Predicting resting-state brain functional connectivity from the structural connectome using the heat diffusion model: a multiple-timescale fusion method.

Objective.Understanding the intricate relationship between structural connectivity (SC) and functional connectivity (FC) is pivotal for understanding the complexities of the human brain. To explore this relationship, the heat diffusion model (HDM) was utilized to predict FC from SC. However, previous studies using the HDM have typically predicted FC at a critical time scale in the heat kernel equation, overlooking the dynamic nature of the diffusion process and providing an incomplete representation of the predicted FC.Approach.In this study, we propose an alternative approach based on the HDM. First, we introduced a multiple-timescale fusion method to capture the dynamic features of the diffusion process. Additionally, to enhance the smoothness of the predicted FC values, we employed the Wavelet reconstruction method to maintain local consistency and remove noise. Moreover, to provide a more accurate representation of the relationship between SC and FC, we calculated the linear transformation between the smoothed FC and the empirical FC.Main results.We conducted extensive experiments in two independent datasets. By fusing different time scales in the diffusion process for predicting FC, the proposed method demonstrated higher predictive correlation compared with method considering only critical time points (Singlescale). Furthermore, compared with other existing methods, the proposed method achieved the highest predictive correlations of 0.6939±0.0079 and 0.7302±0.0117 on the two datasets respectively. We observed that the visual network at the network level and the parietal lobe at the lobe level exhibited the highest predictive correlations, indicating that the functional activity in these regions may be closely related to the direct diffusion of information between brain regions.Significance.The multiple-timescale fusion method proposed in this study provides insights into the dynamic aspects of the diffusion process, contributing to a deeper understanding of how brain structure gives rise to brain function.

A large normative connectome for exploring the tractographic correlates of focal brain interventions.

Diffusion-weighted MRI (dMRI) is a widely used neuroimaging modality that permits the in vivo exploration of white matter connections in the human brain. Normative structural connectomics - the application of large-scale, group-derived dMRI datasets to out-of-sample cohorts - have increasingly been leveraged to study the network correlates of focal brain interventions, insults, and other regions-of-interest (ROIs). Here, we provide a normative, whole-brain connectome in MNI space that enables researchers to interrogate fiber streamlines that are likely perturbed by given ROIs, even in the absence of subject-specific dMRI data. Assembled from multi-shell dMRI data of 985 healthy Human Connectome Project subjects using generalized Q-sampling imaging and multispectral normalization techniques, this connectome comprises ~12 million unique streamlines, the largest to date. It has already been utilized in at least 18 peer-reviewed publications, most frequently in the context of neuromodulatory interventions like deep brain stimulation and focused ultrasound. Now publicly available, this connectome will constitute a useful tool for understanding the wider impact of focal brain perturbations on white matter architecture going forward.

Noninvasive- and invasive mapping reveals similar language network centralities - A function-based connectome analysis.

BACKGROUND: Former comparisons between direct cortical stimulation (DCS) and navigated transcranial magnetic stimulation (nTMS) only focused on cortical mapping. While both can be combined with diffusion tensor imaging, their differences in the visualization of subcortical and even network levels remain unclear. Network centrality is an essential parameter in network analysis to measure the importance of nodes identified by mapping. Those include Degree centrality, Eigenvector centrality, Closeness centrality, Betweenness centrality, and PageRank centrality. While DCS and nTMS have repeatedly been compared on the cortical level, the underlying network identified by both has not been investigated yet. METHOD: 27 patients with brain lesions necessitating preoperative nTMS and intraoperative DCS language mapping during awake craniotomy were enrolled. Function-based connectome analysis was performed based on the cortical nodes obtained through the two mapping methods, and language-related network centralities were compared. RESULTS: Compared with DCS language mapping, the positive predictive value of cortical nTMS language mapping is 74.1%, with good consistency of tractography for the arcuate fascicle and superior longitudinal fascicle. Moreover, network centralities did not differ between the two mapping methods. However, ventral stream tracts can be better traced based on nTMS mappings, demonstrating its strengths in acquiring language-related networks. In addition, it showed lower centralities than other brain areas, with decentralization as an indicator of language function loss. CONCLUSION: This study deepens the understanding of language-related functional anatomy and proves that non-invasive mapping-based network analysis is comparable to the language network identified via invasive cortical mapping.

The functional connectivity of the human claustrum, according to the Human Connectome Project database.

