1
|
Paciorek AM, von Schacky CE, Foreman SC, Gassert FG, Gassert FT, Kirschke JS, Laugwitz KL, Geith T, Hadamitzky M, Nadjiri J. Automated assessment of cardiac pathologies on cardiac MRI using T1-mapping and late gadolinium phase sensitive inversion recovery sequences with deep learning. BMC Med Imaging 2024; 24:43. [PMID: 38350900 PMCID: PMC10865672 DOI: 10.1186/s12880-024-01217-4] [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: 10/17/2023] [Accepted: 01/30/2024] [Indexed: 02/15/2024] Open
Abstract
BACKGROUND A deep learning (DL) model that automatically detects cardiac pathologies on cardiac MRI may help streamline the diagnostic workflow. To develop a DL model to detect cardiac pathologies on cardiac MRI T1-mapping and late gadolinium phase sensitive inversion recovery (PSIR) sequences were used. METHODS Subjects in this study were either diagnosed with cardiac pathology (n = 137) including acute and chronic myocardial infarction, myocarditis, dilated cardiomyopathy, and hypertrophic cardiomyopathy or classified as normal (n = 63). Cardiac MR imaging included T1-mapping and PSIR sequences. Subjects were split 65/15/20% for training, validation, and hold-out testing. The DL models were based on an ImageNet pretrained DenseNet-161 and implemented using PyTorch and fastai. Data augmentation with random rotation and mixup was applied. Categorical cross entropy was used as the loss function with a cyclic learning rate (1e-3). DL models for both sequences were developed separately using similar training parameters. The final model was chosen based on its performance on the validation set. Gradient-weighted class activation maps (Grad-CAMs) visualized the decision-making process of the DL model. RESULTS The DL model achieved a sensitivity, specificity, and accuracy of 100%, 38%, and 88% on PSIR images and 78%, 54%, and 70% on T1-mapping images. Grad-CAMs demonstrated that the DL model focused its attention on myocardium and cardiac pathology when evaluating MR images. CONCLUSIONS The developed DL models were able to reliably detect cardiac pathologies on cardiac MR images. The diagnostic performance of T1 mapping alone is particularly of note since it does not require a contrast agent and can be acquired quickly.
Collapse
Affiliation(s)
- Aleksandra M Paciorek
- Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany.
| | - Claudio E von Schacky
- Department of Diagnostic and Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Sarah C Foreman
- Department of Diagnostic and Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Felix G Gassert
- Department of Diagnostic and Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Florian T Gassert
- Department of Diagnostic and Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Jan S Kirschke
- TUM-Neuroimaging Center, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Karl-Ludwig Laugwitz
- Department of Medicine I, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Tobias Geith
- Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| | - Martin Hadamitzky
- Department of Radiology, German Heart Center Munich, Technical University of Munich, Lazarettstraße 36, 80636, Munich, Germany
| | - Jonathan Nadjiri
- Interventional Radiology, School of Medicine & Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany
| |
Collapse
|
2
|
Jiao R, Zhang Y, Ding L, Xue B, Zhang J, Cai R, Jin C. Learning with limited annotations: A survey on deep semi-supervised learning for medical image segmentation. Comput Biol Med 2024; 169:107840. [PMID: 38157773 DOI: 10.1016/j.compbiomed.2023.107840] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/18/2023] [Revised: 10/30/2023] [Accepted: 12/07/2023] [Indexed: 01/03/2024]
Abstract
Medical image segmentation is a fundamental and critical step in many image-guided clinical approaches. Recent success of deep learning-based segmentation methods usually relies on a large amount of labeled data, which is particularly difficult and costly to obtain, especially in the medical imaging domain where only experts can provide reliable and accurate annotations. Semi-supervised learning has emerged as an appealing strategy and been widely applied to medical image segmentation tasks to train deep models with limited annotations. In this paper, we present a comprehensive review of recently proposed semi-supervised learning methods for medical image segmentation and summarize both the technical novelties and empirical results. Furthermore, we analyze and discuss the limitations and several unsolved problems of existing approaches. We hope this review can inspire the research community to explore solutions to this challenge and further advance the field of medical image segmentation.
