1
|
Shenouda M, Gudmundsson E, Li F, Straus CM, Kindler HL, Dudek AZ, Stinchcombe T, Wang X, Starkey A, Armato Iii SG. Convolutional Neural Networks for Segmentation of Pleural Mesothelioma: Analysis of Probability Map Thresholds (CALGB 30901, Alliance). JOURNAL OF IMAGING INFORMATICS IN MEDICINE 2024:10.1007/s10278-024-01092-z. [PMID: 39266911 DOI: 10.1007/s10278-024-01092-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 12/04/2023] [Revised: 02/28/2024] [Accepted: 02/29/2024] [Indexed: 09/14/2024]
Abstract
The purpose of this study was to evaluate the impact of probability map threshold on pleural mesothelioma (PM) tumor delineations generated using a convolutional neural network (CNN). One hundred eighty-six CT scans from 48 PM patients were segmented by a VGG16/U-Net CNN. A radiologist modified the contours generated at a 0.5 probability threshold. Percent difference of tumor volume and overlap using the Dice Similarity Coefficient (DSC) were compared between the reference standard provided by the radiologist and CNN outputs for thresholds ranging from 0.001 to 0.9. CNN-derived contours consistently yielded smaller tumor volumes than radiologist contours. Reducing the probability threshold from 0.5 to 0.01 decreased the absolute percent volume difference, on average, from 42.93% to 26.60%. Median and mean DSC ranged from 0.57 to 0.59, with a peak at a threshold of 0.2; no distinct threshold was found for percent volume difference. The CNN exhibited deficiencies with specific disease presentations, such as severe pleural effusion or disease in the pleural fissure. No single output threshold in the CNN probability maps was optimal for both tumor volume and DSC. This study emphasized the importance of considering both figures of merit when evaluating deep learning-based tumor segmentations across probability thresholds. This work underscores the need to simultaneously assess tumor volume and spatial overlap when evaluating CNN performance. While automated segmentations may yield comparable tumor volumes to that of the reference standard, the spatial region delineated by the CNN at a specific threshold is equally important.
Collapse
Affiliation(s)
- Mena Shenouda
- Department of Radiology, The University of Chicago, Chicago, IL, 60637, USA
| | | | - Feng Li
- Department of Radiology, The University of Chicago, Chicago, IL, 60637, USA
| | | | - Hedy L Kindler
- Department of Medicine, The University of Chicago, Chicago, IL, 60637, USA
| | - Arkadiusz Z Dudek
- Metro Minnesota Community Oncology Research Consortium, St. Louis Park, MN, 55416, USA
| | | | - Xiaofei Wang
- Alliance Statistics and Data Management Center, Duke University, Durham, NC, 27710, USA
| | - Adam Starkey
- Department of Radiology, The University of Chicago, Chicago, IL, 60637, USA
| | - Samuel G Armato Iii
- Department of Radiology, The University of Chicago, Chicago, IL, 60637, USA.
| |
Collapse
|
2
|
Shenouda M, Gudmundsson E, Li F, Straus CM, Kindler HL, Dudek AZ, Stinchcombe T, Wang X, Starkey A, Armato SG. Convolutional Neural Networks for Segmentation of Malignant Pleural Mesothelioma: Analysis of Probability Map Thresholds (CALGB 30901, Alliance). ARXIV 2023:arXiv:2312.00223v1. [PMID: 38076518 PMCID: PMC10705569] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Indexed: 12/22/2023]
Abstract
Malignant pleural mesothelioma (MPM) is the most common form of malignant mesothelioma, with exposure to asbestos being the primary cause of the disease. To assess response to treatment, tumor measurements are acquired and evaluated based on a patient's longitudinal computed tomography (CT) scans. Tumor volume, however, is the more accurate metric for assessing tumor burden and response. Automated segmentation methods using deep learning can be employed to acquire volume, which otherwise is a tedious task performed manually. The deep learning-based tumor volume and contours can then be compared with a standard reference to assess the robustness of the automated segmentations. The purpose of this study was to evaluate the impact of probability map threshold on MPM tumor delineations generated using a convolutional neural network (CNN). Eighty-eight CT scans from 21 MPM patients were segmented by a VGG16/U-Net CNN. A radiologist modified the contours generated at a 0.5 probability threshold. Percent difference of tumor volume and overlap using the Dice Similarity Coefficient (DSC) were compared between the standard reference provided by the radiologist and CNN outputs for thresholds ranging from 0.001 to 0.9. CNN annotations consistently yielded smaller tumor volumes than radiologist contours. Reducing the probability threshold from 0.5 to 0.1 decreased the absolute percent volume difference, on average, from 43.96% to 24.18%. Median and mean DSC ranged from 0.58 to 0.60, with a peak at a threshold of 0.5; no distinct threshold was found for percent volume difference. The CNN exhibited deficiencies with specific disease presentations, such as severe pleural effusion or disease in the pleural fissure. No single output threshold in the CNN probability maps was optimal for both tumor volume and DSC. This study emphasized the importance of considering both figures of merit when evaluating deep learning-based tumor segmentations across probability thresholds. This work underscores the need to simultaneously assess tumor volume and spatial overlap when evaluating CNN performance. While automated segmentations may yield comparable tumor volumes to that of the reference standard, the spatial region delineated by the CNN at a specific threshold is equally important.
Collapse
Affiliation(s)
- Mena Shenouda
- Department of Radiology, The University of Chicago, Chicago, IL, USA
| | | | - Feng Li
- Department of Radiology, The University of Chicago, Chicago, IL, USA
| | | | - Hedy L Kindler
- Department of Radiology, The University of Chicago, Chicago, IL, USA
| | - Arkadiusz Z Dudek
- Metro Minnesota Community Oncology Research Consortium, St. Louis Park, MN, USA
| | | | - Xiaofei Wang
- Alliance Statistics and Data Management Center, Duke University, Durham, NC, USA
| | - Adam Starkey
- Department of Radiology, The University of Chicago, Chicago, IL, USA
| | - Samuel G Armato
- Department of Radiology, The University of Chicago, Chicago, IL, USA
| |
Collapse
|
3
|
Bos S, Ricciardi S, Caruana EJ, Öztürk NAA, Magouliotis D, Pompili C, Migliore M, Vos R, Meloni F, Elia S, Hellemons M. ERS International Congress 2021: highlights from Assembly 8 Thoracic Surgery and Lung Transplantation. ERJ Open Res 2022; 8:00649-2021. [PMID: 35615414 PMCID: PMC9125043 DOI: 10.1183/23120541.00649-2021] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/22/2021] [Accepted: 02/22/2022] [Indexed: 11/05/2022] Open
Abstract
The thoracic surgery and lung transplantation assembly of the European Respiratory Society (ERS) is delighted to present the highlights from the 2021 International ERS Congress. We have selected four sessions that discussed recent advances across a wide range of topics: including digital health surveillance in thoracic surgery, emerging concepts in pulmonary metastasectomy, advances in mesothelioma care, and novel developments in lung graft allocation and monitoring. The sessions are summarised by early career members in close collaboration with the assembly faculty. We aim to give the reader an update on the highlights of the conference in the fields of thoracic surgery and lung transplantation.
Collapse
|
4
|
Romei C, Fanni SC, Volpi F, Milazzo A, D’Amore CA, Colligiani L, Neri E, De Liperi A, Stella GM, Bortolotto C. New Updates of the Imaging Role in Diagnosis, Staging, and Response Treatment of Malignant Pleural Mesothelioma. Cancers (Basel) 2021; 13:cancers13174377. [PMID: 34503186 PMCID: PMC8430786 DOI: 10.3390/cancers13174377] [Citation(s) in RCA: 7] [Impact Index Per Article: 2.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/27/2021] [Revised: 08/24/2021] [Accepted: 08/27/2021] [Indexed: 11/16/2022] Open
Abstract
Malignant pleural mesothelioma is a rare neoplasm with poor prognosis. CT is the first imaging technique used for diagnosis, staging, and assessment of therapy response. Although, CT has intrinsic limitations due to low soft tissue contrast and the current staging system as well as criteria for evaluating response, it does not consider the complex growth pattern of this tumor. Computer-based methods have proven their potentiality in diagnosis, staging, prognosis, and assessment of therapy response; moreover, computer-based methods can make feasible tasks like segmentation that would otherwise be impracticable. MRI, thanks to its high soft tissue contrast evaluation of contrast enhancement and through diffusion-weighted-images, could replace CT in many clinical settings.