The claustrum is an irregular and fine sheet of grey matter in the basolateral telencephalon present in almost all mammals. The claustrum has been the object of several studies using animal models and, more recently, in human beings using neuroimaging. One of the most extended cognitive processes attributed to the claustrum is the salience process, which is also related to the insular cortex. In the same way, studies with human subjects and functional magnetic resonance imaging have reported the coactivation of the claustrum/insular cortex in the integration of sensory signals. This coactivation has been reported in the left claustrum/insular cortex or in the right claustrum/insular cortex. The asymmetry has been reported in task studies and literature related to neurological disorders such as Alzheimer's disease and schizophrenia, relating the severity of delusions with the reduction in left claustral volume. We present a functional connectivity study of the claustrum. Resting-state functional and anatomical MRI data from 100 healthy subjects were analyzed; taken from the Human Connectome Project (HCP, NIH Blueprint: The Human Connectome Project), with 2x2x2 mm3 voxel resolution. We hypothesize that 1) the claustrum is a node involved in different brain networks, 2) the functional connectivity pattern of the claustrum is different from the insular cortex's pattern, and 3) the asymmetry is present in the claustrum's functional connectivity. Our findings include at least three brain networks related to the claustrum. We found functional connectivity between the claustrum, frontoparietal network, and the default mode network as a distinctive attribute. The functional connectivity between the right claustrum with the frontoparietal network and the dorsal attention network supports the hypothesis of claustral asymmetry. These findings provide functional evidence, suggesting that the claustrum is coupled with the frontoparietal network serving together to instantiate new task states by flexibly modulating and interacting with other control and processing networks.

Higher-order functional brain networks and anterior cingulate glutamate + glutamine (Glx) in antipsychotic-naïve first episode psychosis patients.

Human connectome studies have provided abundant data consistent with the hypothesis that functional dysconnectivity is predominant in psychosis spectrum disorders. Converging lines of evidence also suggest an interaction between dorsal anterior cingulate cortex (dACC) cortical glutamate with higher-order functional brain networks (FC) such as the default mode (DMN), dorsal attention (DAN), and executive control networks (ECN) in healthy controls (HC) and this mechanism may be impaired in psychosis. Data from 70 antipsychotic-medication naïve first-episode psychosis (FEP) and 52 HC were analyzed. 3T Proton magnetic resonance spectroscopy (1H-MRS) data were acquired from a voxel in the dACC and assessed correlations (positive FC) and anticorrelations (negative FC) of the DMN, DAN, and ECN. We then performed regressions to assess associations between glutamate + glutamine (Glx) with positive and negative FC of these same networks and compared them between groups. We found alterations in positive and negative FC in all networks (HC > FEP). A relationship between dACC Glx and positive and negative FC was found in both groups, but when comparing these relationships between groups, we found contrasting associations between these variables in FEP patients compared to HC. We demonstrated that both positive and negative FC in three higher-order resting state networks are already altered in antipsychotic-naïve FEP, underscoring the importance of also considering anticorrelations for optimal characterization of large-scale functional brain networks as these represent biological processes as well. Our data also adds to the growing body of evidence supporting the role of dACC cortical Glx as a mechanism underlying alterations in functional brain network connectivity. Overall, the implications for these findings are imperative as this particular mechanism may differ in untreated or chronic psychotic patients; therefore, understanding this mechanism prior to treatment could better inform clinicians.Clinical trial registration: Trajectories of Treatment Response as Window into the Heterogeneity of Psychosis: A Longitudinal Multimodal Imaging Study, NCT03442101 . Glutamate, Brain Connectivity and Duration of Untreated Psychosis (DUP), NCT02034253 .

The non-decussating and decussating trigeminothalamic tracts in humans: A combination of connectome-based tractography and histological validation.

BACKGROUND: Functional anatomical research proposed the existence of a bilateral trigeminal ascending system although the anatomy trajectories of the trigeminothalamic connections cranial to the pons remain largely elusive. This study therefore aimed to clarify the anatomical distributions of the trigeminothalamic connections in humans. METHODS: Advanced deterministic tractography to an averaged template of diffusion tensor imaging data from 1065 subjects from the Human Connectome Project was used. Seedings masks were placed in Montreal Neurological Institute standard space by use of the BigBrain histological dataset. Waypoint masks of the sensory thalamus was obtained from the Brainnetome Atlas. RESULTS: Tractography results were validated by use of the BigBrain histological dataset and Polarized Light Imaging microscopy. The trigeminothalamic tract bifurcated into a decussating ventral and a non-decussating dorsal tract. The ventral and dorsal tracts ascended to the contralateral thalamus and ipsilateral thalamus and reflected the ventral trigeminothalamic tract and the dorsal trigeminothalamic tract, respectively. The projection of the ventral trigeminothalamic tract and the dorsal trigeminothalamic tract to both thalami confirm the existence of a bilateral trigeminothalamic system in humans. CONCLUSIONS: Because our study is strictly anatomical, no further conclusions can be drawn with regard to physiological functionality. Future research should explore if the dorsal trigeminothalamic tract and the ventral trigeminothalamic tract actually transmit signals from noxious stimuli, this offers potential in understanding and possibly treating neuropathology in the orofacial region.