Collapse
Affiliation(s)
- Rushi Jiao
- School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China; School of Engineering Medicine, Beihang University, Beijing, 100191, China; Shanghai Artificial Intelligence Laboratory, Shanghai, 200232, China.
| | - Yichi Zhang
- School of Data Science, Fudan University, Shanghai, 200433, China; Artificial Intelligence Innovation and Incubation Institute, Fudan University, Shanghai, 200433, China.
| | - Le Ding
- School of Biological Science and Medical Engineering, Beihang University, Beijing, 100191, China.
| | - Bingsen Xue
- School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China; Shanghai Artificial Intelligence Laboratory, Shanghai, 200232, China.
| | - Jicong Zhang
- School of Biological Science and Medical Engineering, Beihang University, Beijing, 100191, China; Hefei Innovation Research Institute, Beihang University, Hefei, 230012, China.
| | - Rong Cai
- School of Engineering Medicine, Beihang University, Beijing, 100191, China; Key Laboratory for Biomechanics and Mechanobiology of Ministry of Education, Beihang University, Beijing, 100191, China.
| | - Cheng Jin
- School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China; Shanghai Artificial Intelligence Laboratory, Shanghai, 200232, China; Beijing Anding Hospital, Capital Medical University, Beijing, 100088, China.
| |
Collapse
|
3
|
Lecesne E, Simon A, Garreau M, Barone-Rochette G, Fouard C. Segmentation of cardiac infarction in delayed-enhancement MRI using probability map and transformers-based neural networks. COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE 2023; 242:107841. [PMID: 37865006 DOI: 10.1016/j.cmpb.2023.107841] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 02/09/2023] [Revised: 09/15/2023] [Accepted: 10/01/2023] [Indexed: 10/23/2023]
Abstract
BACKGROUND AND OBJECTIVE Automatic segmentation of myocardial infarction is of great clinical interest for the quantitative evaluation of myocardial infarction (MI). Late Gadolinium Enhancement cardiac MRI (LGE-MRI) is commonly used in clinical practice to quantify MI, which is crucial for clinical diagnosis and treatment of cardiac diseases. However, the segmentation of infarcted tissue in LGE-MRI is highly challenging due to its high anisotropy and inhomogeneities. METHODS The innovative aspect of our work lies in the utilization of a probability map of the healthy myocardium to guide the localization of infarction, as well as the combination of 2D U-Net and U-Net transformers to achieve the final segmentation. Instead of employing a binary segmentation map, we propose using a probability map of the normal myocardium, obtained through a dedicated 2D U-Net. To leverage spatial information, we employ a U-Net transformers network where we incorporate the probability map into the original image as an additional input. Then, To address the limitations of U-Net in segmenting accurately the contours, we introduce an adapted loss function. RESULTS Our method has been evaluated on the 2020 MICCAI EMIDEC challenge dataset, yielding competitive results. Specifically, we achieved a Dice score of 92.94% for the myocardium and 92.36% for the infarction. These outcomes highlight the competitiveness of our approach. CONCLUSION In the case of the infarction class, our proposed method outperforms state-of-the-art techniques across all metrics evaluated in the challenge, establishing its superior performance in infarction segmentation. This study further reinforces the importance of integrating a contour loss into the segmentation process.
Collapse
Affiliation(s)
- Erwan Lecesne
- Univ Rennes, Inserm, LTSI - UMR 1099, Rennes, 35000, France.
