DenseIncepS115: a novel network-level fusion framework for Alzheimer's disease prediction using MRI images
2024

DenseIncepS115: A New Framework for Predicting Alzheimer's Disease Using MRI

Sample size: 416 publication 10 minutes Evidence: high

Author Information

Author(s): Rauf Fatima, Khan Muhammad Attique, Brahim Ghassen Ben, Abrar Wardah, Alasiry Areej, Marzougui Mehrez, Jeon Seob, Nam Yunyoung

Primary Institution: HITEC University, Taxila, Pakistan; Prince Mohammad Bin Fahd University, Al Khobar, Saudi Arabia; King Khalid University, Abha, Saudi Arabia; Soonchunhyang University, Cheonan, Republic of Korea; Soonchunhyang University, Asan, Republic of Korea

Hypothesis

Can a novel deep learning architecture improve the prediction of Alzheimer's disease from MRI scans?

Conclusion

The proposed DenseIncepS115 architecture achieved high accuracy levels of 99.5% and 98.7% on two publicly available datasets for Alzheimer's disease prediction.

Supporting Evidence

  • The DenseIncepS115 architecture achieved an accuracy of 99.5% on the ADNI dataset.
  • The architecture was validated using two publicly available datasets.
  • Feature extraction was optimized using the Catch Fish Optimization algorithm.
  • The model was trained on a dataset of 416 patients with Alzheimer's disease.

Takeaway

This study created a new computer program that helps doctors predict Alzheimer's disease using brain scans, making it easier to catch the disease early.

Methodology

The study used a novel deep learning architecture that combines inception and dense modules with self-attention for classifying MRI images of Alzheimer's patients.

Potential Biases

Potential biases may arise from the datasets used, which may not represent the broader population.

Limitations

The study may be limited by the class imbalance in the datasets and the reliance on specific MRI datasets for training and testing.

Participant Demographics

The study included 416 patients aged 60 and older, with a focus on those diagnosed with Alzheimer's disease.

Digital Object Identifier (DOI)

10.3389/fonc.2024.1501742

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication