Deep learning enabled brain shunt valve identification using mobile phones.
Deep learning
Magnetic resonance imaging
Mobile phone camera
Programmable cerebrospinal fluid shunt valve
Journal
Computer methods and programs in biomedicine
ISSN: 1872-7565
Titre abrégé: Comput Methods Programs Biomed
Pays: Ireland
ID NLM: 8506513
Informations de publication
Date de publication:
Oct 2021
Oct 2021
Historique:
received:
09
03
2021
accepted:
09
08
2021
pubmed:
2
9
2021
medline:
29
9
2021
entrez:
1
9
2021
Statut:
ppublish
Résumé
Accurate information concerning implanted medical devices prior to a Magnetic resonance imaging (MRI) examination is crucial to assure safety of the patient and to address MRI induced unintended changes in device settings. The identification of these devices still remains a very challenging task. In this paper, with the aim of providing a faster device detection, we propose the adoption of deep learning for medical device detection from X-rays. In particular, we propose a pipeline for the identification of implanted programmable cerebrospinal fluid shunt valves using X-ray images of the radiologist workstation screens captured with mobile phone integrated cameras at different angles and illuminations. We compare the proposed convolutional neural network with published methods. Experimental results show that this approach outperforms methods trained on images digitally transferred directly from the scanners and then applied on mobile phones images (mean accuracy 95% vs 77%, Avg. Precision 0.96 vs 0.77, Avg. Recall 0.95 vs 0.77, Avg. F1-score 0.95 vs 0.77) and existing published methods based on transfer learning fine-tuned directly on the mobile phone images (mean accuracy 94% vs 75%, Avg. Precision 0.94 vs 0.75, Avg. Recall 0.94 vs 0.75, Avg. F1-score 0.94 vs 0.75). An automated shunt valve identification system is a promising safety tool for radiologists to efficiently coordinate the care of patients with implanted devices. An image-based safety system able to be deployed on a mobile phone would have significant advantages over methods requiring direct input from X-ray scanners or clinical picture archiving and communication system (PACS) in terms of ease of integration in the hospital or clinical ecosystems.
Sections du résumé
BACKGROUND AND OBJECTIVE
OBJECTIVE
Accurate information concerning implanted medical devices prior to a Magnetic resonance imaging (MRI) examination is crucial to assure safety of the patient and to address MRI induced unintended changes in device settings. The identification of these devices still remains a very challenging task. In this paper, with the aim of providing a faster device detection, we propose the adoption of deep learning for medical device detection from X-rays.
METHOD
METHODS
In particular, we propose a pipeline for the identification of implanted programmable cerebrospinal fluid shunt valves using X-ray images of the radiologist workstation screens captured with mobile phone integrated cameras at different angles and illuminations. We compare the proposed convolutional neural network with published methods.
RESULTS
RESULTS
Experimental results show that this approach outperforms methods trained on images digitally transferred directly from the scanners and then applied on mobile phones images (mean accuracy 95% vs 77%, Avg. Precision 0.96 vs 0.77, Avg. Recall 0.95 vs 0.77, Avg. F1-score 0.95 vs 0.77) and existing published methods based on transfer learning fine-tuned directly on the mobile phone images (mean accuracy 94% vs 75%, Avg. Precision 0.94 vs 0.75, Avg. Recall 0.94 vs 0.75, Avg. F1-score 0.94 vs 0.75).
CONCLUSION
CONCLUSIONS
An automated shunt valve identification system is a promising safety tool for radiologists to efficiently coordinate the care of patients with implanted devices. An image-based safety system able to be deployed on a mobile phone would have significant advantages over methods requiring direct input from X-ray scanners or clinical picture archiving and communication system (PACS) in terms of ease of integration in the hospital or clinical ecosystems.
Identifiants
pubmed: 34469808
pii: S0169-2607(21)00430-2
doi: 10.1016/j.cmpb.2021.106356
pmc: PMC8478889
mid: NIHMS1733010
pii:
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
106356Subventions
Organisme : NINDS NIH HHS
ID : R01 NS121154
Pays : United States
Organisme : NLM NIH HHS
ID : T15 LM007093
Pays : United States
Organisme : NCATS NIH HHS
ID : UL1 TR003167
Pays : United States
Informations de copyright
Copyright © 2021. Published by Elsevier B.V.
Déclaration de conflit d'intérêts
Declaration of Competing Interest The authors declare that they have no known competing financial interests. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Références
Artif Intell Med. 2019 Sep;100:101707
pubmed: 31607347
Sci Rep. 2018 Oct 30;8(1):16052
pubmed: 30375411
Mult Scler. 2020 Sep;26(10):1217-1226
pubmed: 31190607
P T. 2014 May;39(5):356-64
pubmed: 24883008
J Magn Reson Imaging. 2000 Sep;12(3):510
pubmed: 10992321
J Digit Imaging. 2012 Jun;25(3):352-8
pubmed: 21858592
J Neurosurg Pediatr. 2008 Sep;2(3):222-8
pubmed: 18759607
Med Image Anal. 2019 Feb;52:128-143
pubmed: 30579222
Int J Med Inform. 2007 Jan;76(1):22-33
pubmed: 16478675
J Magn Reson Imaging. 2019 Oct;50(4):1260-1267
pubmed: 30811739
World Neurosurg. 2016 Apr;88:297-299
pubmed: 26455768