Deep Voxel-Guided Morphometry (VGM): Learning Regional Brain Changes in Serial MRI

Schnurr, Alena-Kathrin; Eisele, Philipp; Roßmanith, Christina; Hoffmann, Stefan; Gregori, Johannes; Dabringhaus, Andreas; Kraemer, Matthias; Kern, Raimar; Gass, Achim; Zöllner, Frank G.

doi:10.1007/978-3-030-66843-3_16

Cited by 4 publications

(14 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The predictability of acute type lesions in MS subjects is an important aspect that can cause an evolution in the treatment procedure and patient's quality of life. A follow‐up perfusion study to predict acute lesions in MS subjects utilizing the deep learning approaches [46] can be an idea for the following research works.…”

Section: Discussionmentioning

confidence: 99%

Perfusion MRI in automatic classification of multiple sclerosis lesion subtypes

Homayouny

Khayati

Nabavi

et al. 2022

IET Signal Processing

View full text Add to dashboard Cite

This retrospective and exploratory study investigated the efficiency of the 3T perfusion magnetic resonance imaging (MRI) at the classification of MS lesion subtypes. For the MS lesion subtype classification, firstly, it was necessary to segment all MS lesions. Therefore, a Bayesian classifier based on the adaptive mixture method was used to segment all lesions, and an artificial neural network (ANN) employed a multi‐layer Perceptron as a subtype classifier. The Bayesian classifier accomplished the segmentation of lesions using Fluid Attenuated Inversion Recovery automatically, and the ANN part was used as a subtype classifier that worked based on extracted information from perfusion MRI (i.e. Mean Transit Time and Cerebral Blood Volume maps) along with the intensity information of the conventional multi‐channel MRI in segmented lesions. Adding 3‐Tesla perfusion MRI to the proposed model for the subtype classification led to an increment of about 7% and 13% in the sensitivity of acute and chronic lesion classifications, respectively. The sensitivity of T2 lesions did not meaningfully change. The overall accuracy of the classification for acute, chronic, and T2 lesion classifications was 96.1%, 90.5%, and 92.9%, respectively. The proposed architectures reached high sensitivity in discrimination between MS lesion subtypes when 3T perfusion MRIs were used.

show abstract

Section: Discussionmentioning

confidence: 99%

Perfusion MRI in automatic classification of multiple sclerosis lesion subtypes

Homayouny

Khayati

Nabavi

et al. 2022

IET Signal Processing

View full text Add to dashboard Cite

show abstract

“…In this retrospective study, we analyzed two datasets of patients with multiple sclerosis (MS) from two different centers, following the 2010 diagnostic criteria by Polman et al ( 2011 ). These datasets are referred to as Dataset A and Dataset B. Dataset A, which comprises 71 patients, is the same dataset utilized in the state-of-the-art method proposed by Schnurr et al ( 2020 ). Dataset B consists of 97 patients.…”

Section: Methodsmentioning

confidence: 99%

“…Furthermore, we compare our trained networks' performance with the baseline U-Net (Ronneberger et al, 2015 ) model obtained from the work of Schnurr et al ( 2020 ).…”

Section: Methodsmentioning

confidence: 99%

“…There is a growing need for automated methods capable of generating complete maps quantifying structural brain tissue changes. Such methods are Voxel-Guided Morphometry (VGM) (Schormann and Kraemer, 2003 ) and deep VGM (Schnurr et al, 2020 ), where a neural network approximates a high dimensional deformation field for detecting changes in MS lesions in longitudinal MRI scans. The deep VGM approach by Schnurr et al ( 2020 ) is fast, however, we intended to improve its robustness making it more applicable to a clinical setting.…”

Section: Introductionmentioning

confidence: 99%

“…Such methods are Voxel-Guided Morphometry (VGM) (Schormann and Kraemer, 2003 ) and deep VGM (Schnurr et al, 2020 ), where a neural network approximates a high dimensional deformation field for detecting changes in MS lesions in longitudinal MRI scans. The deep VGM approach by Schnurr et al ( 2020 ) is fast, however, we intended to improve its robustness making it more applicable to a clinical setting. It is vital to develop a robust deep-VGM approach that is independent of the MRI system used.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A generalizable deep voxel-guided morphometry algorithm for the detection of subtle lesion dynamics in multiple sclerosis

Raj,

Gass,

Eisele

et al. 2024

Front. Neurosci.

Self Cite

View full text Add to dashboard Cite

IntroductionMultiple sclerosis (MS) is a chronic neurological disorder characterized by the progressive loss of myelin and axonal structures in the central nervous system. Accurate detection and monitoring of MS-related changes in brain structures are crucial for disease management and treatment evaluation. We propose a deep learning algorithm for creating Voxel-Guided Morphometry (VGM) maps from longitudinal MRI brain volumes for analyzing MS disease activity. Our approach focuses on developing a generalizable model that can effectively be applied to unseen datasets.MethodsLongitudinal MS patient high-resolution 3D T1-weighted follow-up imaging from three different MRI systems were analyzed. We employed a 3D residual U-Net architecture with attention mechanisms. The U-Net serves as the backbone, enabling spatial feature extraction from MRI volumes. Attention mechanisms are integrated to enhance the model's ability to capture relevant information and highlight salient regions. Furthermore, we incorporate image normalization by histogram matching and resampling techniques to improve the networks' ability to generalize to unseen datasets from different MRI systems across imaging centers. This ensures robust performance across diverse data sources.ResultsNumerous experiments were conducted using a dataset of 71 longitudinal MRI brain volumes of MS patients. Our approach demonstrated a significant improvement of 4.3% in mean absolute error (MAE) against the state-of-the-art (SOTA) method. Furthermore, the algorithm's generalizability was evaluated on two unseen datasets (n = 116) with an average improvement of 4.2% in MAE over the SOTA approach.DiscussionResults confirm that the proposed approach is fast and robust and has the potential for broader clinical applicability.

show abstract

Daten sammeln aus der Praxis: der Weg zum digitalen MS-Zwilling

Voigt¹,

Ziemssen²

2021

DNP

View full text Add to dashboard Cite

Deep Voxel-Guided Morphometry (VGM): Learning Regional Brain Changes in Serial MRI

Cited by 4 publications

References 17 publications

Perfusion MRI in automatic classification of multiple sclerosis lesion subtypes

Perfusion MRI in automatic classification of multiple sclerosis lesion subtypes

A generalizable deep voxel-guided morphometry algorithm for the detection of subtle lesion dynamics in multiple sclerosis

Daten sammeln aus der Praxis: der Weg zum digitalen MS-Zwilling

Contact Info

Product

Resources

About