Changes for page Methodology

Last modified by manuelmenendez on 2025/03/14 08:31

From 2.1 to 1.1

From version 28.1

edited by manuelmenendez
on 2025/03/14 08:31

Change comment: There is no comment for this version

To version 2.1

edited by manuelmenendez
on 2025/01/27 23:07

Change comment: There is no comment for this version

Raw
Rendered

Summary

Page properties (1 modified, 0 added, 0 removed)
Attachments (0 modified, 0 added, 1 removed)
- workflow neurodiagnoses.png

Details

Page properties

Content

@@ -1,216 +98,106 @@
--**Neurodiagnoses AI** is an open-source, AI-driven framework designed to enhance the diagnosis and prognosis of central nervous system (CNS) disorders. It encompasses a broader spectrum of neurological conditions. The system integrates multimodal data sources—including EEG, neuroimaging, biomarkers, and genetics—and employs machine learning models to deliver explainable, real-time diagnostic insights. A key feature of this framework is the incorporation of the **Generalized Neuro Biomarker Ontology Categorization (Neuromarker) **and** Disease Knowledge Transfer (DKT)**, which standardizes disease and biomarker classification across all CNS diseases, facilitating cross-disease AI training.
--
--**Neuromarker: Generalized Biomarker Ontology**
--
--Neuromarker extends the Common Alzheimer’s Disease Research Ontology (CADRO) into a comprehensive biomarker categorization framework applicable to all neurodegenerative diseases (NDDs). This ontology enables standardized classification, AI-based feature extraction, and seamless multimodal data integration.
--
--**Recommended Software**
--
--There is a suite of software that can help implement the workflow needed in Neurodiagnoses. Find a list of recommendations [[here>>https://github.com/Fundacion-de-Neurociencias/neurodiagnoses/blob/main/recommended_software]].
--
--**Core Biomarker Categories**
--
--Within the Neurodiagnoses AI framework, biomarkers are categorized as follows:
--
--|=**Category**|=**Description**
--|**Molecular Biomarkers**|Omics-based markers (genomic, transcriptomic, proteomic, metabolomic, lipidomic)
--|**Neuroimaging Biomarkers**|Structural (MRI, CT), Functional (fMRI, PET), Molecular Imaging (tau, amyloid, α-synuclein)
--|**Fluid Biomarkers**|CSF, plasma, blood-based markers for tau, amyloid, α-synuclein, TDP-43, GFAP, NfL, autoantiboides
--|**Neurophysiological Biomarkers**|EEG, MEG, evoked potentials (ERP), sleep-related markers
--|**Digital Biomarkers**|Gait analysis, cognitive/speech biomarkers, wearables data, EHR-based markers
--|**Clinical Phenotypic Markers**|Standardized clinical scores (MMSE, MoCA, CDR, UPDRS, ALSFRS, UHDRS)
--|**Genetic Biomarkers**|Risk alleles (APOE, LRRK2, MAPT, C9orf72, PRNP) and polygenic risk scores
--|**Environmental & Lifestyle Factors**|Toxins, infections, diet, microbiome, comorbidities
--
--**Integrating External Databases into Neurodiagnoses**
--
--To enhance diagnostic precision, Neurodiagnoses AI incorporates data from multiple biomedical and neurological research databases. Researchers can integrate external datasets by following these steps:
--
--1. (((
--**Register for Access**
--
--* Each external database requires individual registration and access approval.
--* Ensure compliance with ethical approvals and data usage agreements before integrating datasets into Neurodiagnoses.
--* Some repositories may require a Data Usage Agreement (DUA) for sensitive medical data.
--)))
--1. (((
--**Download & Prepare Data**
--
--* Download datasets while adhering to database usage policies.
--* (((
--Ensure files meet Neurodiagnoses format requirements:
--
--|=**Data Type**|=**Accepted Formats**
--|**Tabular Data**|.csv, .tsv
--|**Neuroimaging**|.nii, .dcm
--|**Genomic Data**|.fasta, .vcf
--|**Clinical Metadata**|.json, .xml
--)))
--* (((
--**Mandatory Fields for Integration**:
--
--* Subject ID: Unique patient identifier
--* Diagnosis: Standardized disease classification
--* Biomarkers: CSF, plasma, or imaging biomarkers
--* Genetic Data: Whole-genome or exome sequencing