Predictability of intelligence and age from structural connectomes.

In this study, structural images of 1048 healthy subjects from the Human Connectome Project Young Adult study and 94 from ADNI-3 study were processed by an in-house tractography pipeline and analyzed together with pre-processed data of the same subjects from braingraph.org. Whole brain structural connectome features were used to build a simple correlation-based regression machine learning model to predict intelligence and age of healthy subjects. Our results showed that different forms of intelligence as well as age are predictable to a certain degree from diffusion tensor imaging detecting anatomical fiber tracts in the living human brain. Though we did not identify significant differences in the prediction capability for the investigated features depending on the imaging feature extraction method, we did find that crystallized intelligence was consistently better predictable than fluid intelligence from structural connectivity data through all datasets. Our findings suggest a practical and scalable processing and analysis framework to explore broader research topics employing brain MR imaging.

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome.

Brain state classification by applying deep learning techniques on neuroimaging data has become a recent topic of research. However, unlike domains where the data is low dimensional or there are large number of available training samples, neuroimaging data is high dimensional and has few training samples. To tackle these issues, we present a sparse feedforward deep neural architecture for encoding and decoding the structural connectome of the human brain. We use a sparsely connected element-wise multiplication as the first hidden layer and a fixed transform layer as the output layer. The number of trainable parameters and the training time is significantly reduced compared to feedforward networks. We demonstrate superior performance of this architecture in encoding the structural connectome implicated in Alzheimer's disease (AD) and Parkinson's disease (PD) from DTI brain scans. For decoding, we propose recursive feature elimination (RFE) algorithm based on DeepLIFT, layer-wise relevance propagation (LRP), and Integrated Gradients (IG) algorithms to remove irrelevant features and thereby identify key biomarkers associated with AD and PD. We show that the proposed architecture reduces 45.1% and 47.1% of the trainable parameters compared to a feedforward DNN with an increase in accuracy by 2.6 % and 3.1% for cognitively normal (CN) vs AD and CN vs PD classification, respectively. We also show that the proposed RFE method leads to a further increase in accuracy by 2.1% and 4% for CN vs AD and CN vs PD classification, while removing approximately 90% to 95% irrelevant features. Furthermore, we argue that the biomarkers (i.e., key brain regions and connections) identified are consistent with previous literature. We show that relevancy score-based methods can yield high discriminative power and are suitable for brain decoding. We also show that the proposed approach led to a reduction in the number of trainable network parameters, an increase in classification accuracy, and a detection of brain connections and regions that were consistent with earlier studies.

Comprehensive analysis of synthetic learning applied to neonatal brain MRI segmentation.

Brain segmentation from neonatal MRI images is a very challenging task due to large changes in the shape of cerebral structures and variations in signal intensities reflecting the gestational process. In this context, there is a clear need for segmentation techniques that are robust to variations in image contrast and to the spatial configuration of anatomical structures. In this work, we evaluate the potential of synthetic learning, a contrast-independent model trained using synthetic images generated from the ground truth labels of very few subjects. We base our experiments on the dataset released by the developmental Human Connectome Project, for which high-quality images are available for more than 700 babies aged between 26 and 45 weeks postconception. First, we confirm the impressive performance of a standard UNet trained on a few volumes, but also confirm that such models learn intensity-related features specific to the training domain. We then confirm the robustness of the synthetic learning approach to variations in image contrast. However, we observe a clear influence of the age of the baby on the predictions. We improve the performance of this model by enriching the synthetic training set with realistic motion artifacts and over-segmentation of the white matter. Based on extensive visual assessment, we argue that the better performance of the model trained on real T2w data may be due to systematic errors in the ground truth. We propose an original experiment allowing us to show that learning from real data will reproduce any systematic bias affecting the training set, while synthetic models can avoid this limitation. Overall, our experiments confirm that synthetic learning is an effective solution for segmenting neonatal brain MRI. Our adapted synthetic learning approach combines key features that will be instrumental for large multisite studies and clinical applications.

