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.

Patients with cerebral small vessel disease (SVD) frequently show decline in cognitive performance. However, neuroimaging in SVD patients discloses a wide range of brain lesions and alterations so that it is often difficult to understand which of these changes are the most relevant for cognitive decline. It has also become evident that visually-rated alterations do not fully explain the neuroimaging correlates of cognitive decline in SVD. Fractal dimension (FD), a unitless feature of structural complexity that can be computed from high-resolution T1-weighted images, has been recently applied to the neuroimaging evaluation of the human brain. Indeed, white matter (WM) and cortical gray matter (GM) exhibit an inherent structural complexity that can be measured through the FD. In our study, we included 64 patients (mean age ± standard deviation, 74.6 ± 6.9, education 7.9 ± 4.2 years, 53% males) with SVD and mild cognitive impairment (MCI), and a control group of 24 healthy subjects (mean age ± standard deviation, 72.3 ± 4.4 years, 50% males). With the aim of assessing whether the FD values of cerebral WM (WM FD) and cortical GM (GM FD) could be valuable structural predictors of cognitive performance in patients with SVD and MCI, we employed a machine learning strategy based on LASSO (least absolute shrinkage and selection operator) regression applied on a set of standard and advanced neuroimaging features in a nested cross-validation (CV) loop. This approach was aimed at 1) choosing the best predictive models, able to reliably predict the individual neuropsychological scores sensitive to attention and executive dysfunctions (prominent features of subcortical vascular cognitive impairment) and 2) identifying a features ranking according to their importance in the model through the assessment of the out-of-sample error. For each neuropsychological test, using 1000 repetitions of LASSO regression and 5000 random permutations, we found that the statistically significant models were those for the Montreal Cognitive Assessment scores (p-value = .039), Symbol Digit Modalities Test scores (p-value = .039), and Trail Making Test Part A scores (p-value = .025). Significant prediction of these scores was obtained using different sets of neuroimaging features in which the WM FD was the most frequently selected feature. In conclusion, we showed that a machine learning approach could be useful in SVD research field using standard and advanced neuroimaging features. Our study results raise the possibility that FD may represent a consistent feature in predicting cognitive decline in SVD that can complement standard imaging.

The human brain is a highly complex structure, which can be only partially described by conventional metrics derived from magnetic resonance imaging (MRI), such as volume, cortical thickness, and gyrification index. In the last years, the fractal dimension (FD) – a useful quantitative index of fractal geometry – has proven to well express the morphological complexity of the cerebral cortex. However, this complexity is likely higher than that we can observe using MRI scanners with 1.5 T or 3 T field strength. Ultrahigh-field MRI (UHF-MRI) improves imaging of smaller anatomical brain structures by exploring down to a submillimetric spatial resolution with higher signal-to-noise and contrast-to-noise ratios. Accordingly, we hypothesized that UHF-MRI might reveal a higher level of the structural complexity of the cerebral cortex. In this study, using an improved box-counting algorithm, we estimated the FD of the cerebral cortex in six public or private T1-weighted MRI datasets of young healthy subjects (for a total of 87 subjects), acquired at different field strengths (1.5 T, 3 T, and 7 T). Our results showed, for the first time, that MRI-derived FD values of the cerebral cortex imaged at 7 T were significantly higher than those observed at lower field strengths. UHF-MRI provides an anatomical definition not achievable at lower field strengths and can improve unveiling the real structural complexity of the human brain.

To assess the potential of histogram metrics of diffusion-tensor imaging (DTI)-derived indices in revealing neurodegeneration and its progression in spinocerebellar ataxia type 2 (SCA2). Nine SCA2 patients and 16 age-matched healthy controls, were examined twice (SCA2 patients 3.6±0.7 years and controls 3.3±1.0 years apart) on the same 1.5T scanner by acquiring T₁-weighted and diffusion-weighted (b-value = 1000 s/mm²) images. Cerebrum and brainstem-cerebellum regions were segmented using FreeSurfer suite. Histogram analysis of DTI-derived indices, including mean diffusivity (MD), fractional anisotropy (FA), axial (AD) / radial (RD) diffusivity and mode of anisotropy (MO), was performed. At baseline, significant differences between SCA2 patients and controls were confined to brainstem-cerebellum. Median values of MD/AD/RD and FA/MO were significantly (p<0.001) higher and lower, respectively, in SCA2 patients (1.11/1.30/1.03×10⁻³ mm²/s and 0.14/0.19) than in controls (0.80/1.00/0.70×10⁻³ mm²/s and 0.20/0.41). Also, peak location values of MD/AD/RD and FA were significantly (p<0.001) higher and lower, respectively, in SCA2 patients (0.91/1.11/0.81×10⁻³ mm²/s and 0.12) than in controls (0.71/0.91/0.63×10⁻³ mm²/s and 0.18). Peak height values of FA and MD/AD/RD/MO were significantly (p<0.001) higher and lower, respectively, in SCA2 patients (0.20 and 0.07/0.06/0.07×10⁻³ mm²/s/year /0.07) than in controls (0.15 and 0.14/0.11/0.12/×10⁻³ mm²/s/year /0.09). The rate of change of MD median values was significantly (p<0.001) higher (i.e., increased) in SCA2 patients (0.010×10⁻³ mm²/s/year) than in controls (-0.003×10⁻³ mm²/s/year) in the brainstem-cerebellum, whereas no significant difference was found for other indices and in the cerebrum. Histogram analysis of DTI-derived indices is a relatively straightforward approach which reveals microstructural changes associated with pontocerebellar degeneration in SCA2 and the median value of MD is capable to track its progression.

Many patients with multiple sclerosis (MS) experience information processing speed (IPS) deficits, and the Symbol Digit Modalities Test (SDMT) has been recommended as a valid screening test. Magnetic resonance imaging (MRI) has markedly improved the understanding of the mechanisms associated with cognitive deficits in MS. However, which structural MRI markers are the most closely related to cognitive performance is still unclear. We used the multicenter 3T‐MRI data set of the Italian Neuroimaging Network Initiative to extract multimodal data (i.e., demographic, clinical, neuropsychological, and structural MRIs) of 540 MS patients. We aimed to assess, through machine learning techniques, the contribution of brain MRI structural volumes in the prediction of IPS deficits when combined with demographic and clinical features. We trained and tested the eXtreme Gradient Boosting (XGBoost) model following a rigorous validation scheme to obtain reliable generalization performance. We carried out a classification and a regression task based on SDMT scores feeding each model with different combinations of features. For the classification task, the model trained with thalamus, cortical gray matter, hippocampus, and lesions volumes achieved an area under the receiver operating characteristic curve of 0.74. For the regression task, the model trained with cortical gray matter and thalamus volumes, EDSS, nucleus accumbens, lesions, and putamen volumes, and age reached a mean absolute error of 0.95. In conclusion, our results confirmed that damage to cortical gray matter and relevant deep and archaic gray matter structures, such as the thalamus and hippocampus, is among the most relevant predictors of cognitive performance in MS.