Login
Search
Search
0 Dates
2024
2023
2022
2021
2020
2019
2018
0 Events
CPC 2018
CPC 2019
Curso de Atualização em Medicina Cardiovascular 2019
Reunião Anual Conjunta dos Grupos de Estudo de Cirurgia Cardíaca, Doenças Valvulares e Ecocardiografia da SPC
CPC 2020
CPC 2021
CPC 2022
CPC 2023
CPC 2024
0 Topics
A. Basics
B. Imaging
C. Arrhythmias and Device Therapy
D. Heart Failure
E. Coronary Artery Disease, Acute Coronary Syndromes, Acute Cardiac Care
F. Valvular, Myocardial, Pericardial, Pulmonary, Congenital Heart Disease
G. Aortic Disease, Peripheral Vascular Disease, Stroke
H. Interventional Cardiology and Cardiovascular Surgery
I. Hypertension
J. Preventive Cardiology
K. Cardiovascular Disease In Special Populations
L. Cardiovascular Pharmacology
M. Cardiovascular Nursing
N. E-Cardiology / Digital Health, Public Health, Health Economics, Research Methodology
O. Basic Science
P. Other
0 Themes
01. History of Cardiology
02. Clinical Skills
03. Imaging
04. Arrhythmias, General
05. Atrial Fibrillation
06. Supraventricular Tachycardia (non-AF)
07. Syncope and Bradycardia
08. Ventricular Arrhythmias and Sudden Cardiac Death (SCD)
09. Device Therapy
10. Chronic Heart Failure
11. Acute Heart Failure
12. Coronary Artery Disease (Chronic)
13. Acute Coronary Syndromes
14. Acute Cardiac Care
15. Valvular Heart Disease
16. Infective Endocarditis
17. Myocardial Disease
18. Pericardial Disease
19. Tumors of the Heart
20. Congenital Heart Disease and Pediatric Cardiology
21. Pulmonary Circulation, Pulmonary Embolism, Right Heart Failure
22. Aortic Disease
23. Peripheral Vascular and Cerebrovascular Disease
24. Stroke
25. Interventional Cardiology
26. Cardiovascular Surgery
27. Hypertension
28. Risk Factors and Prevention
29. Rehabilitation and Sports Cardiology
30. Cardiovascular Disease in Special Populations
31. Pharmacology and Pharmacotherapy
32. Cardiovascular Nursing
33. e-Cardiology / Digital Health
34. Public Health and Health Economics
35. Research Methodology
36. Basic Science
37. Miscellanea
0 Resources
Abstract
Slides
Vídeo
Report
CLEAR FILTERS
Automatic Interpretation of Point-of-Care Lung Ultrasound
Session:
Posters (Sessão 4 - Écran 3) - Imagem multimodal 1
Speaker:
Bárbara Malainho
Congress:
CPC 2023
Topic:
B. Imaging
Theme:
03. Imaging
Subtheme:
03.6 Cross-Modality and Multi-Modality Imaging Topics
Session Type:
Pósters Electrónicos
FP Number:
---
Authors:
Bárbara Malainho; Catarina Rodrigues; Ana Cláudia Tonelli; André Santanchè; Marco A. Carvalho-Filho; Nuno Sousa; Jaime C. Fonseca; Vítor Hugo Pereira; Sandro Queirós
Abstract
<p style="text-align:justify"><span style="font-size:12pt"><span style="font-family:Calibri,sans-serif"><strong>Introduction:</strong> Point-of-care ultrasound (POCUS) is a safe, portable, and low-cost imaging technique useful for a fast bedside patient examination. Currently, with the COVID-19 pandemic, the necessity for an expeditious image interpretation and associated diagnosis has become clearer than ever. Coincidentally, deep learning-based solutions have increased their presence in the medical imaging field. Notwithstanding their potential, the usage of these techniques in lung POCUS remains underexplored.</span></span></p> <p style="text-align:justify"><span style="font-size:12pt"><span style="font-family:Calibri,sans-serif"><strong>Objectives:</strong> To develop a flexible deep learning (DL) framework for the interpretation of lung ultrasound (LUS) and identification of the most common findings in clinical practice: scattering, A-lines, up to 3 B-lines, positive B-lines pattern and other pathologies (including pleural effusion and consolidations). These labels can also be aggregated into two super-classes: normal findings and findings indicative of pathology. </span></span></p> <p style="text-align:justify"><span style="font-size:12pt"><span style="font-family:Calibri,sans-serif"><strong>Methods:</strong> The dataset is composed by 3649 annotated lung ultrasound videos, with nearly equal proportions between normal and indicative findings, and 4511 videos without annotations. Since data is scarce but a core necessity to train a successful DL model, two learning strategies are investigated: supervised and semi-supervised scenarios. The work culminates with the proposal of a novel model ensembling strategy, which aggregates the outputs of models trained to predict distinct label sets, and an optional dataset-specific post-processing routine, both aimed at leveraging of the hierarchy inherent to LUS interpretation. The proposed framework and its building blocks were evaluated in an extensive set of experiments, considering both multi-class and multi-label models, for both supervised and semi-supervised settings. </span></span></p> <p style="text-align:justify"><span style="font-size:12pt"><span style="font-family:Calibri,sans-serif"><strong>Results:</strong> Our experiments show the framework’s versatility, allowing for a custom combination of the multiple proposed blocks according to the task in question. In a held-out test set, the categorical proposal, which is useful for an expedite triage, achieved an average F1-score of 92.61%, while the multi-label proposal, helpful for patient management and referral, achieved an average F1-score of 70.45% when considering five relevant LUS findings. </span></span></p> <p style="text-align:justify"><span style="font-size:12pt"><span style="font-family:Calibri,sans-serif"><strong>Conclusion:</strong> Overall, the proposal shows promise in an underexplored field, paving the way for an accurate computer-assisted lung ultrasound interpretation in clinical practice.</span></span></p>
Slides
Our mission: To reduce the burden of cardiovascular disease
Visit our site