Tesi etd-01312022-173242 |
Link copiato negli appunti
Tipo di tesi
Tesi di laurea magistrale
Autore
BASHA, MARIS
URN
etd-01312022-173242
Titolo
Interpretable COVID-19 severity prediction from lung ultrasound using Spatio-temporal Neural Networks
Dipartimento
INFORMATICA
Corso di studi
INFORMATICA
Relatori
relatore Bacciu, Davide
Parole chiave
- covid-19
- interpretability
- lung ultrasound
- neural networks
Data inizio appello
25/02/2022
Consultabilità
Non consultabile
Data di rilascio
25/02/2092
Riassunto
The theis deals with severity prediction of pulmunary infections from lung ultrasounds using Spatio-temporal neural networks. The thesis explores the use of predictors based on SlowFast networks, and their comparison with I3D and P3D. It also explores interpretability aspects, considering visualization of weight and layer activations, activation maps, and Gradient-weighted Class Activation Mapping. The models have been trained and validated on real-world data collected at Cisanello hospital.
File
| Nome file | Dimensione |
|---|---|
Tesi non consultabile. |
|