1
|
Shikhare S, Balki I, Shi Y, Kavanagh J, Donahoe L, Xu W, Rozenberg D, de Perrot M, McInnis M. Right-to-left ventricle ratio determined by machine learning algorithms on CT pulmonary angiography images predicts prolonged ICU length of stay in operated chronic thromboembolic pulmonary hypertension. Br J Radiol 2022; 95:20210722. [PMID: 36043477 PMCID: PMC9793468 DOI: 10.1259/bjr.20210722] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/13/2021] [Revised: 05/06/2022] [Accepted: 08/13/2022] [Indexed: 01/09/2023] Open
Abstract
OBJECTIVE Right-to-left ventricle diameter ratio (dRV/dLV) on CT pulmonary angiography (CTPA) is a predictor of outcomes in non-operated chronic thromboembolic pulmonary hypertension (CTEPH) patients. The purpose of this study is to evaluate the performance of a novel machine learning (ML) algorithm for dRV/dLV measurement in operated CTEPH patients and its association with post-operative outcomes. METHODS This retrospective study reviewed consecutive CTEPH patients who underwent pulmonary endarterectomy between 2013 and 2017. ML calculated dRV/dLV on pre-operative CTPA and compared with manual measures. Associations of dRV/dLV with patient characteristics and post-operative outcomes were evaluated including intensive care (ICU) and hospital length of stay (LOS) using multivariable linear regression analysis. Prolonged LOS was defined as greater than median. RESULTS ML segmented the ventricles in 99/125 (79%) patients. The most common cause of failure was misidentification of the moderator band as the interventricular septum (7.9%). Mean dRV/dLV by ML was 1.4 ± 0.4 and strongly correlated with manual measures (r = 0.9-0.96 p < 0.0001). dRV/dLV was moderately correlated with measures of pulmonary hypertension on right heart catheterization and RV dilatation on echocardiogram (r = 0.5-0.6, p < 0.0001). dRV/dLV ≥ 1.2 was associated with proximal Jamieson type disease (p = 0.032), longer cardiopulmonary bypass (p = 0.037), aortic cross-clamp (p = 0.022) and circulatory arrest (p < 0.001) at surgery and dRV/dLV ≥ 1.6 with post-operative ECMO (p = 0.006). dRV/dLV was independently associated with prolonged ICU LOS (OR = 3.79, 95% CI 1.1-13.06, p = 0.035). CONCLUSION dRV/dLV was associated with CTEPH severity and independently associated with prolonged ICU LOS. This CT parameter may therefore assist in perioperative planning. Further refinement of the ML algorithm or CTPA technique is required to avoid errors in ventricular segmentation. ADVANCES IN KNOWLEDGE Automated right-to-left ventricle ratio measurement by machine learning is feasible and is independently associated with outcome after pulmonary endarterectomy.
Collapse
Affiliation(s)
| | - Indranil Balki
- Department of Medicine, University of Toronto, Toronto, Ontario, Canada
| | - Yuliang Shi
- Department of Biostatistics, Princess Margaret Cancer Center, Toronto, Canada
| | - John Kavanagh
- Division of Cardiothoracic Imaging, Joint Department of Medical Imaging, University of Toronto, Toronto, Canada
| | - Laura Donahoe
- Division of Thoracic Surgery, Department of Surgery, University of Toronto, Toronto General Hospital, Toronto, Canada
| | - Wei Xu
- Department of Biostatistics, Princess Margaret Cancer Center, Toronto, Canada
| | - Dmitry Rozenberg
- Division of Respirology, Department of Medicine, University of Toronto, Toronto General Hospital, Toronto, Canada
| | - Marc de Perrot
- Division of Thoracic Surgery, Department of Surgery, University of Toronto, Toronto General Hospital, Toronto, Canada
| | - Micheal McInnis
- Division of Cardiothoracic Imaging, Joint Department of Medical Imaging, University of Toronto, Toronto, Canada
| |
Collapse
|
2
|
Sharkey MJ, Taylor JC, Alabed S, Dwivedi K, Karunasaagarar K, Johns CS, Rajaram S, Garg P, Alkhanfar D, Metherall P, O'Regan DP, van der Geest RJ, Condliffe R, Kiely DG, Mamalakis M, Swift AJ. Fully automatic cardiac four chamber and great vessel segmentation on CT pulmonary angiography using deep learning. Front Cardiovasc Med 2022; 9:983859. [PMID: 36225963 PMCID: PMC9549370 DOI: 10.3389/fcvm.2022.983859] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/01/2022] [Accepted: 09/02/2022] [Indexed: 11/13/2022] Open
Abstract
