Claudia Ottomano
X-ray scatter estimation for arbitrary spectra in digital breast tomosynthesis.
Rel. Kristen Mariko Meiburger, Ioannis Sechopoulos, Pim Van Den Berg. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2026
|
|
PDF (Tesi_di_laurea)
- Tesi
Accesso limitato a: Solo utenti staff fino al 28 Settembre 2027 (data di embargo). Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (12MB) |
Abstract
X-ray scatter radiation in digital breast tomosynthesis (DBT) reduces image contrast and quantitative accuracy. Monte Carlo (MC) simulations accurately estimate scatter but are computationally expensive, motivating deep-learning (DL)-based alternatives. A previous DL model estimated scatter in DBT projections but was limited to the cranio-caudal (CC) view and a fixed set of polychromatic spectra, defined by a thickness-dependent lookup table. This model was extended to generalize across clinically relevant x-ray spectra and to both the CC and medio-lateral oblique (MLO) views, by explicitly providing the incident spectrum as input to the network and extending the training set. MC scatter maps for training were generated from homogeneous CC- and MLO-view breast phantoms, simulated at 16 discrete energies from 10 keV to 50 keV, allowing arbitrary polychromatic spectra to be synthesized.
Validation data was generated similarly but used seven polychromatic spectra outside the training set to test generalization
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
