1
|
Chen Z, Chen J, Zhao J, Liu B, Jiang S, Si D, Ding H, Nian Y, Yang X, Xiao J. What Matters in Radiological Image Segmentation? Effect of Segmentation Errors on the Diagnostic Related Features. J Digit Imaging 2023; 36:2088-2099. [PMID: 37340195 PMCID: PMC10501981 DOI: 10.1007/s10278-023-00865-2] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/03/2023] [Revised: 05/29/2023] [Accepted: 05/31/2023] [Indexed: 06/22/2023] Open
Abstract
Segmentation is a crucial step in extracting the medical image features for clinical diagnosis. Though multiple metrics have been proposed to evaluate the segmentation performance, there is no clear study on how or to what extent the segmentation errors will affect the diagnostic related features used in clinical practice. Therefore, we proposed a segmentation robustness plot (SRP) to build the link between segmentation errors and clinical acceptance, where relative area under the curve (R-AUC) was designed to help clinicians to identify the robust diagnostic related image features. In experiments, we first selected representative radiological series from time series (cardiac first-pass perfusion) and spatial series (T2 weighted images on brain tumors) of magnetic resonance images, respectively. Then, dice similarity coefficient (DSC) and Hausdorff distance (HD), as the widely used evaluation metrics, were used to systematically control the degree of the segmentation errors. Finally, the differences between diagnostic related image features extracted from the ground truth and the derived segmentation were analyzed, using the statistical method large sample size T-test to calculate the corresponding p values. The results are denoted in the SRP, where the x-axis indicates the segmentation performance using the aforementioned evaluation metric, and the y-axis shows the severity of the corresponding feature changes, which are expressed in either the p values for a single case or the proportion of patients without significant change. The experimental results in SRP show that when DSC is above 0.95 and HD is below 3 mm, the segmentation errors will not change the features significantly in most cases. However, when segmentation gets worse, additional metrics are required for further analysis. In this way, the proposed SRP indicates the impact of the segmentation errors on the severity of the corresponding feature changes. By using SRP, one could easily define the acceptable segmentation errors in a challenge. Additionally, the R-AUC calculated from SRP provides an objective reference to help the selection of reliable features in image analysis.
Collapse
Affiliation(s)
- Zihang Chen
- Bioengineering College, Chongqing University, Chongqing, China
| | - Jiafei Chen
- The department of radiology, Southwest Hospital, Chongqing, China
| | - Jun Zhao
- The department of radiology, Southwest Hospital, Chongqing, China
| | - Bowei Liu
- Center for Biomedical Imaging Research, Tsinghua University, Beijing, China
| | - Shuanglong Jiang
- Bio-Med Informatics Research Center & Clinical Research Center, The Second Affiliated Hospital, Army Medical University, Chongqing, China
| | - Dongyue Si
- Center for Biomedical Imaging Research, Tsinghua University, Beijing, China
| | - Haiyan Ding
- Center for Biomedical Imaging Research, Tsinghua University, Beijing, China
| | - Yongjian Nian
- School of Biomedical Engineering, Third Military Medical University, Chongqing, China
| | - Xiaochao Yang
- School of Biomedical Engineering, Third Military Medical University, Chongqing, China
| | - Jingjing Xiao
- Bio-Med Informatics Research Center & Clinical Research Center, The Second Affiliated Hospital, Army Medical University, Chongqing, China
| |
Collapse
|
2
|
Helmy S, Karim S. Multimodality imaging in aortic stenosis. Heart Views 2022; 23:22-32. [PMID: 35757450 PMCID: PMC9231538 DOI: 10.4103/heartviews.heartviews_32_22] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/28/2022] [Accepted: 04/03/2022] [Indexed: 12/02/2022] Open
Abstract
Aortic stenosis (AS) is the most common cardiac valve lesion in the adult population, with an incidence increasing as the population ages. Accurate assessment of AS severity is necessary for clinical decision-making. Echocardiography is currently the diagnostic method of choice for assessing and managing AS. Transthoracic echocardiography is usually sufficient in most situations. Transesophageal echocardiography and stress echocardiography may also be utilized when there is inadequate image quality and/or discordance in the results and the clinical presentation. There is a role for other imaging modalities such as cardiac computed tomography, magnetic resonance imaging, and catheterization in selected cases. The following describes in some detail the role of these modalities in the diagnosis and assessment of AS.
