Machine Learning techniques for microwave brain stroke detection and classification - Degree thesis

Aug 3, 2026

Upload by Maeve

Strokes interrupt oxygenated blood supply to the brain and can cause severe injury, long-term disability, or death, making fast and appropriate diagnosis essential. This degree thesis investigates machine learning methods to detect and classify strokes using microwave-frequency measurements on head phantoms that replicate the electrical properties of human head tissues. The study targets improved precision and effectiveness, enabling faster, more accurate diagnostic support while reducing life-changing stroke impacts. Further work is proposed through alternative classification strategies, algorithm testing, and simulation calibration with measurements.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages119

Reading Time5 h 0 min

AccessFree to download

INTRODUCTION

CHAPTER 1. THEORETICAL FOUNDATION

1.1 Machine Learning

1.1.1 Support Vector Machine

1.1.2 Multi-Layer Perceptron

1.1.3 K-nearest neighbors

1.1.4 Confusion matrices

FINAL DEGREE PROJECT TITLE: Machine Learning techniques for microwave brain stroke detection and classification DEGREE: Degree in Telecommunication Systems Engineering AUTHOR: Emma Rodríguez Punset DIRECTORS: Francesca Vipiana / Ernesto Serrano Finetti DATE: 24th October of 2023 Títol: L’ús de técniques de Machine Learning per a la detecció i classificació d’infarts cerebrals per microones Autor: Emma Rodríguez Punset Director: Francesca Vipiana / Ernesto Serrano Finetti Data: 24 d’octubre del 2023 Resum Els ictus , definits per una interrupció del subministrament de sang oxigenada al cervell, són lesions cerebrals devastadores que poden causar danys profunds, discapacitat temporal o permanent, o fins i tot la mort. Es produeix quan un vas sanguini cerebral esclata (o trenca) o es bloqueja per un coàgul. Els pacients amb ictus representen una emergència mèdica greu, i per tal d'augmentar la probabilitat de recuperació i reduir els danys del pacient, el risc de mort o les discapacitats futures, el diagnòstic i el tractament apropiats ipuntuals són crucials. L'objectiu principal d'aquesta tesi és investigar l'ús de tècniques de Machine Learning (ML) per detectar i classificar ictus. El sistema utilitza “phantoms” jaque tenen les mateixes característiques elèctriques que els teixits del cap humà a freqüències de microones. Aquest projecte explora maneres de millorar la precisió i l'eficàcia de la detecció i classificació de ictus mitjançant l'aplicació de les capacitats computacionals del ML , amb l'objectiu de reduir elsefectes severs d'aquesta emergència mèdica. Per últim , aquest estudi representa un desenvolupament interessant en laincorporació del ML en el diagnòstic de ictus cerebrals, amb el potencial de millorar la precisió i la velocitat del diagnòstic. Es pot desenvolupar encara més investigant estratègies alternatives com millorar els mètodes de classificació, provar diferents algorismes i calibrar simulacions amb mesuraments. Com a activitat de recerca en curs, aquest projecte continua per millorar el diagnòstic i la predicció de ictus , motivat per la reducció dels efectes dels ictus, que canvien la vida als pacients. Title: Machine Learning techniques for microwave brain stroke detection and classification Author: Emma Rodríguez Punset Director: Francesca Vipiana / Ernesto Serrano Finetti Date: 24th October 2023 Abstract Strokes, defined by an interruption of the supply of oxygenated blood to the brain, are devastating brain injuries that can cause profound damage, temporary or permanent disability, or even death. It occurs when a brain bloodvessel bursts (or ruptures) or becomes blocked by a clot. Stroke patients represent a serious medical emergency, and in order to increase the probability of recovery and lower the patient's damages, risk of death, or future disabilities, appropriate and timely diagnosis and treatment are crucial. The main objective of this thesis is to investigate the use of machine learning (ML) techniques to detect and classify strokes. The system used is placed according to head phantoms, which have the same electrical characteristics as human head tissues at microwave frequencies. This project explores ways to improve the precision and efficacy of stroke detection and classification by applying machine learning's computational capabilities, ultimately with the goal of reducing the severe effects of this medical emergency. Ultimately, this study represents an interesting development in the incorporation of machine learning into stroke diagnosis, with the potential to improve the accuracy and speed of diagnosis. It can be further developed by investigating alternative strategies such as improving classification methods, testing different algorithms, and calibrating simulations with measurements. As an ongoing research activity, the continuing journey to improve stroke diagnosis and prediction will continue, motivated by reducing the life-changing eff

FAQs

What is the main objective of this thesis?

To investigate machine learning techniques for detecting and classifying strokes using microwave-based data.

How does the system obtain measurements for the ML model?

Measurements are collected using head phantoms designed to match the electrical characteristics of human head tissues at microwave frequencies.

What approaches are suggested for future improvements?

Improve classification methods, test different algorithms, and calibrate simulations using experimental measurements.

Aug 3, 2026

Upload by Maeve