Biology subjects

Soltaninejad, M.

Publications and source records attributed to Soltaninejad, M..

5 recordsLinked to original sources

bioRxiv2026

Machine-learning MRI stratification of genetic frontotemporal dementia for clinical trial enrichment

BACKGROUNDGenetic frontotemporal dementia (FTD) shows large differences in symptom profiles, brain atrophy patterns, and progression rate, making clinical trials difficult to design and power. There is a need for biomarkers that can model disease progression, identify biologically distinct groups, and support efficient trial enrichment. METHODSWe applied contrastive trajectory inference (cTI), a machine-learning method, to structural MRI, white matter hyperintensity, and demographic data from 736 participants in the GENFI cohort, including non-carriers and carriers of C9orf72, GRN, or MAPT mutations. cTI produced an individual "genetic FTD progression score" (0-1) and grouped mutation carriers into data-driven subtypes. We tested construct validity using correlations between progression score and cognitive/functional measures, examined subtype differences in brain-behavior coupling, plasma neurofilament light (NfL), and longitudinal decline, and compared cTI-based trial enrichment against age, cortical thickness and NfL using analytic and simulation-based power analyses. RESULTSGenetic FTD progression scores correlated strongly with global dementia severity and multiple cognitive domains (all p < 0.001), confirming robust clinical scoring. Two mutation-carrier subtypes emerged: a Progressive Track (Subtype 2) with strong associations between progression score and cognitive/functional impairment, rising NfL, and faster longitudinal decline; and a Dissociated Track (Subtype 3) with comparable levels of structural variation but weak or absent clinical and NfL changes, suggesting relative biological stability. Baseline subtype membership added prognostic value for future decline in processing speed and language beyond baseline severity. Notably, for C9orf72 and GRN, cTI-informed enrichment reduced required recruited sample size per arm by about 61-75% compared with unenriched designs, and outperformed enrichment using age, cortical thickness or NfL in both analytic and simulation-based power analyses. CONCLUSIONSMachine-learning stratification of genetic FTD reveals a progressive and a dissociated disease track and provides individualized progression scores that closely track clinical status. cTI progression scores offer a powerful tool for trial enrichment, enabling smaller, more efficient prevention and early-intervention trials than conventional MRI or NfL markers alone.

neuroscience↗

bioRxiv2025

MRI-based classifier to identify close-to-onset cases in C9orf72 genetic frontotemporal dementia

Predicting symptom onset in genetic frontotemporal dementia (FTD) is crucial for advancing targeted interventions and clinical trial design. Brain changes begin years before clinical symptoms emerge, making neuroimaging a strong candidate for onset prediction. However, FTD is highly heterogeneous, encompassing diverse molecular pathologies, affected brain networks, and symptom trajectories. This variability limits the predictive power of any single imaging biomarker and underscores the need for an integrative, multimodal approach to improve prediction accuracy and generalizability. We used machine learning to integrate diverse neuroimaging features, identifying a robust signature for risk stratification. We analyzed T1-weighted and T2-weighted MRI scans from 71 symptomatic C9orf72 carriers, 90 presymptomatic carriers, and 69 healthy controls from the GENFI cohort. We used FreeSurfer to measure cortical thickness and subcortical volumes, and BISON to quantify white matter hyperintensities (WMH). We applied Principal Component Analysis for dimensionality reduction and trained a random forest classifier to distinguish symptomatic carriers from controls. The model was subsequently applied to the presymptomatic cohort to identify individuals whose brain patterns resembled those of symptomatic cases, under the hypothesis that greater similarity indicated a higher risk of conversion. We validated the model with neuropsychological data and a two-year longitudinal follow-up. The classifier distinguished symptomatic C9orf72 carriers from controls with 87.0% accuracy. When applied to presymptomatic carriers, the model identified 21.1% of the cohort as having brain features comparable to those of symptomatic cases. This "high-risk group" showed significant neuropsychological weaknesses in executive function, language and social cognition compared to the non high-risk group. The model accurately predicted clinical conversion within a two-year period with 84.5% accuracy, a 70% sensitivity and a 93.3% negative predictive value. Our findings demonstrate the utility of a machine learning approach using multi-modal MRI to identify presymptomatic C9orf72 carriers at high risk of disease onset within the next two years. By capturing subtle neuroanatomical patterns associated with disease processes, this approach offers a promising method for stratifying genetic FTD carriers prior to symptom onset. Such predictive models could optimize patient selection in future clinical trials.

neuroscience↗

bioRxiv2025

White Matter Hyperintensities Precede other Biomarkers in GRN Frontotemporal Dementia