Collapse
|
3
|
Woldendorp K, Doyle MP, Black D, Ng M, Keech A, Grieve SM, Bannon PG. Subclinical valve thrombosis in transcatheter aortic valve implantation: A systematic review and meta-analysis. J Thorac Cardiovasc Surg 2021; 162:1491-1499.e2. [DOI: 10.1016/j.jtcvs.2020.01.084] [Citation(s) in RCA: 15] [Impact Index Per Article: 5.0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 02/13/2019] [Revised: 12/21/2019] [Accepted: 01/16/2020] [Indexed: 11/27/2022]
|
4
|
Woldendorp K, Bannon PG, Grieve SM. Evaluation of aortic stenosis using cardiovascular magnetic resonance: a systematic review & meta-analysis. J Cardiovasc Magn Reson 2020; 22:45. [PMID: 32536342 PMCID: PMC7294634 DOI: 10.1186/s12968-020-00633-z] [Citation(s) in RCA: 14] [Impact Index Per Article: 3.5] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/24/2019] [Accepted: 05/08/2020] [Indexed: 11/22/2022] Open
Abstract
BACKGROUND As the average age of patients with severe aortic stenosis (AS) who receive procedural intervention continue to age, the need for non-invasive modalities that provide accurate diagnosis and operative planning is increasingly important. Advances in cardiovascular magnetic resonance (CMR) over the past two decades mean it is able to provide haemodynamic data at the aortic valve, along with high fidelity anatomical imaging. METHODS Electronic databases were searched for studies comparing CMR to transthoracic echocardiography (TTE) and transoesophageal echocardiography (TEE) in the diagnosis of AS. Studies were included only if direct comparison was made on matched patients, and if diagnosis was primarily through measurement of aortic valve area (AVA). RESULTS Twenty-three relevant, prospective articles were included in the meta-analysis, totalling 1040 individual patients. There was no significant difference in AVA measured as by CMR compared to TEE. CMR measurements of AVA size were larger compared to TTE by an average of 10.7% (absolute difference: + 0.14cm2, 95% CI 0.07-0.21, p < 0.001). Reliability was high for both inter- and intra-observer measurements (0.03cm2 +/- 0.04 and 0.02cm2 +/- 0.01, respectively). CONCLUSIONS Our analysis demonstrates the equivalence of AVA measurements using CMR compared to those obtained using TEE. CMR demonstrated a small but significantly larger AVA than TTE. However, this can be attributed to known errors in derivation of left ventricular outflow tract size as measured by TTE. By offering additional anatomical assessment, CMR is warranted as a primary tool in the assessment and workup of patients with severe AS who are candidates for surgical or transcatheter intervention.
Collapse
Affiliation(s)
- Kei Woldendorp
- Sydney Translational Imaging Laboratory, Imaging and Phenotyping Laboratory, Charles Perkins Centre, Faculty of Medicine and Health, The University of Sydney, Camperdown, NSW 2006 Australia
- Sydney Medical School, The University of Sydney, Camperdown, NSW 2050 Australia
- Baird Institute of Applied Heart & Lung Surgical Research, Newtown, NSW 2042 Australia
- Department of Cardiothoracic Surgery, Royal Prince Alfred Hospital, Camperdown, NSW 2006 Australia
| | - Paul G. Bannon
- Sydney Translational Imaging Laboratory, Imaging and Phenotyping Laboratory, Charles Perkins Centre, Faculty of Medicine and Health, The University of Sydney, Camperdown, NSW 2006 Australia
- Sydney Medical School, The University of Sydney, Camperdown, NSW 2050 Australia
- Baird Institute of Applied Heart & Lung Surgical Research, Newtown, NSW 2042 Australia
| | - Stuart M. Grieve
- Sydney Translational Imaging Laboratory, Imaging and Phenotyping Laboratory, Charles Perkins Centre, Faculty of Medicine and Health, The University of Sydney, Camperdown, NSW 2006 Australia
- Sydney Medical School, The University of Sydney, Camperdown, NSW 2050 Australia
- Department of Radiology, Royal Prince Alfred Hospital, Camperdown, NSW 2006 Australia
| |
Collapse
|
5
|
Future perspectives of nanoparticle-based contrast agents for cardiac magnetic resonance in myocardial infarction. NANOMEDICINE-NANOTECHNOLOGY BIOLOGY AND MEDICINE 2019; 17:329-341. [PMID: 30802547 DOI: 10.1016/j.nano.2019.02.003] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Received: 06/15/2018] [Revised: 01/11/2019] [Accepted: 02/02/2019] [Indexed: 12/23/2022]
Abstract
Cardiac Magnetic Resonance (CMR), thanks to high spatial resolution and absence of ionizing radiation, has been widely used in myocardial infarction (MI) assessment to evaluate cardiac structure, function, perfusion and viability. Nevertheless, it suffers from limitations in tissue and assessment of myocardial pathophysiological changes subsequent to MI. In this issue, nanoparticle-based contrast agents offer the possibility to track biological processes at cellular and molecular level underlying the various phases of MI, infarct healing and tissue repair. In this paper, first we examine the conventional CMR protocol and its findings in MI patients. Next, we looked at how nanoparticles can help in the imaging of MI and give an overview of the major approaches currently explored. Based on the presentation of successful nanoparticle applications as contrast agents (CAs) in preclinical and clinical models, we discuss promises and outstanding challenges facing the field of CMR in MI, their translational potential and clinical application.
