Tesi etd-06252021-173635 |
Link copiato negli appunti
Tipo di tesi
Tesi di laurea magistrale
Autore
CARLONI, GIANLUCA
URN
etd-06252021-173635
Titolo
Study and development of advanced models integrating radiomic features and clinical data for outcome prediction in non-small cell lung cancer patients treated for brain metastases with stereotactic radiotherapy
Dipartimento
INGEGNERIA DELL'INFORMAZIONE
Corso di studi
INGEGNERIA BIOMEDICA
Relatori
relatore Ing. Positano, Vincenzo
correlatore Prof.ssa Jereczek, Barbara Alicja
correlatore Dott.ssa Garibaldi, Cristina
correlatore Prof.ssa Jereczek, Barbara Alicja
correlatore Dott.ssa Garibaldi, Cristina
Parole chiave
- brain metastases
- features
- immagini biomediche
- magnetic resonance
- medical imaging
- metastasi cerebrali
- modello predittivo
- non small cell lung cancer
- predictive model
- Python
- radiomica
- radiomics
- radioterapia
- radiotherapy
- risonanza magnetica
- tumore polmonare non a piccole cellule
Data inizio appello
16/07/2021
Consultabilità
Completa
Riassunto
In questo lavoro sono stati sviluppati in Python modelli di previsione dei risultati in pazienti con metastasi cerebrali da tumore al polmone trattati con radiochirurgia. I modelli predittivi sono stati elaborati a partire da dati clinici e da features radiomiche estratte da immagini RMN precedenti al trattamento. Si sono indagati tre principali endpoint: controllo locale e progressione a distanza delle lesioni encefaliche, e sopravvivenza globale dei pazienti. Dopo aver sviluppato tecniche di normalizzazione delle intensità delle immagini, sono state estratte le features utilizzando due diverse piattaforme permettendo un confronto end-to-end sulla potenza predittiva. Infine, l'analisi statistica dei risultati tramite analisi univariata/multivariata con modelli Cox e LASSO.
In this work, outcome prediction models were developed in Python for patients with brain metastases from lung cancer treated with radiosurgery. Predictive models were developed from clinical data and radiomic features extracted from pre-treatment MRI images. Three main endpoints were investigated: local control and distant progression of brain lesions, and overall survival of patients. After developing techniques to normalise image intensities, features were extracted using two different platforms allowing an end-to-end comparison of predictive power. Finally, statistical analysis of the results by univariate/multivariate analysis with Cox and LASSO models
In this work, outcome prediction models were developed in Python for patients with brain metastases from lung cancer treated with radiosurgery. Predictive models were developed from clinical data and radiomic features extracted from pre-treatment MRI images. Three main endpoints were investigated: local control and distant progression of brain lesions, and overall survival of patients. After developing techniques to normalise image intensities, features were extracted using two different platforms allowing an end-to-end comparison of predictive power. Finally, statistical analysis of the results by univariate/multivariate analysis with Cox and LASSO models
File
Nome file | Dimensione |
---|---|
TesiCarl...nluca.pdf | 4.33 Mb |
Contatta l’autore |