Impaired topology and connectivity of grey matter structural networks in major depressive disorder: evidence from a multi-site neuroimaging data-set.

BACKGROUND: Major depressive disorder (MDD) has been increasingly understood as a disruption of brain connectome. Investigating grey matter structural networks with a large sample size can provide valuable insights into the structural basis of network-level neuropathological underpinnings of MDD. AIMS: Using a multisite MRI data-set including nearly 2000 individuals, this study aimed to identify robust topology and connectivity abnormalities of grey matter structural network linked to MDD and relevant clinical phenotypes. METHOD: A total of 955 MDD patients and 1009 healthy controls were included from 23 sites. Individualised structural covariance networks (SCN) were established based on grey matter volume maps. Following data harmonisation, network topological metrics and focal connectivity were examined for group-level comparisons, individual-level classification performance and association with clinical ratings. Various validation strategies were applied to confirm the reliability of findings. RESULTS: Compared with healthy controls, MDD individuals exhibited increased global efficiency, abnormal regional centralities (i.e. thalamus, precentral gyrus, middle cingulate cortex and default mode network) and altered circuit connectivity (i.e. ventral attention network and frontoparietal network). First-episode drug-naive and recurrent patients exhibited different patterns of deficits in network topology and connectivity. In addition, the individual-level classification of topological metrics outperforms that of structural connectivity. The thalamus-insula connectivity was positively associated with the severity of depressive symptoms. CONCLUSIONS: Based on this high-powered data-set, we identified reliable patterns of impaired topology and connectivity of individualised SCN in MDD and relevant subtypes, which adds to the current understanding of neuropathology of MDD and might guide future development of diagnostic and therapeutic markers.

Diverse and asymmetric patterns of single-neuron projectome in regulating interhemispheric connectivity.

The corpus callosum, historically considered primarily for homotopic connections, supports many heterotopic connections, indicating complex interhemispheric connectivity. Understanding this complexity is crucial yet challenging due to diverse cell-specific wiring patterns. Here, we utilized public AAV bulk tracing and single-neuron tracing data to delineate the anatomical connection patterns of mouse brains and conducted wide-field calcium imaging to assess functional connectivity across various brain states in male mice. The single-neuron data uncovered complex and dense interconnected patterns, particularly for interhemispheric-heterotopic connections. We proposed a metric "heterogeneity" to quantify the complexity of the connection patterns. Computational modeling of these patterns suggested that the heterogeneity of upstream projections impacted downstream homotopic functional connectivity. Furthermore, higher heterogeneity observed in interhemispheric-heterotopic projections would cause lower strength but higher stability in functional connectivity than their intrahemispheric counterparts. These findings were corroborated by our wide-field functional imaging data, underscoring the important role of heterotopic-projection heterogeneity in interhemispheric communication.

Fibration symmetries and cluster synchronization in the Caenorhabditis elegans connectome.

Capturing how the Caenorhabditis elegans connectome structure gives rise to its neuron functionality remains unclear. It is through fiber symmetries found in its neuronal connectivity that synchronization of a group of neurons can be determined. To understand these we investigate graph symmetries and search for such in the symmetrized versions of the forward and backward locomotive sub-networks of the Caenorhabditi elegans worm neuron network. The use of ordinarily differential equations simulations admissible to these graphs are used to validate the predictions of these fiber symmetries and are compared to the more restrictive orbit symmetries. Additionally fibration symmetries are used to decompose these graphs into their fundamental building blocks which reveal units formed by nested loops or multilayered fibers. It is found that fiber symmetries of the connectome can accurately predict neuronal synchronization even under not idealized connectivity as long as the dynamics are within stable regimes of simulations.

A multi-demand operating system underlying diverse cognitive tasks.

The existence of a multiple-demand cortical system with an adaptive, domain-general, role in cognition has been proposed, but the underlying dynamic mechanisms and their links to cognitive control abilities are poorly understood. Here we use a probabilistic generative Bayesian model of brain circuit dynamics to determine dynamic brain states across multiple cognitive domains, independent datasets, and participant groups, including task fMRI data from Human Connectome Project, Dual Mechanisms of Cognitive Control study and a neurodevelopment study. We discover a shared brain state across seven distinct cognitive tasks and found that the dynamics of this shared brain state predicted cognitive control abilities in each task. Our findings reveal the flexible engagement of dynamic brain processes across multiple cognitive domains and participant groups, and uncover the generative mechanisms underlying the functioning of a domain-general cognitive operating system. Our computational framework opens promising avenues for probing neurocognitive function and dysfunction.