Collapse
|
6
|
Kozor R, Grieve SM, Buchholz S, Kaye S, Darke S, Bhindi R, Figtree GA. Regular cocaine use is associated with increased systolic blood pressure, aortic stiffness and left ventricular mass in young otherwise healthy individuals. PLoS One 2014; 9:e89710. [PMID: 24717541 PMCID: PMC3981670 DOI: 10.1371/journal.pone.0089710] [Citation(s) in RCA: 31] [Impact Index Per Article: 3.1] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/15/2013] [Accepted: 01/22/2014] [Indexed: 12/31/2022] Open
Abstract
BACKGROUND The cardiovascular impact of cocaine use in otherwise healthy individuals who consider themselves 'social' users is not well established. METHODS/RESULTS Twenty regular cocaine users and 20 control subjects were recruited by word-of-mouth. Cardiovascular magnetic resonance was performed to assess cardiac and vascular structure and function. Cocaine users had higher systolic blood pressure compared to non-users (134±11 vs 126±11 mmHg, p = 0.036), a finding independent of age, body surface area, smoking and alcohol consumption. Cocaine use was associated with increased arterial stiffness - reflected by reduced aortic compliance (1.3±0.2 vs 1.7±0.5 cm2×10-2.mmHg-1, p = 0.004), decreased distensibility (3.8±0.9 vs 5.1±1.4 mmHg-1.10-3, p = 0.001), increased stiffness index (2.6±0.6 vs 2.1±0.6, p = 0.005), and higher pulse wave velocity (5.1±0.6 vs 4.4±0.6 m.s-1, p = 0.001). This change in aortic stiffness was independent of vessel wall thickness. Left ventricular mass was 18% higher in cocaine users (124±25 vs 105±16 g, p = 0.01), a finding that was independent of body surface area, and left atrial diameter was larger in the user group than controls (3.8±0.6 vs 3.5±0.3 cm, p = 0.04). The increased left ventricular mass, systolic blood pressure and vascular stiffness measures were all associated with duration and/or frequency of cocaine use. No late gadolinium enhancement or segmental wall motion abnormalities were seen in any of the subjects. CONCLUSIONS Compared with the non-user control cohort, cocaine users had increased aortic stiffness and systolic blood pressure, associated with greater left ventricular mass. These measures are all well known risk factors for premature cardiovascular events, highlighting the dangers of cocaine use, even in a 'social' setting, and have important public health implications.
Collapse
Affiliation(s)
- Rebecca Kozor
- North Shore Heart Research Group, Kolling Institute of Medical Research, University of Sydney, Sydney, Australia
- Department of Cardiology, Royal North Shore Hospital, Sydney, Australia
| | - Stuart M. Grieve
- North Shore Heart Research Group, Kolling Institute of Medical Research, University of Sydney, Sydney, Australia
- Sydney Translational Imaging Laboratory, Sydney Medical School, University of Sydney, Sydney, Australia
| | - Stefan Buchholz
- Department of Cardiology, Royal North Shore Hospital, Sydney, Australia
| | - Sharlene Kaye
- National Drug and Alcohol Research Centre, University of New South Wales, Sydney, Australia
| | - Shane Darke
- National Drug and Alcohol Research Centre, University of New South Wales, Sydney, Australia
| | - Ravinay Bhindi
- North Shore Heart Research Group, Kolling Institute of Medical Research, University of Sydney, Sydney, Australia
- Department of Cardiology, Royal North Shore Hospital, Sydney, Australia
| | - Gemma A. Figtree
- North Shore Heart Research Group, Kolling Institute of Medical Research, University of Sydney, Sydney, Australia
- Department of Cardiology, Royal North Shore Hospital, Sydney, Australia
| |
Collapse
|
7
|
In the search of coronary calcium. Int J Cardiol 2013; 167:310-7. [DOI: 10.1016/j.ijcard.2012.06.051] [Citation(s) in RCA: 6] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 01/05/2012] [Revised: 04/23/2012] [Accepted: 06/09/2012] [Indexed: 01/07/2023]
|
8
|
Lønborg J, Ward M, Gill A, Grieve SM, Figtree GA. Utility of cardiac magnetic resonance in assessing right-sided heart failure in sarcoidosis. BMC Med Imaging 2013; 13:2. [PMID: 23311586 PMCID: PMC3600993 DOI: 10.1186/1471-2342-13-2] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/22/2012] [Accepted: 12/31/2012] [Indexed: 11/10/2022] Open
Abstract
BACKGROUND Cardiac involvement in sarcoidosis is associated with a poor prognosis. In patients with right sided heart failure, differentiating between cor-pulmonale, or cardiac sarcoidosis has important implications to management. CASE PRESENTATION We present the case of a patient with severe but stable pulmonary sarcoidosis and new onset right sided heart failure despite only mild elevations of pulmonary artery pressure. CMR demonstration of extensive right ventricular fibrosis with associated dilatation and hypokinesis was a key finding for prognosis and management of the patient. CONCLUSION Cardiac magnetic resonance (CMR) is the preferred investigation in the diagnosis of cardiac sarcoidosis, allowing assessment of myocardial inflammation and fibrosis, as well as function, in a manner not matched by other technologies.
Collapse
Affiliation(s)
- Jacob Lønborg
- North Shore Heart Research Group, Kolling Institute of Medical Research, University of Sydney, Sydney, Australia
| | | | | | | | | |
Collapse
|