Comparative Analysis of Different Machine Learning Algorithms for Detection of Alzheimer Disease from Medical images

Alzheimer's disease is a neurological, progressive brain disorder and multifaceted disease that slowly destroys brain cells. In this disease brain cells are affected and due to that communication between cells are not taken properly, which ultimately causes the memory loss. Early detection of the Alzheimer is very crucial to slow down the progress of the disease. For diagnosing the disease many imaging techniques are there like magnetic resonance imaging (MRI), commuted tomography (CT) scan, Positron emission tomography (PET). These imaging techniques are good but still have some limitations in the form of accuracy. So now a day's different machine learning algorithms are available for improvising the same. In our proposed research title we want to implement transfer learning and compare different classifier on the bases of parameters like Accuracy, precision, recall and F1 score. To perform the research, we got the data set from ADNI i.e. Alzheimer Disease Neuroimaging Initiatives. Objectives Alzheimer's disease is the main dementia in old person, slowly which causes memory loss and which tends to the death. AD is currently not recoverable but diagnosis is done by two parameters i.e. MSME score and by manual examination of the MRI scan images. As there are different modalities are available for the diagnosis like MRI and PET. MRI is a good imaging technique for the diagnosis but not everyone can go for the MRI. For example, a person with metal implanted in body cannot go for MRI because in the MRI chamber very strong magnetic field is present, another problem is that if a person is having tattoo then he/she cannot go in MRI because many of the tattoo material contain small magnet particle in it and also MRI cost is very high. So another option is go with PET scan. In PET scan a liquid is injected in the body then this liquid travels through the body and a scanner is placed to get the PET images. In this paper we want to compare different Machine learning and transfer learning techniques with the PET images so that researchers will get to that which classifier will be best suited for the Alzheimer disease detection using PET images. So another objective of the paper is to compare different classifiers used now a days for detection of the Alzheimer disease, currently the disease getting diagnosed by manually seeing an studying the MRI or CT scan images, in many cases MSE is calculated to diagnose the AD, Researchers are now a days giving their most of the time to find the best suited algorithm for the detection of Alzheimer. Here in this paper, we are going to compare different machine learning algorithm for the Alzheimer detection. Materials and method: Here We present a comparison of different transfer learning classifiers with the mechanism of Alzheimer's disease from ADNI database of Alzheimer's disease; the dataset used in this project contains PET images of Alzheimer's patients along with normal person's image. We have applied the transfer learning methods to get the comparison between different classifier such as support vector machine, linear discriminate, KNN, Naive Bayes and decision trees.