Collapse
Affiliation(s)
- Chiara Romei
- 2nd Radiology Unit, Radiology Department, Pisa University Hospital, 56124 Pisa, Italy;
- Correspondence: (C.R.); (S.C.F.)
| | - Salvatore Claudio Fanni
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
- Correspondence: (C.R.); (S.C.F.)
| | - Federica Volpi
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
| | - Alessio Milazzo
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
| | - Caterina Aida D’Amore
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
| | - Leonardo Colligiani
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
| | - Emanuele Neri
- Department of Translational Research, Academic Radiology, University of Pisa, 56124 Pisa, Italy; (F.V.); (A.M.); (C.A.D.); (L.C.); (E.N.)
| | - Annalisa De Liperi
- 2nd Radiology Unit, Radiology Department, Pisa University Hospital, 56124 Pisa, Italy;
| | - Giulia Maria Stella
- Unit of Respiratory Diseases, Department of Medical Sciences and Infective Diseases, IRCCS Policlinico San Matteo Foundation, University of Pavia Medical School, 27100 Pavia, Italy;
| | - Chandra Bortolotto
- Unit of Radiology, Department of Intensive Medicine, IRCCS Policlinico San Matteo Foundation, University of Pavia Medical School, 27100 Pavia, Italy;
| |
Collapse
|
5
|
Fu Y, Lei Y, Wang T, Curran WJ, Liu T, Yang X. A review of deep learning based methods for medical image multi-organ segmentation. Phys Med 2021; 85:107-122. [PMID: 33992856 PMCID: PMC8217246 DOI: 10.1016/j.ejmp.2021.05.003] [Citation(s) in RCA: 73] [Impact Index Per Article: 24.3] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 11/27/2020] [Revised: 03/12/2021] [Accepted: 05/03/2021] [Indexed: 12/12/2022] Open
Abstract
Deep learning has revolutionized image processing and achieved the-state-of-art performance in many medical image segmentation tasks. Many deep learning-based methods have been published to segment different parts of the body for different medical applications. It is necessary to summarize the current state of development for deep learning in the field of medical image segmentation. In this paper, we aim to provide a comprehensive review with a focus on multi-organ image segmentation, which is crucial for radiotherapy where the tumor and organs-at-risk need to be contoured for treatment planning. We grouped the surveyed methods into two broad categories which are 'pixel-wise classification' and 'end-to-end segmentation'. Each category was divided into subgroups according to their network design. For each type, we listed the surveyed works, highlighted important contributions and identified specific challenges. Following the detailed review, we discussed the achievements, shortcomings and future potentials of each category. To enable direct comparison, we listed the performance of the surveyed works that used thoracic and head-and-neck benchmark datasets.
Collapse
Affiliation(s)
- Yabo Fu
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA
| | - Yang Lei
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA
| | - Tonghe Wang
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA
| | - Walter J Curran
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA
| | - Tian Liu
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA
| | - Xiaofeng Yang
- Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA.
