Dataset Information

Deep learning features from diffusion tensor imaging improve glioma stratification and identify risk groups with distinct molecular pathway activities.

ABSTRACT:

Background

To develop and validate a deep learning signature (DLS) from diffusion tensor imaging (DTI) for predicting overall survival in patients with infiltrative gliomas, and to investigate the biological pathways underlying the developed DLS.

Methods

The DLS was developed based on a deep learning cohort (n = 688). The key pathways underlying the DLS were identified on a radiogenomics cohort with paired DTI and RNA-seq data (n=78), where the prognostic value of the pathway genes was validated in public databases (TCGA, n = 663; CGGA, n = 657).

Findings

The DLS was associated with survival (log-rank P < 0.001) and was an independent predictor (P < 0.001). Incorporating the DLS into existing risk system resulted in a deep learning nomogram predicting survival better than either the DLS or the clinicomolecular nomogram alone, with a better calibration and classification accuracy (net reclassification improvement 0.646, P < 0.001). Five kinds of pathways (synaptic transmission, calcium signaling, glutamate secretion, axon guidance, and glioma pathways) were significantly correlated with the DLS. Average expression value of pathway genes showed prognostic significance in our radiogenomics cohort and TCGA/CGGA cohorts (log-rank P < 0.05).

Interpretation

DTI-derived DLS can improve glioma stratification by identifying risk groups with dysregulated biological pathways that contributed to survival outcomes. Therapies inhibiting neuron-to-brain tumor synaptic communication may be more effective in high-risk glioma defined by DTI-derived DLS.

Funding

A full list of funding bodies that contributed to this study can be found in the Acknowledgements section.

SUBMITTER: Yan J

PROVIDER: S-EPMC8479635 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Diffusion tensor magnetic resonance imaging (DTI) is unsurpassed in its ability to map tissue microstructure and structural connectivity in the living human brain. Nonetheless, the angular sampling requirement for DTI leads to long scan times and poses a critical barrier to performing high-quality DTI in routine clinical practice and large-scale research studies. In this work we present a new processing framework for DTI entitled DeepDTI that minimizes the data requirement of DTI to six diffusion-weighted images (DWIs) required by conventional voxel-wise fitting methods for deriving the six unique unknowns in a diffusion tensor using data-driven supervised deep learning. DeepDTI maps the input non-diffusion-weighted (b = 0) image and six DWI volumes sampled along optimized diffusion-encoding directions, along with T1-weighted and T2-weighted image volumes, to the residuals between the input and high-quality output b = 0 image and DWI volumes using a 10-layer three-dimensional convolutional neural network (CNN). The inputs and outputs of DeepDTI are uniquely formulated, which not only enables residual learning to boost CNN performance but also enables tensor fitting of resultant high-quality DWIs to generate orientational DTI metrics for tractography. The very deep CNN used by DeepDTI leverages the redundancy in local and non-local spatial information and across diffusion-encoding directions and image contrasts in the data. The performance of DeepDTI was systematically quantified in terms of the quality of the output images, DTI metrics, DTI-based tractography and tract-specific analysis results. We demonstrate rotationally-invariant and robust estimation of DTI metrics from DeepDTI that are comparable to those obtained with two b = 0 images and 21 DWIs for the primary eigenvector derived from DTI and two b = 0 images and 26-30 DWIs for various scalar metrics derived from DTI, achieving 3.3-4.6 × acceleration, and twice as good as those of a state-of-the-art denoising algorithm at the group level. The twenty major white-matter tracts can be accurately identified from the tractography of DeepDTI results. The mean distance between the core of the major white-matter tracts identified from DeepDTI results and those from the ground-truth results using 18 b = 0 images and 90 DWIs measures around 1-1.5 mm. DeepDTI leverages domain knowledge of diffusion MRI physics and power of deep learning to render DTI, DTI-based tractography, major white-matter tracts identification and tract-specific analysis more feasible for a wider range of neuroscientific and clinical studies.

Project description:Musculoskeletal modelling is an important platform on which to study the biomechanics of morphological structures in vertebrates and is widely used in clinical, zoological and palaeontological fields. The popularity of this approach stems from the potential to non-invasively quantify biologically important but difficult-to-measure functional parameters. However, while it is known that model predictions are highly sensitive to input values, it is standard practice to build models by combining musculoskeletal data from different sources resulting in 'generic' models for a given species. At present, there are little quantitative data on how merging disparate anatomical data in models impacts the accuracy of these functional predictions. This issue is addressed herein by quantifying the accuracy of both subject-specific human limb models containing individualised muscle force-generating properties and models built using generic properties from both elderly and young individuals, relative to experimental muscle torques obtained from an isokinetic dynamometer. The results show that subject-specific models predict isokinetic muscle torques to a greater degree of accuracy than generic models at the ankle (root-mean-squared error - 7.9% vs. 49.3% in elderly anatomy-based models), knee (13.2% vs. 57.3%) and hip (21.9% vs. 32.8%). These results have important implications for the choice of musculoskeletal properties in future modelling studies, and the relatively high level of accuracy achieved in the subject-specific models suggests that such models can potentially address questions about inter-subject variations of muscle functions. However, despite relatively high levels of overall accuracy, models built using averaged generic muscle architecture data from young, healthy individuals may lack the resolution and accuracy required to study such differences between individuals, at least in certain circumstances. The results do not wholly discourage the continued use of averaged generic data in musculoskeletal modelling studies but do emphasise the need for to maximise the accuracy of input values if studying intra-species form-function relationships in the musculoskeletal system.