Consiglio Nazionale delle Ricerche

Tipo di prodottoArticolo in rivista
TitoloDeep Learning-Based Methods for Prostate Segmentation in Magnetic Resonance Imaging
Anno di pubblicazione2021
Formato-
Autore/iAlbert Comelli 1,2, Navdeep Dahiya 3, Alessandro Stefano 2, Federica Vernuccio 4, Marzia Portoghese 4, Giuseppe Cutaia 4, Alberto Bruno 4, Giuseppe Salvaggio 4 and Anthony Yezzi 3
Affiliazioni autori1. Ri.MED Foundation, Via Bandiera, 11, 90133 Palermo, Italy; 2. Institute of Molecular Bioimaging and Physiology, National Research Council (IBFM-CNR), 90015 Cefalù, Italy; 3. Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA; 4. Dipartimento di Biomedicina, Neuroscienze e Diagnostica avanzata (BIND), University of Palermo, 90127 Palermo, Italy;
Autori CNR e affiliazioni
  • ALBERT COMELLI
  • ALESSANDRO STEFANO
Lingua/e
  • inglese
AbstractMagnetic Resonance Imaging-based prostate segmentation is an essential task for adaptive radiotherapy and for radiomics studies whose purpose is to identify associations between imaging features and patient outcomes. Because manual delineation is a time-consuming task, we present hree deep-learning (DL) approaches, namely UNet, efficient neural network (ENet), and efficient residual factorized convNet (ERFNet), whose aim is to tackle the fully-automated, real-time, and 3D delineation process of the prostate gland on T2-weighted MRI. While UNet is used in many biomedical image delineation applications, ENet and ERFNet are mainly applied in self-driving cars to compensate for limited hardware availability while still achieving accurate segmentation. We apply these models to a limited set of 85 manual prostate segmentations using the k-fold validation strategy and the Tversky loss function and we compare their results. We find that ENet and UNet are more accurate than ERFNet, with ENet much faster than UNet. Specifically, ENet obtains a dice similarity coefficient of 90.89% and a segmentation time of about 6 s using central processing unit (CPU) hardware to simulate real clinical conditions where graphics processing unit (GPU) is not always available. In conclusion, ENet could be efficiently applied for prostate delineation even in small image training datasets with potential benefit for patient management personalization.
Lingua abstractinglese
Altro abstract-
Lingua altro abstract-
Pagine da782
Pagine a-
Pagine totali13
RivistaApplied sciences
Attiva dal 2010
Editore: Molecular Diversity Preservation International - Basel
Lingua: inglese
ISSN: 2076-3417
Titolo chiave: Applied sciences
Titolo proprio: Applied sciences.
Titolo abbreviato: Appl. sci.
Numero volume della rivista11
Fascicolo della rivista-
DOI10.3390/app11020782
Verificato da referee-
Stato della pubblicazionePublished version
Indicizzazione (in banche dati controllate)-
Parole chiavedeep learning; segmentation; prostate; MRI; ENet; UNet; ERFNet; radiomics
Link (URL, URI)-
Titolo parallelo-
Licenza-
Scadenza embargo-
Data di accettazione12/01/2021
Note/Altre informazioni-
Strutture CNR
  • IBFM — Istituto di bioimmagini e fisiologia molecolare
Moduli/Attività/Sottoprogetti CNR
  • DSB.AD008.166.001 : TERAPIE CON RADIAZIONI IONIZZANTI E EFFETTI BIOLOGICI
Progetti Europei-
Allegati