Decomposing cortical activity through neuronal tracing connectome-eigenmodes in marmosets.

Deciphering the complex relationship between neuroanatomical connections and functional activity in primate brains remains a daunting task, especially regarding the influence of monosynaptic connectivity on cortical activity. Here, we investigate the anatomical-functional relationship and decompose the neuronal-tracing connectome of marmoset brains into a series of eigenmodes using graph signal processing. These cellular connectome eigenmodes effectively constrain the cortical activity derived from resting-state functional MRI, and uncover a patterned cellular-functional decoupling. This pattern reveals a spatial gradient from coupled dorsal-posterior to decoupled ventral-anterior cortices, and recapitulates micro-structural profiles and macro-scale hierarchical cortical organization. Notably, these marmoset-derived eigenmodes may facilitate the inference of spontaneous cortical activity and functional connectivity of homologous areas in humans, highlighting the potential generalizing of the connectomic constraints across species. Collectively, our findings illuminate how neuronal-tracing connectome eigenmodes constrain cortical activity and improve our understanding of the brain's anatomical-functional relationship.

Modeling the neurocognitive dynamics of language across the lifespan.

Healthy aging is associated with a heterogeneous decline across cognitive functions, typically observed between language comprehension and language production (LP). Examining resting-state fMRI and neuropsychological data from 628 healthy adults (age 18-88) from the CamCAN cohort, we performed state-of-the-art graph theoretical analysis to uncover the neural mechanisms underlying this variability. At the cognitive level, our findings suggest that LP is not an isolated function but is modulated throughout the lifespan by the extent of inter-cognitive synergy between semantic and domain-general processes. At the cerebral level, we show that default mode network (DMN) suppression coupled with fronto-parietal network (FPN) integration is the way for the brain to compensate for the effects of dedifferentiation at a minimal cost, efficiently mitigating the age-related decline in LP. Relatedly, reduced DMN suppression in midlife could compromise the ability to manage the cost of FPN integration. This may prompt older adults to adopt a more cost-efficient compensatory strategy that maintains global homeostasis at the expense of LP performances. Taken together, we propose that midlife represents a critical neurocognitive juncture that signifies the onset of LP decline, as older adults gradually lose control over semantic representations. We summarize our findings in a novel synergistic, economical, nonlinear, emergent, cognitive aging model, integrating connectomic and cognitive dimensions within a complex system perspective.

Community detection in the human connectome: Method types, differences and their impact on inference.

Community structure is a fundamental topological characteristic of optimally organized brain networks. Currently, there is no clear standard or systematic approach for selecting the most appropriate community detection method. Furthermore, the impact of method choice on the accuracy and robustness of estimated communities (and network modularity), as well as method-dependent relationships between network communities and cognitive and other individual measures, are not well understood. This study analyzed large datasets of real brain networks (estimated from resting-state fMRI from n $$ n $$ = 5251 pre/early adolescents in the adolescent brain cognitive development [ABCD] study), and n $$ n $$ = 5338 synthetic networks with heterogeneous, data-inspired topologies, with the goal to investigate and compare three classes of community detection methods: (i) modularity maximization-based (Newman and Louvain), (ii) probabilistic (Bayesian inference within the framework of stochastic block modeling (SBM)), and (iii) geometric (based on graph Ricci flow). Extensive comparisons between methods and their individual accuracy (relative to the ground truth in synthetic networks), and reliability (when applied to multiple fMRI runs from the same brains) suggest that the underlying brain network topology plays a critical role in the accuracy, reliability and agreement of community detection methods. Consistent method (dis)similarities, and their correlations with topological properties, were estimated across fMRI runs. Based on synthetic graphs, most methods performed similarly and had comparable high accuracy only in some topological regimes, specifically those corresponding to developed connectomes with at least quasi-optimal community organization. In contrast, in densely and/or weakly connected networks with difficult to detect communities, the methods yielded highly dissimilar results, with Bayesian inference within SBM having significantly higher accuracy compared to all others. Associations between method-specific modularity and demographic, anthropometric, physiological and cognitive parameters showed mostly method invariance but some method dependence as well. Although method sensitivity to different levels of community structure may in part explain method-dependent associations between modularity estimates and parameters of interest, method dependence also highlights potential issues of reliability and reproducibility. These findings suggest that a probabilistic approach, such as Bayesian inference in the framework of SBM, may provide consistently reliable estimates of community structure across network topologies. In addition, to maximize robustness of biological inferences, identified network communities and their cognitive, behavioral and other correlates should be confirmed with multiple reliable detection methods.

