|
| 1 | +--- |
| 2 | +type: "project" # DON'T TOUCH THIS ! :) |
| 3 | +date: "2025-06-03" # Date you first upload your project. |
| 4 | +# Title of your project (we like creative title) |
| 5 | +title: "Functional Connectivity in ADHD: Group Differences and Predictive Modeling During Spatial Working Memory Task" |
| 6 | + |
| 7 | +# List the names of the collaborators within the [ ]. If alone, simple put your name within [] |
| 8 | +names: [Nilay Ozdemir Haksever] |
| 9 | + |
| 10 | +# Your project GitHub repository URL |
| 11 | +github_repo: https://github.com/nilayoh/swm_fmri_project.git |
| 12 | + |
| 13 | +# If you are working on a project that has website, indicate the full url including "https://" below or leave it empty. |
| 14 | +website: |
| 15 | + |
| 16 | +# List +- 4 keywords that best describe your project within []. Note that the project summary also involves a number of key words. Those are listed on top of the [github repository](https://github.com/brainhack-school2020/project_template), click `manage topics`. |
| 17 | +# Please only lowercase letters |
| 18 | +tags: [spatial working memory, functional connectivity, machine learning, ADHD] |
| 19 | + |
| 20 | +# Summarize your project in < ~75 words. This description will appear at the top of your page and on the list page with other projects.. |
| 21 | + |
| 22 | +summary: "Firstly, this project investigates differences in frontoparietal brain connectivity between individuals diagnosed with ADHD and control participants during Spatial Working Memory Task, using fMRI-based connectivity data. In the second part of this project, To classify individuals as either having ADHD or being in control group based on functional connectivity data features machine learning models was tested by using k-fold cross validation. " |
| 23 | + |
| 24 | +# If you want to add a cover image (listpage and image in the right), add it to your directory and indicate the name |
| 25 | +# below with the extension. |
| 26 | +image: "Image.jpg" |
| 27 | +--- |
| 28 | +<!-- This is an html comment and this won't appear in the rendered page. You are now editing the "content" area, the core of your description. Everything that you can do in markdown is allowed below. We added a couple of comments to guide your through documenting your progress. --> |
| 29 | + |
| 30 | +## Project definition |
| 31 | + |
| 32 | +### Background |
| 33 | +Working memory is a core component of executive function, and spatial working memory is often impaired in ADHD. |
| 34 | +Deficits in working memory contribute to: |
| 35 | +* Inattention, impulsivity, and poor task performance |
| 36 | +* Difficulty sustaining attention in spatial contexts |
| 37 | +Neural Mechanisms: |
| 38 | +SWM tasks typically activate: |
| 39 | +* Dorsolateral prefrontal cortex (DLPFC): for maintaining and manipulating information |
| 40 | +* Parietal cortex: for spatial processing |
| 41 | + |
| 42 | +In ADHD, these areas show reduced activation. |
| 43 | + |
| 44 | +### Tools |
| 45 | + |
| 46 | +* The "project template" project will rely on the following technologies: |
| 47 | +* Git and Github |
| 48 | +* Python libraries: |
| 49 | +* ScikitLearn for machine learning, |
| 50 | +* Nilearn for fMRI connectivity analysis |
| 51 | +* Pyplot, Matplotlib, Seaborn for data visualization |
| 52 | +* Jupyter Notebook |
| 53 | + |
| 54 | +### Data |
| 55 | +The dataset obtained from following project. |
| 56 | + |
| 57 | +Alpha oscillations and working memory deficits in ADHD: A multimodal imaging investigation (R01MH116268) |
| 58 | + |
| 59 | +https://nda.nih.gov/edit_collection.html?id=3101 |
| 60 | + |
| 61 | +It was a multimodal data containing both fMRI and EEG. This project only focused on the fMRI data. A total of 68 young adults was used in these analyses (39 ADHD, 29 Control; Mage: 21.35) |
| 62 | +The data was based on a computerized version of the spatial working memory (SWM) task. |
| 63 | + |
| 64 | + |
| 65 | + |
| 66 | +### Deliverables |
| 67 | + |
| 68 | +At the end of this project, I have: |
| 69 | +- A Github repository with code scripts (https://github.com/nilayoh/swm_fmri_project.git) |
| 70 | +- Figures and results summarizing the performance of the model and showing the functional connectivity across regions |
| 71 | +- Jupyter notebook that includes analysis codes and visualizations |
| 72 | + |
| 73 | + |
| 74 | +## Results |
| 75 | + |
| 76 | + |
| 77 | +These images show mean functional connectivity across frontoparietal regions of the brain. Blue color indicate higher functional connectivity for Control than ADHD patients. Red color indicates higher functional connectivity for ADHD patients than control group. |
| 78 | + |
| 79 | + |
| 80 | + |
| 81 | +These images show significant differences between ADHD and control group across frontoparietall regions of the brain before FDR correction. |
| 82 | + |
| 83 | + |
| 84 | +No significant difference between ADHD and Control group across these areas after FDR correction. |
| 85 | + |
| 86 | +Even though I could not find significant difference for practice I would like to continue to investigate my second part of the project. |
| 87 | +I used machine learning methods to classify participants as ADHD or control. I used k-fold cross validation (k=5) to test my model. |
| 88 | +The cross validation accuracy across folds can be seen in following graph. |
| 89 | + |
| 90 | + |
| 91 | +## Conclusion and acknowledgement |
| 92 | + |
| 93 | +What I’ve Learned: |
| 94 | + |
| 95 | +Understanding fMRI Data: |
| 96 | +* Downloading, organizing datasets |
| 97 | +* Managing confounding variables |
| 98 | +Functional Connectivity Analysis: |
| 99 | +* Comparing ADHD vs. Control groups using connectivity matrices |
| 100 | +* Using atlases for brain parcellation |
| 101 | +Debugging & Code Practice: |
| 102 | +* Troubleshooting analysis pipelines in Python |
| 103 | +* Gaining confidence in working with neuroimaging libraries (e.g., Nilearn, Nibabel) |
| 104 | +Introduction to Machine Learning: |
| 105 | +* Applying classification techniques to distinguish between ADHD and Control groups |
| 106 | +* Exploring pipelines for diagnosis-based predictions |
| 107 | + |
| 108 | +Thank you to all TAs, Dr. Erin Dickie, for all your help in the process and this opportunity. Also, thank you to Joel Diaz for providing the data and helping. |
0 commit comments