--* Neuroimaging Metadata: MRI/PET acquisition parameters
--)))
--)))
--1. (((
--**Upload Data to Neurodiagnoses**
--
--* (((
--**Option 1: Upload to EBRAINS Bucket**
--
--* Location: EBRAINS Neurodiagnoses Bucket
--* Ensure correct metadata tagging before submission.
--)))
--* (((
--**Option 2: Contribute via GitHub Repository**
--
--* Location: GitHub Data Repository
--* Create a new folder under /data/ and include a dataset description.
--* For large datasets, contact project administrators before uploading.
--)))
--)))
--1. (((
--**Integrate Data into AI Models**
--
--* Open Jupyter Notebooks on EBRAINS to run preprocessing scripts.
--* Standardize neuroimaging and biomarker formats using harmonization tools.
--* Utilize machine learning models to handle missing data and feature extraction.
--* Train AI models with newly integrated patient cohorts.
--
--**Reference**: See docs/data_processing.md for detailed instructions.
--)))
--
--**AI-Driven Biomarker Categorization**
--
--Neurodiagnoses employs advanced AI models for biomarker classification:
--
--|=**Model Type**|=**Application**
--|**Graph Neural Networks (GNNs)**|Identify shared biomarker pathways across diseases
--|**Contrastive Learning**|Distinguish overlapping vs. unique biomarkers
--|**Multimodal Transformer Models**|Integrate imaging, omics, and clinical data
--
--=== **Jupyter Integration with EBRAINS** ===
--
  === **Overview** ===
--Neurodiagnoses Open Source leverages **Jupyter Notebooks from EBRAINS** to facilitate neurodiagnostic research, biomarker analysis, and AI-driven data processing. This integration provides an interactive and reproducible environment for developing machine learning models, analyzing neuroimaging data, and exploring multimodal biomarkers. Jupyter integration in EBRAINS empowers **Neurodiagnoses Open Source** to: ✅ **Analyze MRI, EEG, and biomarker data efficiently**. ✅ **Train machine learning models with high-performance computing**. ✅ **Ensure transparency with interactive explainability tools**. ✅ **Enable collaborative neurodiagnostic research with reproducible notebooks**.
++This section describes the step-by-step process used in the **Neurodiagnoses** project to develop a novel diagnostic framework for neurological diseases. The methodology integrates artificial intelligence (AI), biomedical ontologies, and computational neuroscience to create a structured, interpretable, and scalable diagnostic system.
--=== **Key Capabilities of Jupyter in Neurodiagnoses** ===
++----
--==== **1. Neuroimaging Analysis (MRI, fMRI, PET)** ====
++=== **1. Data Integration** ===
--* **Preprocessing Pipelines:**
--** Use **Nipype, NiLearn, ANTs, and FreeSurfer** for structural and functional MRI analysis.
--** Skull stripping, segmentation, and registration of MRI scans.
--** Entropy-based slice selection for training deep learning models.
--* **Deep Learning for Neuroimaging:**
--** Implement **CNN-based models (ResNet, VGG16, Autoencoders)** for biomarker extraction.
--** Feature-based classification of **Alzheimer’s, Parkinson’s, and MCI** from neuroimaging data.
++==== **Data Sources** ====
--==== **2. EEG and MEG Signal Processing** ====
++* **Biomedical Ontologies**:
++** Human Phenotype Ontology (HPO) for phenotypic abnormalities.
++** Gene Ontology (GO) for molecular and cellular processes.
++* **Neuroimaging Datasets**:
++** Example: Alzheimer’s Disease Neuroimaging Initiative (ADNI), OpenNeuro.
++* **Clinical and Biomarker Data**:
++** Anonymized clinical reports, molecular biomarkers, and test results.
--* **Data Preprocessing & Artifact Removal:**
--** Use **MNE-Python** for filtering, ICA-based artifact rejection, and time-series normalization.
--** Extract frequency and time-domain features from EEG/MEG signals.
--* **Feature Engineering & Connectivity Analysis:**
--** Functional connectivity analysis using **coherence and phase synchronization metrics**.
--** Graph-theory-based EEG biomarkers for neurodegenerative disease classification.
--* **Deep Learning for EEG Analysis:**
