Early Detection of Alzheimer Using Deep Learning

Project Title

Early Detection of Alzheimer Using Deep Learning

Project Area of Specialization Artificial IntelligenceProject Summary

Alzheimer’s is a disease related to loss of memory and intellectual abilities which leads to dementia. Every three seconds a patient falls prey to it and the cost to combat it has surpassed $818 billion worldwide. In Pakistan alone, more than 2 million people are suffering from Alzheimer's disease (AD). The early detection of Alzheimer's disease and proper care could reduce the risk of the disease. But only 10 percent of the cases who suffer from this disease are diagnosed early by medical experts in developing countries like Pakistan. In our country, loss of memory and intellectual abilities is considered an aging factor and there is a serious lack of research and knowledge regarding this disease. Even though Alzheimer’s growth rate is alarming in Pakistan, very little research effort is focused on issues related to it.

In this research, the primary objective is to develop a deep learning based architecture using neuro-imaging data for computer aided early diagnosis (CAD) of Alzheimer’s disease. Standard benchmark datasets will help in fine tuning the system. Moreover, the coordination with local hospitals and laboratories in Pakistan will be an integral part of the research which shall facilitate the neurologists in the early diagnosis of Alzheimer’s disease. A further refinement will be to categorize the Alzheimer’s disease in different stages. State-of-the-art deep learning methods will be employed to achieve high performance of this CAD system. The secondary objective of proposed research based implementation model is to create awareness among people about this disease. The research and development in this area will improve health-care facilities of Pakistani medical institutions.

Project Objectives

The main objective is to propose a fast, reliable and efficient Deep Learning based Computer aided diagnosis of Alzheimer disease.

• The scope of this project is early diagnosis of AD, with clear classification of severity of the disease.
• To get higher accuracy in computer aided diagnosis (CAD) as compared to other state of the art techniques.

• To investigate current clinical practices in the identification and intervention for Alzheimer disease.
• To explore the development of computer aided diagnosis for AD in Pakistan.

A secondary objective of the project is to create awareness among people who have a chance to become victim of AD by early diagnosis, so the effected people and their caretakers should take preliminary measures to control the severity of the disease.

Project Implementation Method

In this research, the primary objective is to develop a deep learning based architecture using                        neuro-imaging data for computer aided early diagnosis (CAD) of Alzheimer’s disease. Standard benchmark datasets will help in fine tuning the system. Moreover, the coordination with local hospitals and laboratories in Pakistan (Rahila Research & Reference Lab) will be an integral part of the research which shall facilitate the neurologists in the early diagnosis of Alzheimer’s disease. A further refinement will be to categorize the Alzheimer’s disease in different stages. State-of-the-art deep learning methods will be employed to achieve high performance of this CAD system. The secondary objective of proposed research based implementation model is to create awareness among people about this disease. The research and development in this area will improve health-care facilities of Pakistani medical institutions.

Benefits of the Project

• The foremost expected benefit of this research based implementation project is higher accuracy in computer aided diagnosis of Alzheimer disease at an early stage to lower the severity of the disease.
• We expect the outcome model of this research will be a second opinion alongside neurologists for the early diagnosis of Alzheimer disease.

• This will help to create awareness about Alzheimer disease among people and care takers of the Alzheimer patients.

Technical Details of Final Deliverable

Figure: Block Diagram of proposed Model

In our proposed research, we will use a deep learning architecture, consisting of stacked sparse auto encoders and a softmax regression layer. This proposed method will work for multi-class classification. This will not only classify AD but also predict the risk of Mild cognitive impairment (MCI). Prediction is categorized into four classes, Alzheimer Disease (AD), normal control (NC), MCI non converters (ncMCI) and MCI converters (cMCI). The sparse auto encoder is an encoding structure, which consists of a neural network with multiple hidden layers. The proposed model will be trained and test using Graphical Processing Unit (GPU).

Final Deliverable of the Project HW/SW integrated systemCore Industry MedicalOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Good Health and Well-Being for People, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	54839
GIGABYTE GA-Z390X AORUS ELITE LGA	Equipment	1	32973	32973
ADATA SSD 256GB M.2 NVME	Equipment	1	9866	9866
12 V Adapters	Equipment	4	500	2000
Cloud services Licencing	Miscellaneous	1	8000	8000
Cables wires	Miscellaneous	10	200	2000