1
|
Marzi C, Giannelli M, Barucci A, Tessa C, Mascalchi M, Diciotti S. Efficacy of MRI data harmonization in the age of machine learning: a multicenter study across 36 datasets. Sci Data 2024; 11:115. [PMID: 38263181 PMCID: PMC10805868 DOI: 10.1038/s41597-023-02421-7] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/06/2022] [Accepted: 07/27/2023] [Indexed: 01/25/2024] Open
Abstract
Pooling publicly-available MRI data from multiple sites allows to assemble extensive groups of subjects, increase statistical power, and promote data reuse with machine learning techniques. The harmonization of multicenter data is necessary to reduce the confounding effect associated with non-biological sources of variability in the data. However, when applied to the entire dataset before machine learning, the harmonization leads to data leakage, because information outside the training set may affect model building, and potentially falsely overestimate performance. We propose a 1) measurement of the efficacy of data harmonization; 2) harmonizer transformer, i.e., an implementation of the ComBat harmonization allowing its encapsulation among the preprocessing steps of a machine learning pipeline, avoiding data leakage by design. We tested these tools using brain T1-weighted MRI data from 1740 healthy subjects acquired at 36 sites. After harmonization, the site effect was removed or reduced, and we showed the data leakage effect in predicting individual age from MRI data, highlighting that introducing the harmonizer transformer into a machine learning pipeline allows for avoiding data leakage by design.
Collapse
Affiliation(s)
- Chiara Marzi
- Department of Statistics, Computer Science and Applications "Giuseppe Parenti", University of Florence, 50134, Florence, Italy
- "Nello Carrara" Institute of Applied Physics (IFAC), National Research Council (CNR), 50019, Sesto Fiorentino, Florence, Italy
| | - Marco Giannelli
- Unit of Medical Physics, Pisa University Hospital "Azienda Ospedaliero-Universitaria Pisana", 56126, Pisa, Italy
| | - Andrea Barucci
- "Nello Carrara" Institute of Applied Physics (IFAC), National Research Council (CNR), 50019, Sesto Fiorentino, Florence, Italy
| | - Carlo Tessa
- Radiology Unit Apuane e Lunigiana, Azienda USL Toscana Nord Ovest, 54100, Massa, Italy
| | - Mario Mascalchi
- Department of Experimental and Clinical Biomedical Sciences "Mario Serio", University of Florence, 50139, Florence, Italy
- Division of Epidemiology and Clinical Governance, Institute for Study, Prevention and netwoRk in Oncology (ISPRO), 50139, Florence, Italy
| | - Stefano Diciotti
- Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi" - DEI, University of Bologna, 47522, Cesena, Italy.
- Alma Mater Research Institute for Human-Centered Artificial Intelligence, University of Bologna, 40121, Bologna, Italy.
