×
  • Select the area you would like to search.
  • ACTIVE INVESTIGATIONS Search for current projects using the investigator's name, institution, or keywords.
  • EXPERTS KNOWLEDGE BASE Enter keywords to search a list of questions and answers received and processed by the ADNI team.
  • ADNI PDFS Search any ADNI publication pdf by author, keyword, or PMID. Use an asterisk only to view all pdfs.
Principal Investigator  
Principal Investigator's Name: Ghadah Fadhl
Institution: Sana'a university
Department: Faculty of computer and information technology
Country:
Proposed Analysis: Alzheimer disease is an irreversible and progressive disease. This disease has become lethal for many elder people around the world. One of them was my grandfather who died after two years of suffering. With the continuous breakthroughs in deep learning techniques specially in image processing and big data analysis, it become possible to develop more efficient methods for detecting AD in early stages which can lead to many medical diagnostic and therapeutic breakthroughs that could cure this complex disease or delay its symptoms. The proposed methodology will apply a multimodal deep learning for predicting Alzheimer's disease susceptibility. It uses longitudinal data of three different modalities: T1-weighted MRI, amyloid PET and genetic sequencing data (SNPs). Furthermore, the proposal will use two types of neural networks: deep neural network (DNN) for processing genetic dataset and convolutional neural network (CNN) for processing MRI (structural MRI) and AV-45 amyloid PET images. The two types of the networks will be merged at high levels in which the features extracted by the neural networks will be merged and classified. Additionally, the output layer of the fused neural network will consist of four neurons: MCIc (mild cognitive impairment converter to AD), MCInc (mild cognitive impartment not converter to AD), NC (normal control), and AD. This means that the model, in addition to detecting AD patients, will predicate if the patient has the susceptibility to develop AD in future, if the patient will only suffer of MCI, or if the patient will not develop any cognitive impartment. Moreover, in order to validate the performance accuracy of the networks, the proposal will run a comparison study between three different scenarios as follows: 1) fusing MRI and SNPs, 2) fusing AV-45 PET and SNPs, and 3) fusing MRI, AV-45 PET and SNPs Integrating two different types of data, images and genetic sequencing, will be challenging as both have different characteristics and manner. However, finding a way to link brain scans with the patient genotype sequence will help decipher the connection between genetic variations and brain structure's abnormality. The motive of using genetic dataset is to find the correlation between the genetic variations and the effect they may cause on the brain structure in order to know if a patient is susceptible to developing AD in future or not. In addition, integrating imaging data with bioinformatics can help find the genetic variants that have a strong influence on late-onset AD. Moreover, the hybrid neural network can be also used to discover the correlation between tau protein aggregation and Amyloid-B accumulation, which are believed to have a contribution to AD based on amyloid and tau hypotheses, with the aid of genetic data to identify the associated disease genes causing this aggregation and to predict patients who are susceptible to AD more accurately.
Additional Investigators