--** Train LSTMs and CNNs for automatic EEG-based classification of MCI and cognitive decline.
++==== **Data Preprocessing** ====
--==== **3. Machine Learning for Biomarker Discovery** ====
++1. **Standardization**: Ensure all data sources are normalized to a common format.
++1. **Feature Selection**: Identify relevant features for diagnosis (e.g., biomarkers, imaging scores).
++1. **Data Cleaning**: Handle missing values and remove duplicates.
--* **SHAP-based Explainability for Biomarkers:**
--** Use **Random Forest + SHAP** to rank the most predictive CSF, blood, and imaging biomarkers.
--** Generate SHAP summary plots to interpret the impact of individual biomarkers.
--* **Multi-Modal Feature Selection:**
--** Implement **Anchor-Graph Feature Selection** to combine MRI, EEG, and CSF data.
--** PCA and autoencoders for dimensionality reduction and feature extraction.
--* **Automated Risk Prediction Models:**
--** Train ensemble models combining **deep learning and classical ML algorithms**.
--** Apply **subject-level cross-validation** to prevent data leakage and ensure reproducibility.
++----
--==== **4. Computational Simulations & Virtual Brain Models** ====
++=== **2. AI-Based Analysis** ===
--* **Integration with The Virtual Brain (TVB):**
--** Simulate large-scale brain networks based on individual neuroimaging data.
--** Model the effect of neurodegenerative progression on brain activity.
--* **Cortical and Subcortical Connectivity Analysis:**
--** Generate connectivity matrices using diffusion MRI and functional MRI correlations.
--** Validate computational simulations with real patient data from EBRAINS datasets.
++==== **Model Development** ====
--==== **5. Interactive Data Visualization & Reporting** ====
++* **Embedding Models**: Use pre-trained models like BioBERT or BioLORD for text data.
++* **Classification Models**:
++** Algorithms: Random Forest, Support Vector Machines (SVM), or neural networks.
++** Purpose: Predict the likelihood of specific neurological conditions based on input data.
--* **Dynamic Plots & Dashboards:**
--** Use **Matplotlib, Seaborn, Plotly** for interactive visualizations of biomarkers.
--** Implement real-time MRI slice rendering and EEG signal visualization.
--* **Automated Report Generation:**
--** Generate **Jupyter-based PDF reports** summarizing key findings.
--** Export analysis results in JSON, CSV, and interactive web dashboards.
++==== **Dimensionality Reduction and Interpretability** ====
--=== **How to Use Neurodiagnoses with Jupyter in EBRAINS** ===
++* Leverage DEIBO (Data-driven Embedding Interpretation Based on Ontologies) to connect model dimensions to ontology concepts.
++* Evaluate interpretability using metrics like the Area Under the Interpretability Curve (AUIC).
--==== **1. Access EBRAINS Jupyter Environment** ====
++----
--1. Create an **EBRAINS account** at [[EBRAINS.eu>>url:https://ebrains.eu/]].
--1. Navigate to the **Collaboratory** and open the Jupyter Lab interface.
--1. Clone the Neurodiagnoses repository:
++=== **3. Diagnostic Framework** ===
--{{{git clone https://github.com/neurodiagnoses
--cd neurodiagnoses
--pip install -r requirements.txt
--}}}
++==== **Axes of Diagnosis** ====
--==== **2. Run Prebuilt Neurodiagnoses Notebooks** ====
++The framework organizes diagnostic data into three axes:
--1. Open the **notebooks/** directory inside Jupyter.
--1. Run any of the available notebooks:
--1*. mri_biomarker_analysis.ipynb → Extracts MRI-based biomarkers.
--1*. eeg_preprocessing.ipynb → Cleans and processes EEG signals.
--1*. shap_biomarker_explainability.ipynb → Visualizes biomarker importance.
--1*. disease_risk_prediction.ipynb → Runs ML models for disease classification.
++1. **Etiology**: Genetic and environmental risk factors.
++1. **Molecular Markers**: Biomarkers such as amyloid-beta, tau, and alpha-synuclein.
++1. **Neuroanatomical Correlations**: Results from neuroimaging (e.g., MRI, PET).
--==== **3. Train Custom AI Models on EBRAINS HPC Resources** ====
++==== **Recommendation System** ====
