Radiomics and machine learning applied to STIR sequence for prediction of quantitative parameters in facioscapulohumeral disease

Quantitative Muscle MRI (qMRI) is a valuable and non-invasive tool to assess disease involvement and progression in neuromuscular disorders being able to detect even subtle changes in muscle pathology. The aim of this study is to evaluate the feasibility of using a conventional short-tau inversion r...

Full description

Saved in:
Bibliographic Details
Published in:Frontiers in neurology Vol. 14; p. 1105276
Main Authors: Colelli, Giulia, Barzaghi, Leonardo, Paoletti, Matteo, Monforte, Mauro, Bergsland, Niels, Manco, Giulia, Deligianni, Xeni, Santini, Francesco, Ricci, Enzo, Tasca, Giorgio, Mira, Antonietta, Figini, Silvia, Pichiecchio, Anna
Format: Journal Article
Language:English
Published: Switzerland Frontiers Media S.A 24.02.2023
Subjects:
FSHD
machine learning
muscle MRI
Neurology
radiomics
stir
FSHD
radiomics
muscle MRI
machine learning
stir
ISSN:1664-2295, 1664-2295
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Quantitative Muscle MRI (qMRI) is a valuable and non-invasive tool to assess disease involvement and progression in neuromuscular disorders being able to detect even subtle changes in muscle pathology. The aim of this study is to evaluate the feasibility of using a conventional short-tau inversion recovery (STIR) sequence to predict fat fraction (FF) and water T2 (wT2) in skeletal muscle introducing a radiomic workflow with standardized feature extraction combined with machine learning algorithms. Twenty-five patients with facioscapulohumeral muscular dystrophy (FSHD) were scanned at calf level using conventional STIR sequence and qMRI techniques. We applied and compared three different radiomics workflows ( ), combined with seven Machine Learning regression algorithms (linear, ridge and lasso regression, tree, random forest, k-nearest neighbor and support vector machine), on conventional STIR images to predict FF and wT2 for six calf muscles. The combination of WF3 and K-nearest neighbor resulted to be the best predictor model of qMRI parameters with a mean absolute error about ± 5 for FF and ± 1.8 for wT2. This pilot study demonstrated the possibility to predict qMRI parameters in a cohort of FSHD subjects starting from conventional STIR sequence.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
Edited by: Jordi Diaz-Manera, University of Newcastle, United Kingdom
These authors share last authorship
Reviewed by: Teresa Gerhalter, University Hospital Erlangen, Germany; Jorge Alonso-Pérez, Hospital Santa Creu i Sant Pau, Spain; José Verdú-Díaz, Newcastle University, United Kingdom
These authors share first authorship
This article was submitted to Applied Neuroimaging, a section of the journal Frontiers in Neurology
ISSN:1664-2295
1664-2295
DOI:10.3389/fneur.2023.1105276