Szczegóły publikacji
Opis bibliograficzny
An $U-Net$ - based regression model incorporating a parametric description of the prostate / Artur KOS // W: MIUA 2021 [Dokument elektroniczny] : 25th UK conference on Medical Image Understanding and Analysis : 12-14 July 2021, Oxford. — Wersja do Windows. — Dane tekstowe. — [Oxford] : [Oxford Institute of Biomedical Engineering], [2021]. — S. [1–4]. — Wymagania systemowe: Adobe Reader. — Tryb dostępu: https://miua2021.com/abstracts/121.pdf [2021-10-18]. — Bibliogr. s. [4]
Autor
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 137090 |
|---|---|
| Data dodania do BaDAP | 2021-10-22 |
| Rok publikacji | 2021 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
Abstract
Accurate prostate segmentation in the CT data remains a challenging task due to organ variation, typically unclear boarder between neighbouring structures and low signal-to-noise ratio of the CT data. We propose an U-Net based algorithm, which combines its encoder part with the parametric shape model of prostate. The presented algorithm acts as a regression model, has a linear activation function of the output layer and could be directly optimized by minimization of the MSE distance to ground-truth parameters. Segmentation accuracy of the proposed method was measured using Dice Similarity Coefficient (DSC) and was equal 0.73 with STD equal 0.18.