A revision of the dorsal origin of the frontal aslant tract (FAT) in the superior frontal gyrus: a DWI-tractographic study.

The frontal aslant tract (FAT) is a white matter tract connecting the superior frontal gyrus (SFG) to the inferior frontal gyrus (IFG). Its dorsal origin is identified in humans in the medial wall of the SFG, in the supplementary motor complex (SM-complex). However, empirical observation shows that many FAT fibres appear to originate from the dorsal, rather than medial, portion of the SFG. We quantitatively investigated the actual origin of FAT fibres in the SFG, specifically discriminating between terminations in the medial wall and in the convexity of the SFG. We analysed data from 105 subjects obtained from the Human Connectome Project (HCP) database. We parcelled the cortex of the IFG, dorsal SFG and medial SFG in several regions of interest (ROIs) ordered in a caudal-rostral direction, which served as seed locations for the generation of streamlines. Diffusion imaging data (DWI) was processed using a multi-shell multi-tissue CSD-based algorithm. Results showed that the number of streamlines originating from the dorsal wall of the SFG significantly exceeds those from the medial wall of the SFG. Connectivity patterns between ROIs indicated that FAT sub-bundles are segregated in parallel circuits ordered in a caudal-rostral direction. Such high degree of coherence in the streamline trajectory allows to establish pairs of homologous cortical parcels in the SFG and IFG. We conclude that the frontal origin of the FAT is found in both dorsal and medial surfaces of the superior frontal gyrus.

MaPPeRTrac: A Massively Parallel, Portable, and Reproducible Tractography Pipeline.

Large-scale diffusion MRI tractography remains a significant challenge. Users must orchestrate a complex sequence of instructions that requires many software packages with complex dependencies and high computational costs. We developed MaPPeRTrac, an edge-centric tractography pipeline that simplifies and accelerates this process in a wide range of high-performance computing (HPC) environments. It fully automates either probabilistic or deterministic tractography, starting from a subject's magnetic resonance imaging (MRI) data, including structural and diffusion MRI images, to the edge density image (EDI) of their structural connectomes. Dependencies are containerized with Singularity (now called Apptainer) and decoupled from code to enable rapid prototyping and modification. Data derivatives are organized with the Brain Imaging Data Structure (BIDS) to ensure that they are findable, accessible, interoperable, and reusable following FAIR principles. The pipeline takes full advantage of HPC resources using the Parsl parallel programming framework, resulting in the creation of connectome datasets of unprecedented size. MaPPeRTrac is publicly available and tested on commercial and scientific hardware, so it can accelerate brain connectome research for a broader user community. MaPPeRTrac is available at: https://github.com/LLNL/mappertrac .

Transformer based multi-modal MRI fusion for prediction of post-menstrual age and neonatal brain development analysis.

The brain development during the perinatal period is characterized by rapid changes in both structure and function, which have significant impact on the cognitive and behavioral abilities later in life. Accurate assessment of brain age is a crucial indicator for brain development maturity and can help predict the risk of neonatal pathology. However, evaluating neonatal brains using magnetic resonance imaging (MRI) is challenging due to its complexity, multi-dimension, and noise with subtle alterations. In this paper, we propose a multi-modal deep learning framework based on transformers for precise post-menstrual age (PMA) estimation and brain development analysis using T2-weighted structural MRI (T2-sMRI) and diffusion MRI (dMRI) data. First, we build a two-stream dense network to learn modality-specific features from T2-sMRI and dMRI of brain individually. Then, a transformer module based on self-attention mechanism integrates these features for PMA prediction and preterm/term classification. Finally, saliency maps on brain templates are used to enhance the interpretability of results. Our method is evaluated on the multi-modal MRI dataset of the developing Human Connectome Project (dHCP), which contains 592 neonates, including 478 term-born and 114 preterm-born subjects. The results demonstrate that our method achieves a 0.5-week mean absolute error (MAE) in PMA estimation for term-born subjects. Notably, preterm-born subjects exhibit delayed brain development, worsening with increasing prematurity. Our method also achieves 95% accuracy in classification of term-born and preterm-born subjects, revealing significant group differences.