| |
Collapse
|
2
|
Marzi C, Scheda R, Salvadori E, Giorgio A, De Stefano N, Poggesi A, Inzitari D, Pantoni L, Mascalchi M, Diciotti S. Fractal dimension of the cortical gray matter outweighs other brain MRI features as a predictor of transition to dementia in patients with mild cognitive impairment and leukoaraiosis. Front Hum Neurosci 2023; 17:1231513. [PMID: 37822707 PMCID: PMC10562576 DOI: 10.3389/fnhum.2023.1231513] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/30/2023] [Accepted: 08/31/2023] [Indexed: 10/13/2023] Open
Abstract
Background The relative contribution of changes in the cerebral white matter (WM) and cortical gray matter (GM) to the transition to dementia in patients with mild cognitive impairment (MCI) is not yet established. In this longitudinal study, we aimed to analyze MRI features that may predict the transition to dementia in patients with MCI and T2 hyperintensities in the cerebral WM, also known as leukoaraiosis. Methods Sixty-four participants with MCI and moderate to severe leukoaraiosis underwent baseline MRI examinations and annual neuropsychological testing over a 2 year period. The diagnosis of dementia was based on established criteria. We evaluated demographic, neuropsychological, and several MRI features at baseline as predictors of the clinical transition. The MRI features included visually assessed MRI features, such as the number of lacunes, microbleeds, and dilated perivascular spaces, and quantitative MRI features, such as volumes of the cortical GM, hippocampus, T2 hyperintensities, and diffusion indices of the cerebral WM. Additionally, we examined advanced quantitative features such as the fractal dimension (FD) of cortical GM and WM, which represents an index of tissue structural complexity derived from 3D-T1 weighted images. To assess the prediction of transition to dementia, we employed an XGBoost-based machine learning system using SHapley Additive exPlanations (SHAP) values to provide explainability to the machine learning model. Results After 2 years, 18 (28.1%) participants had transitioned from MCI to dementia. The area under the receiving operator characteristic curve was 0.69 (0.53, 0.85) [mean (90% confidence interval)]. The cortical GM-FD emerged as the top-ranking predictive feature of transition. Furthermore, aggregated quantitative neuroimaging features outperformed visually assessed MRI features in predicting conversion to dementia. Discussion Our findings confirm the complementary roles of cortical GM and WM changes as underlying factors in the development of dementia in subjects with MCI and leukoaraiosis. FD appears to be a biomarker potentially more sensitive than other brain features.
Collapse
Affiliation(s)
- Chiara Marzi
- Department of Statistics, Computer Science, Applications "Giuseppe Parenti, " University of Florence, Florence, Italy
| | - Riccardo Scheda
- Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi, " University of Bologna, Cesena, Italy
| | - Emilia Salvadori
- NEUROFARBA Department, Neuroscience Section, University of Florence, Florence, Italy
| | - Antonio Giorgio
- Department of Medicine, Surgery, and Neuroscience, University of Siena, Siena, Italy
| | - Nicola De Stefano
- Department of Medicine, Surgery, and Neuroscience, University of Siena, Siena, Italy
| | - Anna Poggesi
- NEUROFARBA Department, Neuroscience Section, University of Florence, Florence, Italy
| | - Domenico Inzitari
- NEUROFARBA Department, Neuroscience Section, University of Florence, Florence, Italy
| | - Leonardo Pantoni
- Department of Biomedical and Clinical Sciences, University of Milan, Milan, Italy
| | - Mario Mascalchi
- Department of Experimental and Clinical Biomedical Sciences "Mario Serio, " University of Florence, Florence, Italy
- Division of Epidemiology and Clinical Governance, Institute for Study, Prevention and Network in Oncology (ISPRO), Florence, Italy
| | - Stefano Diciotti
- Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi, " University of Bologna, Cesena, Italy
- Alma Mater Research Institute for Human-Centered Artificial Intelligence, University of Bologna, Bologna, Italy
| |
Collapse
|
3
|
Elsherbini A, Zhu Z, Quadri Z, Crivelli SM, Ren X, Vekaria HJ, Tripathi P, Zhang L, Zhi W, Bieberich E. Novel Isolation Method Reveals Sex-Specific Composition and Neurotoxicity of Small Extracellular Vesicles in a Mouse Model of Alzheimer's Disease. Cells 2023; 12:1623. [PMID: 37371093 PMCID: PMC10297289 DOI: 10.3390/cells12121623] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/11/2023] [Revised: 06/01/2023] [Accepted: 06/07/2023] [Indexed: 06/29/2023] Open
Abstract
We developed a new method to isolate small extracellular vesicles (sEVs) from male and female wild-type and 5xFAD mouse brains to investigate the sex-specific functions of sEVs in Alzheimer's disease (AD). A mass spectrometric analysis revealed that sEVs contained proteins critical for EV formation and Aβ. ExoView analysis showed that female mice contained more GFAP and Aβ-labeled sEVs, suggesting that a larger proportion of sEVs from the female brain is derived from astrocytes and/or more likely to bind to Aβ. Moreover, sEVs from female brains had more acid sphingomyelinase (ASM) and ceramide, an enzyme and its sphingolipid product important for EV formation and Aβ binding to EVs, respectively. We confirmed the function of ASM in EV formation and Aβ binding using co-labeling and proximity ligation assays, showing that ASM inhibitors prevented complex formation between Aβ and ceramide in primary cultured astrocytes. Finally, our study demonstrated that sEVs from female 5xFAD mice were more neurotoxic than those from males, as determined by impaired mitochondrial function (Seahorse assays) and LDH cytotoxicity assays. Our study suggests that sex-specific sEVs are functionally distinct markers for AD and that ASM is a potential target for AD therapy.