Introduction Computed tomography pulmonary angiography (CTPA) is an essential test in the work-up of suspected pulmonary vascular disease including pulmonary hypertension and pulmonary embolism. Cardiac and great vessel assessments on CTPA are based on visual assessment and manual measurements which are known to have poor reproducibility. The primary aim of this study was to develop an automated whole heart segmentation (four chamber and great vessels) model for CTPA. Methods A nine structure semantic segmentation model of the heart and great vessels was developed using 200 patients (80/20/100 training/validation/internal testing) with testing in 20 external patients. Ground truth segmentations were performed by consultant cardiothoracic radiologists. Failure analysis was conducted in 1,333 patients with mixed pulmonary vascular disease. Segmentation was achieved using deep learning via a convolutional neural network. Volumetric imaging biomarkers were correlated with invasive haemodynamics in the test cohort. Results Dice similarity coefficients (DSC) for segmented structures were in the range 0.58-0.93 for both the internal and external test cohorts. The left and right ventricle myocardium segmentations had lower DSC of 0.83 and 0.58 respectively while all other structures had DSC >0.89 in the internal test cohort and >0.87 in the external test cohort. Interobserver comparison found that the left and right ventricle myocardium segmentations showed the most variation between observers: mean DSC (range) of 0.795 (0.785-0.801) and 0.520 (0.482-0.542) respectively. Right ventricle myocardial volume had strong correlation with mean pulmonary artery pressure (Spearman's correlation coefficient = 0.7). The volume of segmented cardiac structures by deep learning had higher or equivalent correlation with invasive haemodynamics than by manual segmentations. The model demonstrated good generalisability to different vendors and hospitals with similar performance in the external test cohort. The failure rates in mixed pulmonary vascular disease were low (<3.9%) indicating good generalisability of the model to different diseases. Conclusion Fully automated segmentation of the four cardiac chambers and great vessels has been achieved in CTPA with high accuracy and low rates of failure. DL volumetric biomarkers can potentially improve CTPA cardiac assessment and invasive haemodynamic prediction.
Collapse
Affiliation(s)
- Michael J Sharkey
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,3D Imaging Lab, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Jonathan C Taylor
- 3D Imaging Lab, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Samer Alabed
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom
| | - Krit Dwivedi
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Insigneo Institute for in Silico Medicine, University of Sheffield, Sheffield, United Kingdom
| | - Kavitasagary Karunasaagarar
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Radiology Department, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Christopher S Johns
- Radiology Department, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Smitha Rajaram
- Radiology Department, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Pankaj Garg
- Norwich Medical School, University of East Anglia, Norwich, United Kingdom
| | - Dheyaa Alkhanfar
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom
| | - Peter Metherall
- 3D Imaging Lab, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom
| | - Declan P O'Regan
- MRC London Institute of Medical Sciences, Imperial College London, London, United Kingdom
| | - Rob J van der Geest
- Department of Radiology, Leiden University Medical Center, Leiden, Netherlands
| | - Robin Condliffe
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Sheffield Pulmonary Vascular Disease Unit, Sheffield Teaching Hospitals NHS Trust, Sheffield, United Kingdom
| | - David G Kiely
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Insigneo Institute for in Silico Medicine, University of Sheffield, Sheffield, United Kingdom.,Sheffield Pulmonary Vascular Disease Unit, Sheffield Teaching Hospitals NHS Trust, Sheffield, United Kingdom
| | - Michail Mamalakis
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Insigneo Institute for in Silico Medicine, University of Sheffield, Sheffield, United Kingdom.,Department of Computer Science, University of Sheffield, Sheffield, United Kingdom
| | - Andrew J Swift
- Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.,Insigneo Institute for in Silico Medicine, University of Sheffield, Sheffield, United Kingdom
| |
Collapse
|
3
|
Automated calculation of the right ventricle to left ventricle ratio on CT for the risk stratification of patients with acute pulmonary embolism. Eur Radiol 2021; 31:6013-6020. [PMID: 33459854 DOI: 10.1007/s00330-020-07605-y] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/28/2020] [Revised: 11/22/2020] [Accepted: 12/04/2020] [Indexed: 12/11/2022]
Abstract