| | - Antoine Simon
- Univ Rennes, Inserm, LTSI - UMR 1099, Rennes, 35000, France
| | | | - Gilles Barone-Rochette
- Clinic of Cardiology, Cardiovascular and Thoracic Department, University Hospital of Grenoble, Grenoble, 38000, France
| | - Céline Fouard
- Univ. Grenoble Alpes, CNRS, UMR 5525, VetAgro Sup, Grenoble INP, TIMC, Grenoble, 38000, France
| |
Collapse
|
4
|
Duchateau N, Viallon M, Petrusca L, Clarysse P, Mewton N, Belle L, Croisille P. Pixel-wise statistical analysis of myocardial injury in STEMI patients with delayed enhancement MRI. Front Cardiovasc Med 2023; 10:1136760. [PMID: 37396590 PMCID: PMC10313104 DOI: 10.3389/fcvm.2023.1136760] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/03/2023] [Accepted: 06/06/2023] [Indexed: 07/04/2023] Open
Abstract
Objectives Myocardial injury assessment from delayed enhancement magnetic resonance images is routinely limited to global descriptors such as size and transmurality. Statistical tools from computational anatomy can drastically improve this characterization, and refine the assessment of therapeutic procedures aiming at infarct size reduction. Based on these techniques, we propose a new characterization of myocardial injury up to the pixel resolution. We demonstrate it on the imaging data from the Minimalist Immediate Mechanical Intervention randomized clinical trial (MIMI: NCT01360242), which aimed at comparing immediate and delayed stenting in acute ST-Elevation Myocardial Infarction (STEMI) patients. Methods We analyzed 123 patients from the MIMI trial (62 ± 12 years, 98 male, 65 immediate 58 delayed stenting). Early and late enhancement images were transported onto a common geometry using techniques inspired by statistical atlases, allowing pixel-wise comparisons across population subgroups. A practical visualization of lesion patterns against specific clinical and therapeutic characteristics was also proposed using state-of-the-art dimensionality reduction. Results Infarct patterns were roughly comparable between the two treatments across the whole myocardium. Subtle but significant local differences were observed for the LCX and RCA territories with higher transmurality for delayed stenting at lateral and inferior/inferoseptal locations, respectively (15% and 23% of myocardial locations with a p-value <0.05, mainly in these regions). In contrast, global measurements were comparable for all territories (no statistically significant differences for all-except-one measurements before standardization / for all after standardization), although immediate stenting resulted in more subjects without reperfusion injury. Conclusion Our approach substantially empowers the analysis of lesion patterns with standardized comparisons up to the pixel resolution, and may reveal subtle differences not accessible with global observations. On the MIMI trial data as illustrative case, it confirmed its general conclusions regarding the lack of benefit of delayed stenting, but revealed subgroups differences thanks to the standardized and finer analysis scale.
Collapse
Affiliation(s)
- Nicolas Duchateau
- Univ Lyon, CREATIS, INSA, CNRS UMR 5220, INSERM U1294, Université Lyon 1, UJM Saint-Etienne, Lyon, France
- Institut Universitaire de France (IUF), Paris, France
| | - Magalie Viallon
- Univ Lyon, CREATIS, INSA, CNRS UMR 5220, INSERM U1294, Université Lyon 1, UJM Saint-Etienne, Lyon, France
- Department of Radiology, Hôpital Nord, University Hospital of Saint-Étienne, Saint-Étienne, France
| | - Lorena Petrusca
- Univ Lyon, CREATIS, INSA, CNRS UMR 5220, INSERM U1294, Université Lyon 1, UJM Saint-Etienne, Lyon, France
| | - Patrick Clarysse
- Univ Lyon, CREATIS, INSA, CNRS UMR 5220, INSERM U1294, Université Lyon 1, UJM Saint-Etienne, Lyon, France
| | - Nathan Mewton
- Department of Cardiology, Clinical Investigation Center, INSERM 1407, Hôpital Cardiovasculaire Louis Pradel, Lyon, France
| | - Loic Belle
- Department of Cardiology, Centre Hospitalier Annecy-Genevois, Annecy, France
| | - Pierre Croisille
- Univ Lyon, CREATIS, INSA, CNRS UMR 5220, INSERM U1294, Université Lyon 1, UJM Saint-Etienne, Lyon, France
- Department of Radiology, Hôpital Nord, University Hospital of Saint-Étienne, Saint-Étienne, France
| |
Collapse
|
5
|
Iqbal S, N. Qureshi A, Li J, Mahmood T. On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks. ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING : STATE OF THE ART REVIEWS 2023; 30:3173-3233. [PMID: 37260910 PMCID: PMC10071480 DOI: 10.1007/s11831-023-09899-9] [Citation(s) in RCA: 9] [Impact Index Per Article: 9.0] [Reference Citation Analysis] [Abstract] [Grants] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 12/01/2022] [Accepted: 02/19/2023] [Indexed: 06/02/2023]
Abstract
Convolutional neural network (CNN) has shown dissuasive accomplishment on different areas especially Object Detection, Segmentation, Reconstruction (2D and 3D), Information Retrieval, Medical Image Registration, Multi-lingual translation, Local language Processing, Anomaly Detection on video and Speech Recognition. CNN is a special type of Neural Network, which has compelling and effective learning ability to learn features at several steps during augmentation of the data. Recently, different interesting and inspiring ideas of Deep Learning (DL) such as different activation functions, hyperparameter optimization, regularization, momentum and loss functions has improved the performance, operation and execution of CNN Different internal architecture innovation of CNN and different representational style of CNN has significantly improved the performance. This survey focuses on internal taxonomy of deep learning, different models of vonvolutional neural network, especially depth and width of models and in addition CNN components, applications and current challenges of deep learning.