Collapse
Affiliation(s)
- Ahmed Elsherbini
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Zhihui Zhu
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Zainuddin Quadri
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Simone M. Crivelli
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Xiaojia Ren
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Hemendra J. Vekaria
- Spinal Cord and Brain Injury Research Center (SCoBIRC), University of Kentucky, Lexington, KY 40536, USA;
- Veterans Affairs Medical Center, Lexington, KY 40502, USA
| | - Priyanka Tripathi
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Liping Zhang
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
| | - Wenbo Zhi
- Department of Center for Biotechnology and Genomic Medicine, Augusta University, Augusta, GA 30912, USA;
| | - Erhard Bieberich
- Department of Physiology, University of Kentucky College of Medicine, Lexington, KY 40536, USA; (A.E.); (Z.Z.); (Z.Q.); (S.M.C.); (X.R.); (P.T.); (L.Z.)
- Veterans Affairs Medical Center, Lexington, KY 40502, USA
| |
Collapse
|
4
|
Nazlee N, Waiter GD, Sandu A. Age-associated sex and asymmetry differentiation in hemispheric and lobar cortical ribbon complexity across adulthood: A UK Biobank imaging study. Hum Brain Mapp 2022; 44:49-65. [PMID: 36574599 PMCID: PMC9783444 DOI: 10.1002/hbm.26076] [Citation(s) in RCA: 8] [Impact Index Per Article: 4.0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/14/2021] [Revised: 07/28/2022] [Accepted: 08/21/2022] [Indexed: 02/01/2023] Open
Abstract
Cortical morphology changes with ageing and age-related neurodegenerative diseases. Previous studies suggest that the age effect is more pronounced in the frontal lobe. However, our knowledge of structural complexity changes in male and female brains is still limited. We measured cortical ribbon complexity through fractal dimension (FD) analysis at the hemisphere and lobe level in 7010 individuals from the UK Biobank imaging cohort to study age-related sex differences (3332 males, age ranged 45-79 years). FD decreases significantly with age and sexual dimorphism exists. With correction for brain size, females showed higher complexity in the left hemisphere and left and right parietal lobes whereas males showed higher complexity in the right temporal and left and right occipital lobes. A nonlinear age effect was observed in the left and right frontal, and right temporal lobes. Differential patterns of age effects were observed in both sexes with relatively more age-affected regions in males. Significantly higher rightward asymmetries at hemisphere, frontal, parietal, and occipital lobe level and higher leftward asymmetry in temporal lobe were observed. There was no age-by-sex-by asymmetry interaction in any region. When controlling for brain size, the leftward hemispheric, and temporal lobe asymmetry decreased with age. Males had significantly lower asymmetry between hemispheres and higher asymmetry in the parietal and occipital lobes than females. This work provides distinct patterns of age-related sex and asymmetry differences that can aid in the future development of sex-specific models of the normal brain to ascribe cognitive functional significance of these patterns in ageing.
Collapse
Affiliation(s)
- Nafeesa Nazlee
- Aberdeen Biomedical Imaging CentreUniversity of AberdeenAberdeenScotland
| | - Gordon D. Waiter
- Aberdeen Biomedical Imaging CentreUniversity of AberdeenAberdeenScotland
| | - Anca‐Larisa Sandu
- Aberdeen Biomedical Imaging CentreUniversity of AberdeenAberdeenScotland
| |
Collapse
|