الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 128 |  |  |
| | | from | 128 | | |
Abstract
Two analysis approaches were performed on simulated and experimental animal data based on the analysis of motoneuron spike train data to recognize the firing activity of ALS disease. In the simulated data, spiking latency, spike-triggered average signal, and inter-spike interval histogram were extracted and classified using a machine learning approach. Moreover, MN connectivity was investigated using a statistical-based method. The results achieved accuracies of ~99% using both extracted features and MN connectivity, showing a premise to be suggested as markers for ALS cellular changes. In the experimental data, a classification approach was proposed, based on the classical method of spike generation and features from the simulated datasets. Additionally, the spike generation method was modified to design a new method of classification. The results obtained achieved accuracies of ~81%. demonstrating the feasibility of discriminating between ALS and control experimental data of ALS progression.
The thesis is organized as follows: Chapter 1 gives an introduction to our research and presents the main research contributions. Chapter 2 discusses the theoretical foundations related to the thesis. Chapter 3 illustrates the analysis methods and results on the simulated datasets, highlighting the extracted markers and the classification results, in addition to analyzing the connectivity between motoneurons. Chapter 4 demonstrates our analysis approach applied to experimental data recoded from mice, presenting the classification accuracy obtained using the proposed approach that utilizes a novel method of spike generation. Chapter 5 concludes the thesis and discusses potential future work.