Politecnico di Torino (logo)

Analysis of the pupillary response in Amyotrophic Lateral Sclerosis patients

Sara Jlassi

Analysis of the pupillary response in Amyotrophic Lateral Sclerosis patients.

Rel. Luca Mesin. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2022

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (8MB) | Preview

Amyotrophic Lateral Sclerosis (ALS) is an extremely disabling pathology that leads to complete paralysis in a short time. In particular, the patient could reach a condition known as "Locked-In Syndrome" (LIS), in which all voluntary movements are prevented (or "Complete Locked-In Syndrome", if also ocular movements are compromised). In this state, communication with the outside world is impossible and support strategies are required. The disease affects only upper and lower motor neurons, whereas other neurological functions remain unchanged. The pupillary activity is under the control of the Autonomic Nervous System (ANS), whose functions do not appear to be affected as the disease progresses. The pupil size varies in response to external stimuli, such as an ambient light change, or to a variation in mental load. Moreover, pupillary diameter oscillations are present even in the absence of stimuli, which represent the so-called "Pupillary Hippus". The purpose of this study is to understand if the pupillary activity is not altered by the ALS progression, to exploit it as a mean of communication with the outside. To achieve this goal, the pupillographic signals of 33 ALS patients and 10 healthy subjects have been analysed, by separately studying the first part of the time series, that represents the Hippus, and the second one, which shows the pupil response to light stimuli (that is the "Pupillary Light Reflex"). In both cases, a series of indices has been extracted from the signal in the time domain and from its power spectral density (PSD), allowing to create a data set of variables and divided into classes according to the severity of the disease (dictated by the ALSFRS-R score). Multiple tests were carried out: initially, only the data set of ALS patients was considered, which was divided into three classes (severely, moderately and poorly invalidated subjects); then, the recordings of healthy people were included and some tests on both two (healthy and pathological subjects) and three (ALS patients with severe disease or moderate symptoms and healthy subjects) classes subdivision have been carried out. For any index, the frequency distributions of each class have been compared, through the Kruskall-Wallis and the Wilcoxon tests, in order to assess if the groups belonged to the same population. Moreover, a binary tree classifier has been exploited to predict the class of each subject. From this analysis, it resulted that some indices show significant difference among the classes: variance, skewness and kurtosis of the Hippus frequency distribution and other features like the PSD mean frequency (MNF) and the PLR amplitude have been taken into account to discriminate the groups. Furthermore, it has been found that, given a good combination of features as input, the two classes subdivision of data set is the best one, with a classification accuracy of 93.75%. However, the high data variability and the limited size of the data set do not allow to affirm that there are evident variations in pupillary activity and, consequently, the hypothesis that the ANS activity is not altered by the progression of the disease cannot be rejected. Therefore, the communication through pupil size variations could be an effective strategy for LIS people.

Relators: Luca Mesin
Academic year: 2021/22
Publication type: Electronic
Number of Pages: 100
Corso di laurea: Corso di laurea magistrale in Ingegneria Biomedica
Classe di laurea: New organization > Master science > LM-21 - BIOMEDICAL ENGINEERING
Aziende collaboratrici: UNSPECIFIED
URI: http://webthesis.biblio.polito.it/id/eprint/23775
Modify record (reserved for operators) Modify record (reserved for operators)