OBJECTIVES To assess the feasibility and reliability of the use of artificial intelligence post-processing to calculate the RV:LV diameter ratio on computed tomography pulmonary angiography (CTPA) and to investigate its prognostic value in patients with acute PE. METHODS Single-centre, retrospective study of 101 consecutive patients with CTPA-proven acute PE. RV and LV volumes were segmented on 1-mm contrast-enhanced axial slices and maximal ventricular diameters were derived for RV:LV ratio using automated post-processing software (IMBIO LLC, USA) and compared to manual analysis in two observers, via intraclass coefficient correlation analysis. Each CTPA report was analysed for mention of the RV:LV ratio and compared to the automated RV:LV ratio. Thirty-day all-cause mortality post-CTPA was recorded. RESULTS Automated RV:LV analysis was feasible in 87% (n = 88). RV:LV ratios ranged from 0.67 to 2.43, with 64% (n = 65) > 1.0. There was very strong agreement between manual and automated RV:LV ratios (ICC = 0.83, 0.77-0.88). The use of automated analysis led to a change in risk stratification in 45% of patients (n = 40). The AUC of the automated measurement for the prediction of all-cause 30-day mortality was 0.77 (95% CI: 0.62-0.99). CONCLUSION The RV:LV ratio on CTPA can be reliably measured automatically in the majority of real-world cases of acute PE, with perfect reproducibility. The routine use of this automated analysis in clinical practice would add important prognostic information in patients with acute PE. KEY POINTS • Automated calculation of the right ventricle to left ventricle ratio was feasible in the majority of patients and demonstrated perfect intraobserver variability. • Automated analysis would have added important prognostic information and altered risk stratification in the majority of patients. • The optimal cut-off value for the automated right ventricle to left ventricle ratio was 1.18, with a sensitivity of 100% and specificity of 54% for the prediction of 30-day mortality.
Collapse
|
4
|
Medic G, Kosaner Kließ M, Atallah L, Weichert J, Panda S, Postma M, EL-Kerdi A. Evidence-based Clinical Decision Support Systems for the prediction and detection of three disease states in critical care: A systematic literature review. F1000Res 2019; 8:1728. [PMID: 31824670 PMCID: PMC6894361 DOI: 10.12688/f1000research.20498.1] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Accepted: 09/24/2019] [Indexed: 10/21/2023] Open
Abstract
Background: Clinical decision support (CDS) systems have emerged as tools providing intelligent decision making to address challenges of critical care. CDS systems can be based on existing guidelines or best practices; and can also utilize machine learning to provide a diagnosis, recommendation, or therapy course. Methods: This research aimed to identify evidence-based study designs and outcome measures to determine the clinical effectiveness of clinical decision support systems in the detection and prediction of hemodynamic instability, respiratory distress, and infection within critical care settings. PubMed, ClinicalTrials.gov and Cochrane Database of Systematic Reviews were systematically searched to identify primary research published in English between 2013 and 2018. Studies conducted in the USA, Canada, UK, Germany and France with more than 10 participants per arm were included. Results: In studies on hemodynamic instability, the prediction and management of septic shock were the most researched topics followed by the early prediction of heart failure. For respiratory distress, the most popular topics were pneumonia detection and prediction followed by pulmonary embolisms. Given the importance of imaging and clinical notes, this area combined Machine Learning with image analysis and natural language processing. In studies on infection, the most researched areas were the detection, prediction, and management of sepsis, surgical site infections, as well as acute kidney injury. Overall, a variety of Machine Learning algorithms were utilized frequently, particularly support vector machines, boosting techniques, random forest classifiers and neural networks. Sensitivity, specificity, and ROC AUC were the most frequently reported performance measures. Conclusion: This review showed an increasing use of Machine Learning for CDS in all three areas. Large datasets are required for training these algorithms; making it imperative to appropriately address, challenges such as class imbalance, correct labelling of data and missing data. Recommendations are formulated for the development and successful adoption of CDS systems.
