Automatic semantic segmentation of the lumbar spine: Clinical applicability in a multi-parametric and multi-center study on magnetic resonance images.

Autor: Sáenz-Gamboa JJ; FISABIO-CIPF Joint Research Unit in Biomedical Imaging, Fundaciò per al Foment de la Investigaciò Sanitària i Biomèdica (FISABIO), Av. de Catalunya 21, 46020 València, Spain. Electronic address: jsaenz@laberit.com., Domenech J; Orthopedic Surgery Department, Hospital Arnau de Vilanova, Carrer de San Clemente s/n, 46015, València, Spain., Alonso-Manjarrés A; Radiology Department, Hospital Arnau de Vilanova, Carrer de San Clemente s/n, 46015, València, Spain., Gómez JA; Pattern Recognition and Human Language Technology research center, Universitat Politècnica de València, Camí de Vera, s/n, 46022, València, Spain., de la Iglesia-Vayá M; FISABIO-CIPF Joint Research Unit in Biomedical Imaging, Fundaciò per al Foment de la Investigaciò Sanitària i Biomèdica (FISABIO), Av. de Catalunya 21, 46020 València, Spain; Regional ministry of Universal Health and Public Health in Valencia, Carrer de Misser Mascó 31, 46010 València, Spain. Electronic address: delaiglesia_mar@gva.es.
Jazyk: angličtina
Zdroj: Artificial intelligence in medicine [Artif Intell Med] 2023 Jun; Vol. 140, pp. 102559. Date of Electronic Publication: 2023 Apr 26.
DOI: 10.1016/j.artmed.2023.102559
Abstrakt: Significant difficulties in medical image segmentation include the high variability of images caused by their origin (multi-center), the acquisition protocols (multi-parametric), the variability of human anatomy, illness severity, the effect of age and gender, and notable other factors. This work addresses problems associated with the automatic semantic segmentation of lumbar spine magnetic resonance images using convolutional neural networks. We aimed to assign a class label to each pixel of an image, with classes defined by radiologists corresponding to structural elements such as vertebrae, intervertebral discs, nerves, blood vessels, and other tissues. The proposed network topologies represent variants of the U-Net architecture, and we used several complementary blocks to define the variants: three types of convolutional blocks, spatial attention models, deep supervision, and multilevel feature extractor. Here, we describe the topologies and analyze the results of the neural network designs that obtained the most accurate segmentation. Several proposed designs outperform the standard U-Net used as a baseline, primarily when used in ensembles, where the outputs of multiple neural networks are combined according to different strategies.
Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
(Copyright © 2023 The Authors. Published by Elsevier B.V. All rights reserved.)
Databáze: MEDLINE
Externí odkaz: