1
|
Cruciani F, Brusini L, Zucchelli M, Retuci Pinheiro G, Setti F, Boscolo Galazzo I, Deriche R, Rittner L, Calabrese M, Menegaz G. Interpretable deep learning as a means for decrypting disease signature in multiple sclerosis. J Neural Eng 2021; 18. [PMID: 34181581 DOI: 10.1088/1741-2552/ac0f4b] [Citation(s) in RCA: 4] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/09/2021] [Accepted: 06/28/2021] [Indexed: 11/12/2022]
Abstract
Objective.The mechanisms driving multiple sclerosis (MS) are still largely unknown, calling for new methods allowing to detect and characterize tissue degeneration since the early stages of the disease. Our aim is to decrypt the microstructural signatures of the Primary Progressive versus the Relapsing-Remitting state of disease based on diffusion and structural magnetic resonance imaging data.Approach.A selection of microstructural descriptors, based on the 3D-Simple Harmonics Oscillator Based Reconstruction and Estimation and the set of new algebraically independent Rotation Invariant spherical harmonics Features, was considered and used to feed convolutional neural networks (CNNs) models. Classical measures derived from diffusion tensor imaging, that are fractional anisotropy and mean diffusivity, were used as benchmark for diffusion MRI (dMRI). Finally, T1-weighted images were also considered for the sake of comparison with the state-of-the-art. A CNN model was fit to each feature map and layerwise relevance propagation (LRP) heatmaps were generated for each model, target class and subject in the test set. Average heatmaps were calculated across correctly classified patients and size-corrected metrics were derived on a set of regions of interest to assess the LRP contrast between the two classes.Main results.Our results demonstrated that dMRI features extracted in grey matter tissues can help in disambiguating primary progressive multiple sclerosis from relapsing-remitting multiple sclerosis patients and, moreover, that LRP heatmaps highlight areas of high relevance which relate well with what is known from literature for MS disease.Significance.Within a patient stratification task, LRP allows detecting the input voxels that mostly contribute to the classification of the patients in either of the two classes for each feature, potentially bringing to light hidden data properties which might reveal peculiar disease-state factors.
Collapse
Affiliation(s)
- F Cruciani
- Department of Computer Science, University of Verona, Verona, Italy
| | - L Brusini
- Department of Computer Science, University of Verona, Verona, Italy
| | - M Zucchelli
- Athena Project-Team, Inria Sophia Antipolis-Méditerranée, Université Côte d'Azur, Sophia Antipolis, France
| | - G Retuci Pinheiro
- MICLab, School of Electrical and Computer Engineering (FEEC), UNICAMP, Campinas, Brazil
| | - F Setti
- Department of Computer Science, University of Verona, Verona, Italy
| | | | - R Deriche
- Athena Project-Team, Inria Sophia Antipolis-Méditerranée, Université Côte d'Azur, Sophia Antipolis, France
| | - L Rittner
- MICLab, School of Electrical and Computer Engineering (FEEC), UNICAMP, Campinas, Brazil
| | - M Calabrese
- Department of Neurosciences, Biomedicine and Movement, University of Verona, Verona, Italy
| | - G Menegaz
- Department of Computer Science, University of Verona, Verona, Italy
| |
Collapse
|
2
|
Microstructural MRI Correlates of Cognitive Impairment in Multiple Sclerosis: The Role of Deep Gray Matter. Diagnostics (Basel) 2021; 11:diagnostics11061103. [PMID: 34208650 PMCID: PMC8234586 DOI: 10.3390/diagnostics11061103] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/08/2021] [Revised: 06/07/2021] [Accepted: 06/10/2021] [Indexed: 11/24/2022] Open
Abstract
Although cognitive impairment (CI) is frequently observed in people with multiple sclerosis (pwMS), its pathogenesis is still controversial. Conflicting results emerged concerning the role of microstructural gray matter (GM) damage especially when involving the deep GM structures. In this study, we aimed at evaluating whether differences in cortical and deep GM structures between apparently cognitively normal (ACN) and CI pwMS (36 subjects in total) are present, using an extensive set of diffusion MRI (dMRI) indices and conventional morphometry measures. The results revealed increased anisotropy and restriction over several deep GM structures in CI compared with ACN pwMS, while no changes in volume were present in the same areas. Conversely, reduced anisotropy/restriction values were detected in cortical regions, mostly the pericalcarine cortex and precuneus, combined with reduced thickness of the superior frontal gyrus and insula. Most of the dMRI metrics but none of the morphometric indices correlated with the Symbol Digit Modality Test. These results suggest that deep GM microstructural damage can be a strong anatomical substrate of CI in pwMS and might allow identifying pwMS at higher risk of developing CI.
Collapse
|