INTRODUCTIONIncreased white matter hyperintensities (WMHs) have been reported in genetic frontotemporal dementia (FTD) in small studies, but the sequence of WMH abnormalities relative to other biomarkers is unclear. METHODSUsing a large dataset (n=763 GENFI2 participants), we measured WMHs and examined them across genetic FTD variants and stages. Cortical and subcortical volumes were parcellated, and serum neurofilament light chain (NfL) levels were measured. Biomarker progression was assessed with discriminative event-based and regression modeling. RESULTSSymptomatic GRN carriers showed elevated WMHs, primarily in the frontal lobe, while no significant increase was observed in C9orf72 or MAPT carriers. WMH abnormalities preceded NfL elevation, ventricular enlargement, and cortical atrophy. Longitudinally, baseline WMHs predicted subcortical changes, while subcortical volumes did not predict WMH changes, suggesting WMHs may precede neurodegeneration. DISCUSSIONWMHs are elevated in a subset of GRN-related FTD. When present, they appear early and should be considered in disease progression models. HighlightsO_LIElevated WMH volumes in symptomatic GRN carriers, but not in other mutations. C_LIO_LIWMH accumulation is mostly observed in the frontal lobe. C_LIO_LIWMH abnormalities appear early in GRN-FTD, before NfL, atrophy, and ventriculomegaly. C_LIO_LILongitudinally, WMH volumes can predict subcortical changes, but not vice versa. C_LIO_LIWMHs are key early markers in GRN-FTD and should be included in progression models. C_LI RESEARCH IN CONTEXTO_ST_ABSSystematic reviewC_ST_ABSWe systematically reviewed the literature on white matter hyperintensities (WMHs) in frontotemporal dementia (FTD) using PubMed. While a few small studies reported increased WMHs in GRN mutation carriers, their sample sizes were limited, and they did not assess the timing of WMHs within disease progression or their temporal relationship to other biomarkers. InterpretationWe identified a sequence of key biomarkers in GRN-related FTD and demonstrated that WMHs are among the earliest biomarkers, preceding cortical and subcortical atrophy as well as blood biomarkers. This aligns with neuropathological evidence of early white matter involvement in FTLD-GRN. Additionally, using a larger dataset, we validated previous reports of elevated WMHs in GRN carriers, confirming their reliability. Future directionsFuture studies should integrate WMHs into FTD progression models to enhance early diagnosis. Understanding why only a subset of GRN carriers exhibit high WMH volumes remains a key research priority.

neuroscience↗

bioRxiv2024

Brain Age Prediction: Deep Models Need a Hand to Generalize

Predicting brain age from T1-weighted MRI is a promising marker for understanding brain aging and its associated conditions. While deep learning models have shown success in reducing the Mean Absolute Error (MAE) of predicted brain age, concerns about robust and accurate generalization in new data limit their clinical applicability. The large number of trainable parameters, combined with limited medical imaging training data, contribute to this challenge, often resulting in a generalization gap where there is a significant discrepancy between model performance on training data versus unseen data. In this study, we assess a deep model, SFCN-reg, based on the VGG-16 architecture, and address the generalization gap through comprehensive preprocessing, extensive data augmentation, and model regularization. Using training data from the UK Biobank, we demonstrate substantial improvements in model performance. Specifically, our approach reduces the generalization MAE by 44% (from 5.25 to 2.96 years) in the Alzheimers Disease Neuroimaging Initiative dataset and by 22% (from 4.35 to 3.40 years) in the Australian Imaging, Biomarker and Lifestyle dataset. Furthermore, we achieve a 29% reduction in scan-rescan error (from 0.86 to 0.61 years) while enhancing the models robustness to registration errors. Feature importance maps highlight anatomical regions used to predict age. These results highlight the critical role of high-quality preprocessing and robust training techniques in improving accuracy and narrowing the generalization gap, both necessary steps towards the clinical use of brain age prediction models. Our study makes valuable contributions to neuroimaging research by offering a potential pathway to improve the clinical applicability of deep learning models.

neuroscience↗

bioRxiv2019

Three Dimensional Root CT Segmentation using Multi-Resolution Encoder-Decoder Networks

We address the complex problem of reliably segmenting root structure from soil in X-ray Computed Tomography (CT) images. We utilise a deep learning approach, and propose a state-of-the-art multi-resolution architecture based on encoder-decoders. While previous work in encoder-decoders implies the use of multiple resolutions simply by downsampling and upsampling images, we make this process explicit, with branches of the network tasked separately with obtaining local high-resolution segmentation, and wider low-resolution contextual information. The complete network is a memory efficient implementation that is still able to resolve small root detail in large volumetric images. We evaluate our approach by comparing against a number of different encoder-decoder based architectures from the literature, as well as a popular existing image analysis tool designed for root CT segmentation. We show qualitatively and quantitatively that a multi-resolution approach offers substantial accuracy improvements over a both a small receptive field size in a deep network, or a larger receptive field in a shallower network. We obtain a Dice score of 0.59 compared with 0.41 for the closest competing method. We then further improve performance using an incremental learning approach, in which failures in the original network are used to generate harder negative training examples. Results of this process raise the precision of the network, and improve the Dice score to 0.66. Our proposed method requires no user interaction, is fully automatic, and identifies large and fine root material throughout the whole volume. The 3D segmented output of our method is well-connected, allowing the recovery of structured representations of root system architecture, and so may be successfully utilised in root phenotyping.

bioinformatics↗