--* Use EBRAINS **GPU and HPC clusters** for deep learning training:
++* Suggests additional tests or biomarkers if gaps are detected in the data.
++* Prioritizes tests based on clinical impact and cost-effectiveness.
--{{{from neurodiagnoses.models import train_cnn_model
--train_cnn_model(data_path='data/mri/', model_type='ResNet50')
--}}}
--* Save trained models for deployment:
++----
--{{{model.save('models/neurodiagnoses_cnn.h5')
--}}}
++=== **4. Computational Workflow** ===
--For further developments, contribute to the **[[Neurodiagnoses GitHub Repository>>url:https://github.com/neurodiagnoses]]**.
++1. **Data Loading**: Import data from storage (Drive or Bucket).
++1. **Feature Engineering**: Generate derived features from the raw data.
++1. **Model Training**:
++1*. Split data into training, validation, and test sets.
++1*. Train models with cross-validation to ensure robustness.
++1. **Evaluation**:
++1*. Metrics: Accuracy, F1-Score, AUIC for interpretability.
++1*. Compare against baseline models and domain benchmarks.
--**Collaboration & Partnerships**
++----
--Neurodiagnoses actively seeks partnerships with data providers to:
++=== **5. Validation** ===
--* Enable API-based data integration for real-time processing.
--* Co-develop harmonized AI-ready datasets with standardized annotations.
--* Secure funding opportunities through joint grant applications.
++==== **Internal Validation** ====
--**Interested in Partnering?**
++* Test the system using simulated datasets and known clinical cases.
++* Fine-tune models based on validation results.
--If you represent a research consortium or database provider, reach out to explore data-sharing agreements.
++==== **External Validation** ====
--**Contact**: [[info@neurodiagnoses.com>>mailto:info@neurodiagnoses.com]]
++* Collaborate with research institutions and hospitals to test the system in real-world settings.
++* Use anonymized patient data to ensure privacy compliance.
--**Final Notes**
++----
--Neurodiagnoses AI is committed to advancing the integration of artificial intelligence in neurodiagnostic processes. By continuously expanding our data ecosystem and incorporating standardized biomarker classifications through the Neuromarker ontology, we aim to enhance cross-disease AI training and improve diagnostic accuracy across neurodegenerative disorders.
++=== **6. Collaborative Development** ===
--We encourage researchers and institutions to contribute new datasets and methodologies to further enrich this collaborative platform. Your participation is vital in driving innovation and fostering a deeper understanding of complex neurological conditions.
++The project is open to contributions from researchers, clinicians, and developers. Key tools include:
--**For additional technical documentation and collaboration opportunities:**
++* **Jupyter Notebooks**: For data analysis and pipeline development.
++* **Wiki Pages**: For documenting methods and results.
++* **Drive and Bucket**: For sharing code, data, and outputs.
--* **GitHub Repository:** [[Neurodiagnoses GitHub>>url:https://github.com/neurodiagnoses]]
--* **EBRAINS Collaboration Page:** [[EBRAINS Neurodiagnoses>>url:https://ebrains.eu/collabs/neurodiagnoses]]
++----
--If you encounter any issues during data integration or have suggestions for improvement, please open a GitHub Issue or consult the EBRAINS Neurodiagnoses Forum. Together, we can advance the field of neurodiagnostics and contribute to better patient outcomes.
++=== **7. Tools and Technologies** ===
++
++* **Programming Languages**: Python for AI and data processing.
++* **Frameworks**:
++** TensorFlow and PyTorch for machine learning.
++** Flask or FastAPI for backend services.
++* **Visualization**: Plotly and Matplotlib for interactive and static visualizations.
++* **EBRAINS Services**:
++** Collaboratory Lab for running Notebooks.
++** Buckets for storing large datasets.

workflow neurodiagnoses.png

Author

...	...	@@ -1,1 +1,0 @@
1		-XWiki.manuelmenendez

Size

...	...	@@ -1,1 +1,0 @@
1		-157.5 KB

Content