Detail publikačního výsledku

Comparison of spine segmentation algorithms on clinical data from spectral CT of patients with multiple myeloma

NOHEL, M.; JAKUBÍČEK, R.; BLAŽKOVÁ, L.; VÁLEK, V.; DOSTÁL, M.; OUŘEDNÍČEK, P.; CHMELÍK, J.

Originální název

Comparison of spine segmentation algorithms on clinical data from spectral CT of patients with multiple myeloma

Anglický název

Comparison of spine segmentation algorithms on clinical data from spectral CT of patients with multiple myeloma

Druh

Stať ve sborníku v databázi WoS či Scopus

Originální abstrakt

This article presents an evaluation of spine segmentation models using clinical data obtained from multiple myeloma patients. The performance of the models is compared based on the classical Dice score. The results show that the Payer and nnU-Net models show the highest level of similarity in segmentation. However, when it comes to the challenging task of segmenting cervical vertebrae, the Payer algorithm provides more accurate results. On the other hand, the nnU-Net model achieves better results in cases of extensive vertebral deformation. We also observed that convolutional neural networks have problems in segmenting metal surgical implants. Research highlights the strengths and weaknesses of different models and can help select appropriate segmentation algorithms for specific clinical scenarios.

Anglický abstrakt

This article presents an evaluation of spine segmentation models using clinical data obtained from multiple myeloma patients. The performance of the models is compared based on the classical Dice score. The results show that the Payer and nnU-Net models show the highest level of similarity in segmentation. However, when it comes to the challenging task of segmenting cervical vertebrae, the Payer algorithm provides more accurate results. On the other hand, the nnU-Net model achieves better results in cases of extensive vertebral deformation. We also observed that convolutional neural networks have problems in segmenting metal surgical implants. Research highlights the strengths and weaknesses of different models and can help select appropriate segmentation algorithms for specific clinical scenarios.

Klíčová slova

spine segmentation, spectral CT, multiple myeloma, nnU-Net, deep learning

Klíčová slova v angličtině

spine segmentation, spectral CT, multiple myeloma, nnU-Net, deep learning

Autoři

NOHEL, M.; JAKUBÍČEK, R.; BLAŽKOVÁ, L.; VÁLEK, V.; DOSTÁL, M.; OUŘEDNÍČEK, P.; CHMELÍK, J.

Rok RIV

2025

Vydáno

04.01.2024

Nakladatel

Springer Nature Switzerland

Místo

Cham

ISBN

978-3-031-49061-3

Kniha

MEDICON'23 & CMBEBIH'23

Edice

93

ISSN

1680-0737

Periodikum

IFMBE Proceedings

Svazek

93

Stát

Francouzská republika

Strany od

309

Strany do

317

Strany počet

9

URL

https://link.springer.com/chapter/10.1007/978-3-031-49062-0_34

BibTex

@inproceedings{BUT184739,
  author="Michal {Nohel} and Roman {Jakubíček} and Lenka {Blažková} and Vlastimil {Válek} and Marek {Dostál} and Petr {Ouředníček} and Jiří {Chmelík}",
  title="Comparison of spine segmentation algorithms on clinical data from spectral CT of patients with multiple myeloma",
  booktitle="MEDICON'23 & CMBEBIH'23",
  year="2024",
  series="93",
  journal="IFMBE Proceedings",
  volume="93",
  number="1",
  pages="309--317",
  publisher="Springer Nature Switzerland",
  address="Cham",
  doi="10.1007/978-3-031-49062-0\{_}34",
  isbn="978-3-031-49061-3",
  issn="1680-0737",
  url="https://link.springer.com/chapter/10.1007/978-3-031-49062-0_34"
}