percentage variations among features as the intensity bin size varies. A second analysis relates to the implementation of the predictive model based on the clinical outcome, namely the Expanded Disability Status Scale (EDSS) of MS. To this aim, the extracted features are reduced, selected, and then used to build the predictive model based on a machine learning algorithm, namely the Linear Discriminant Analysis (LDA). The k-fold strategy is used to split data into training and validation sets. Performance metrics of each model associated with the four MR images (T1W, T2W, FLAIR, and T1WKS) as the bin size varies resulted in optimal predictive values close to 80%.

MRI-based radiomics analysis for identification of features correlated with the expanded disability status scale of multiple sclerosis patients / Nepi, V; Pasini, G; Bini, F; Marinozzi, F; Russo, G; Stefano, A. - 13373:(2022), pp. 362-373. (Intervento presentato al convegno 21st International conference on image analysis and processing , ICIAP 2022 tenutosi a Lecce, Italy) [10.1007/978-3-031-13321-3_32].

MRI-based radiomics analysis for identification of features correlated with the expanded disability status scale of multiple sclerosis patients

Pasini, G;Bini, F;Marinozzi, F;
2022

Abstract

Multiple Sclerosis (MS) is considered a neurodegenerative disease that can cause multiple injuries within the Central Nervous System (CNS). MS can be diagnosed and qualitatively investigated with nuclear Magnetic Resonance (MR). In this study, a radiomics analysis is carried out considering 30 patients who underwent through MR studies consisting of Tl-weighted (T1W), T-2-weighted (T2W), Fluid Attenuated Inversion Recovery (FLAIR), and post-contrast administration T1W (T1WKS) images. Since radiomics features can vary greatly in relation to experimental extraction conditions, a first analysis is conducted by calculating the average percentage variations among features as the intensity bin size varies. A second analysis relates to the implementation of the predictive model based on the clinical outcome, namely the Expanded Disability Status Scale (EDSS) of MS. To this aim, the extracted features are reduced, selected, and then used to build the predictive model based on a machine learning algorithm, namely the Linear Discriminant Analysis (LDA). The k-fold strategy is used to split data into training and validation sets. Performance metrics of each model associated with the four MR images (T1W, T2W, FLAIR, and T1WKS) as the bin size varies resulted in optimal predictive values close to 80%.
2022
21st International conference on image analysis and processing , ICIAP 2022
radiomics; multiple sclerosis; machine learning; image analysis; MRI
04 Pubblicazione in atti di convegno::04b Atto di convegno in volume
MRI-based radiomics analysis for identification of features correlated with the expanded disability status scale of multiple sclerosis patients / Nepi, V; Pasini, G; Bini, F; Marinozzi, F; Russo, G; Stefano, A. - 13373:(2022), pp. 362-373. (Intervento presentato al convegno 21st International conference on image analysis and processing , ICIAP 2022 tenutosi a Lecce, Italy) [10.1007/978-3-031-13321-3_32].
File allegati a questo prodotto
File Dimensione Formato  
Nepi_MRI_2022.pdf

solo gestori archivio

Tipologia: Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 639.99 kB
Formato Adobe PDF
639.99 kB Adobe PDF   Contatta l'autore

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1659656
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 4
  • ???jsp.display-item.citation.isi??? 3
social impact