Collapse
Affiliation(s)
- Goran Medic
- Health Economics, Philips, Eindhoven, Noord-Brabant, 5621JG, The Netherlands
- Department of Pharmacy, Unit of PharmacoTherapy, -Epidemiology & -Economics, University of Groningen, Groningen, 9700 AB, The Netherlands
| | | | | | | | - Saswat Panda
- Global Market Access Solutions Sàrl, St-Prex, 1162, Switzerland
| | - Maarten Postma
- Department of Pharmacy, Unit of PharmacoTherapy, -Epidemiology & -Economics, University of Groningen, Groningen, 9700 AB, The Netherlands
- Department of Health Sciences, University Medical Centre Groningen, University of Groningen, Groningen, 9700 AB, The Netherlands
- Department of Economics, Econometrics & Finance, University of Groningen, Groningen, 9700 AB, The Netherlands
| | | |
Collapse
|
5
|
Medic G, Kosaner Kließ M, Atallah L, Weichert J, Panda S, Postma M, EL-Kerdi A. Evidence-based Clinical Decision Support Systems for the prediction and detection of three disease states in critical care: A systematic literature review. F1000Res 2019; 8:1728. [PMID: 31824670 PMCID: PMC6894361 DOI: 10.12688/f1000research.20498.2] [Citation(s) in RCA: 14] [Impact Index Per Article: 2.8] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Accepted: 11/22/2019] [Indexed: 02/01/2023] Open
Abstract
Background: Clinical decision support (CDS) systems have emerged as tools providing intelligent decision making to address challenges of critical care. CDS systems can be based on existing guidelines or best practices; and can also utilize machine learning to provide a diagnosis, recommendation, or therapy course. Methods: This research aimed to identify evidence-based study designs and outcome measures to determine the clinical effectiveness of clinical decision support systems in the detection and prediction of hemodynamic instability, respiratory distress, and infection within critical care settings. PubMed, ClinicalTrials.gov and Cochrane Database of Systematic Reviews were systematically searched to identify primary research published in English between 2013 and 2018. Studies conducted in the USA, Canada, UK, Germany and France with more than 10 participants per arm were included. Results: In studies on hemodynamic instability, the prediction and management of septic shock were the most researched topics followed by the early prediction of heart failure. For respiratory distress, the most popular topics were pneumonia detection and prediction followed by pulmonary embolisms. Given the importance of imaging and clinical notes, this area combined Machine Learning with image analysis and natural language processing. In studies on infection, the most researched areas were the detection, prediction, and management of sepsis, surgical site infections, as well as acute kidney injury. Overall, a variety of Machine Learning algorithms were utilized frequently, particularly support vector machines, boosting techniques, random forest classifiers and neural networks. Sensitivity, specificity, and ROC AUC were the most frequently reported performance measures. Conclusion: This review showed an increasing use of Machine Learning for CDS in all three areas. Large datasets are required for training these algorithms; making it imperative to appropriately address, challenges such as class imbalance, correct labelling of data and missing data. Recommendations are formulated for the development and successful adoption of CDS systems.
Collapse
Affiliation(s)
- Goran Medic
- Health Economics, Philips, Eindhoven, Noord-Brabant, 5621JG, The Netherlands
- Department of Pharmacy, Unit of PharmacoTherapy, -Epidemiology & -Economics, University of Groningen, Groningen, 9700 AB, The Netherlands
| | | | | | | | - Saswat Panda
- Global Market Access Solutions Sàrl, St-Prex, 1162, Switzerland
| | - Maarten Postma
- Department of Pharmacy, Unit of PharmacoTherapy, -Epidemiology & -Economics, University of Groningen, Groningen, 9700 AB, The Netherlands
- Department of Health Sciences, University Medical Centre Groningen, University of Groningen, Groningen, 9700 AB, The Netherlands
- Department of Economics, Econometrics & Finance, University of Groningen, Groningen, 9700 AB, The Netherlands
| | | |
Collapse
|
6
|
Rahaghi FN, Minhas JK, Heresi GA. Diagnosis of Deep Venous Thrombosis and Pulmonary Embolism: New Imaging Tools and Modalities. Clin Chest Med 2018; 39:493-504. [PMID: 30122174 PMCID: PMC6317734 DOI: 10.1016/j.ccm.2018.04.003] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/11/2022]
Abstract
Imaging continues to be the modality of choice for the diagnosis of venous thromboembolic disease, particularly when incorporated into diagnostic algorithms. Improvement in imaging techniques as well as new imaging modalities and processing methods have improved diagnostic accuracy and additionally are being leveraged in prognostication and decision making for choice of intervention. In this article, we review the role of imaging in diagnosis and prognostication of venous thromboembolism. We also discuss emerging imaging approaches that may in the near future find clinical usefulness in improving diagnosis and prognostication as well as differentiating disease phenotypes.
Collapse
Affiliation(s)
- Farbod N. Rahaghi
- Pulmonary and Critical Care Medicine, Brigham and Women’s Hospital, Harvard Medical School. 15 Francis Street, Boston MA 02115, ; Phone: 617-632-6770
| | - Jasleen K. Minhas
- Department of Medicine, North Shore Medical Center, 81 highland Ave Salem MA 10970, Phone: 978-354-4801
| | - Gustavo A. Heresi
- Respiratory Institute, Cleveland Clinic, Mail code A90, 9500 Euclid Ave, OH 44195, Phone: 216-636-5327
| |
Collapse
|