Collapse
Affiliation(s)
- Saeed Iqbal
- Department of Computer Science, Faculty of Information Technology & Computer Science, University of Central Punjab, Lahore, Punjab 54000 Pakistan
- Faculty of Information Technology, Beijing University of Technology, Beijing, 100124 Beijing China
| | - Adnan N. Qureshi
- Department of Computer Science, Faculty of Information Technology & Computer Science, University of Central Punjab, Lahore, Punjab 54000 Pakistan
| | - Jianqiang Li
- Faculty of Information Technology, Beijing University of Technology, Beijing, 100124 Beijing China
- Beijing Engineering Research Center for IoT Software and Systems, Beijing University of Technology, Beijing, 100124 Beijing China
| | - Tariq Mahmood
- Artificial Intelligence and Data Analytics (AIDA) Lab, College of Computer & Information Sciences (CCIS), Prince Sultan University, Riyadh, 11586 Kingdom of Saudi Arabia
| |
Collapse
|
6
|
Uncertainty-guided mutual consistency learning for semi-supervised medical image segmentation. Artif Intell Med 2022; 138:102476. [PMID: 36990583 DOI: 10.1016/j.artmed.2022.102476] [Citation(s) in RCA: 5] [Impact Index Per Article: 2.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/20/2021] [Revised: 11/20/2022] [Accepted: 12/13/2022] [Indexed: 12/23/2022]
Abstract
Medical image segmentation is a fundamental and critical step in many clinical approaches. Semi-supervised learning has been widely applied to medical image segmentation tasks since it alleviates the heavy burden of acquiring expert-examined annotations and takes the advantage of unlabeled data which is much easier to acquire. Although consistency learning has been proven to be an effective approach by enforcing an invariance of predictions under different distributions, existing approaches cannot make full use of region-level shape constraint and boundary-level distance information from unlabeled data. In this paper, we propose a novel uncertainty-guided mutual consistency learning framework to effectively exploit unlabeled data by integrating intra-task consistency learning from up-to-date predictions for self-ensembling and cross-task consistency learning from task-level regularization to exploit geometric shape information. The framework is guided by the estimated segmentation uncertainty of models to select out relatively certain predictions for consistency learning, so as to effectively exploit more reliable information from unlabeled data. Experiments on two publicly available benchmark datasets showed that: (1) Our proposed method can achieve significant performance improvement by leveraging unlabeled data, with up to 4.13% and 9.82% in Dice coefficient compared to supervised baseline on left atrium segmentation and brain tumor segmentation, respectively. (2) Compared with other semi-supervised segmentation methods, our proposed method achieve better segmentation performance under the same backbone network and task settings on both datasets, demonstrating the effectiveness and robustness of our method and potential transferability for other medical image segmentation tasks.
Collapse
|
7
|
Zhang Y, Liao Q, Ding L, Zhang J. Bridging 2D and 3D segmentation networks for computation-efficient volumetric medical image segmentation: An empirical study of 2.5D solutions. Comput Med Imaging Graph 2022; 99:102088. [DOI: 10.1016/j.compmedimag.2022.102088] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/25/2020] [Revised: 05/24/2022] [Accepted: 05/26/2022] [Indexed: 11/28/2022]
|