Deep Vision based Digital Diagnostic Aide for the Brain Diseases using fMRI
Previously a lot of work has been done to diagnose brain diseases using CT scans, PET and MRI but they diagnose the disease very late when it becomes incurable. Diagnosing Alzheimer's Disease (AD) in the very first stage using FMRI processing is really a difficult task and requires very efficient se
2025-06-28 16:31:06 - Adil Khan
Deep Vision based Digital Diagnostic Aide for the Brain Diseases using fMRI
Project Area of Specialization Artificial IntelligenceProject SummaryPreviously a lot of work has been done to diagnose brain diseases using CT scans, PET and MRI but they diagnose the disease very late when it becomes incurable. Diagnosing Alzheimer's Disease (AD) in the very first stage using FMRI processing is really a difficult task and requires very efficient segmentation and classification. Our motivation is to make our system as efficient as possible so it may become able to detect ’Mild Cognitive Impairment’ (MCI) before converting into AD as AD cannot be completely cured where MCI can be medicated and also diminish the risk of Alzheimer’s disease in senior citizens by diagnosing it at an early stage.
Project Objectives- Along with the diagnosis of the disease, our objective is to classify the different stages of the disease.
- To provide a system having better accuracy.
- To provide the new-comers with work having increased accuracy.
- Analyze the Patient’s report to detect infected parts.
Development Methodology :
1. Data Acquisition
2. Pre-Processing
3-Modeling
4- Evaluation
5. Ensembling
Data Acquisition In data acquisition we will collect data from different organization databases including
- ADNI (Alzheimer’s disease Neuroimaging Initiative)
- IBSR (Internet Brain Segmentation Repository)
- OASIS (Open Access Series of Imaging Studies)
- MICCAI (Medical Image Computing and Computer-Assisted Intervention)
Then we will define the subjects of each dataset and the image type of our dataset will be fMRI. Then we will define the number of all images that our dataset has and also related to each subject.
2- Pre-Processing
The fMRI preprocessing includes steps which are given below:
- Re-orientation
- Brain Extraction
- Motion Correction
- Slice timing Correction
- Spatial Smoothing
- Enhancement
- Image Conversion
3- Modeling:
The model we will use in this is to classify Alzheimer’s patients and Healthy Patient. Deep Learning techniques are used to diagnose Alzheimer’s disease which includes three main approaches.
- LeNet
- Alexnet
- ZFNet
- VGG16
- VGG19
- DenseNet
- ResNet
- Inception-Net
4- Evaluation:
In the end, we will evaluate the model on the basis of the following metrics.
- Accuracy
- Sensitivity
- Specificity
- Precision
- Recall
- F1 Score
5- Ensemble :
Ensembling helps improve Deep learning results by combining several models. Ensemble methods are meta-algorithms that combine several models' results into one predictive model in order to decrease loss (cross_entropy) or improve predictions (accuracy).
6-Development:
The end product is a web-based application where patients can read their reports after uploading.
Benefits of the ProjectOur Project will have the following benefits
- By using our system patients can read their own reports.
- Now on the spot, there is no need for an appointment with the doctor for a Report consultation.
- To provide a fast and efficient platform to patients to diagnose their disease along with the classification of the dementia stages
The end product of our project would be such an application where the user will be able to get his fMRI report read automatically by uploading his fMRI imaging report in our application. Our system will not only diagnose the disease but will also tell the stage of dementia that patient has reached
Final Deliverable of the Project Software SystemCore Industry ITOther Industries Medical Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 70000 | |||
| gtx 1060 ti | Equipment | 2 | 25000 | 50000 |
| Samsung SSD 970 EVO PLUS NVME M.2 500GB - MZ-V7S500BW | Equipment | 1 | 18000 | 18000 |
| Printing Documents and other expenditures | Miscellaneous | 1 | 2000 | 2000 |