| |
Collapse
|
6
|
Kiser KJ, Ahmed S, Stieb S, Mohamed ASR, Elhalawani H, Park PYS, Doyle NS, Wang BJ, Barman A, Li Z, Zheng WJ, Fuller CD, Giancardo L. PleThora: Pleural effusion and thoracic cavity segmentations in diseased lungs for benchmarking chest CT processing pipelines. Med Phys 2020; 47:5941-5952. [PMID: 32749075 PMCID: PMC7722027 DOI: 10.1002/mp.14424] [Citation(s) in RCA: 13] [Impact Index Per Article: 3.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/07/2020] [Revised: 07/22/2020] [Accepted: 07/27/2020] [Indexed: 12/19/2022] Open
Abstract
This manuscript describes a dataset of thoracic cavity segmentations and discrete pleural effusion segmentations we have annotated on 402 computed tomography (CT) scans acquired from patients with non-small cell lung cancer. The segmentation of these anatomic regions precedes fundamental tasks in image analysis pipelines such as lung structure segmentation, lesion detection, and radiomics feature extraction. Bilateral thoracic cavity volumes and pleural effusion volumes were manually segmented on CT scans acquired from The Cancer Imaging Archive "NSCLC Radiomics" data collection. Four hundred and two thoracic segmentations were first generated automatically by a U-Net based algorithm trained on chest CTs without cancer, manually corrected by a medical student to include the complete thoracic cavity (normal, pathologic, and atelectatic lung parenchyma, lung hilum, pleural effusion, fibrosis, nodules, tumor, and other anatomic anomalies), and revised by a radiation oncologist or a radiologist. Seventy-eight pleural effusions were manually segmented by a medical student and revised by a radiologist or radiation oncologist. Interobserver agreement between the radiation oncologist and radiologist corrections was acceptable. All expert-vetted segmentations are publicly available in NIfTI format through The Cancer Imaging Archive at https://doi.org/10.7937/tcia.2020.6c7y-gq39. Tabular data detailing clinical and technical metadata linked to segmentation cases are also available. Thoracic cavity segmentations will be valuable for developing image analysis pipelines on pathologic lungs - where current automated algorithms struggle most. In conjunction with gross tumor volume segmentations already available from "NSCLC Radiomics," pleural effusion segmentations may be valuable for investigating radiomics profile differences between effusion and primary tumor or training algorithms to discriminate between them.
Collapse
Affiliation(s)
- Kendall J. Kiser
- John P. and Kathrine G. McGovern Medical SchoolHoustonTXUSA
- Center for Precision HealthUTHealth School of Biomedical InformaticsHoustonTXUSA
- Department of Radiation OncologyUniversity of Texas MD Anderson Cancer CenterHoustonTXUSA
| | - Sara Ahmed
- Department of Radiation OncologyUniversity of Texas MD Anderson Cancer CenterHoustonTXUSA
| | - Sonja Stieb
- Department of Radiation OncologyUniversity of Texas MD Anderson Cancer CenterHoustonTXUSA
| | - Abdallah S. R. Mohamed
- Department of Radiation OncologyUniversity of Texas MD Anderson Cancer CenterHoustonTXUSA
- MD Anderson Cancer Center‐UTHealth Graduate School of Biomedical SciencesHoustonTXUSA
| | - Hesham Elhalawani
- Department of Radiation OncologyCleveland Clinic Taussig Cancer CenterClevelandOHUSA
| | - Peter Y. S. Park
- Department of Diagnostic and Interventional ImagingJohn P. and Kathrine G. McGovern Medical SchoolHoustonTXUSA
| | - Nathan S. Doyle
- Department of Diagnostic and Interventional ImagingJohn P. and Kathrine G. McGovern Medical SchoolHoustonTXUSA
| | - Brandon J. Wang
- Department of Diagnostic and Interventional ImagingJohn P. and Kathrine G. McGovern Medical SchoolHoustonTXUSA
| | - Arko Barman
- Center for Precision HealthUTHealth School of Biomedical InformaticsHoustonTXUSA
| | - Zhao Li
- Center for Precision HealthUTHealth School of Biomedical InformaticsHoustonTXUSA
| | - W. Jim Zheng
- Center for Precision HealthUTHealth School of Biomedical InformaticsHoustonTXUSA
| | - Clifton D. Fuller
- Department of Radiation OncologyUniversity of Texas MD Anderson Cancer CenterHoustonTXUSA
- MD Anderson Cancer Center‐UTHealth Graduate School of Biomedical SciencesHoustonTXUSA
| | - Luca Giancardo
- Center for Precision HealthUTHealth School of Biomedical InformaticsHoustonTXUSA
- Department of Radiation OncologyCleveland Clinic Taussig Cancer CenterClevelandOHUSA
| |
Collapse
|
7
|
Gudmundsson E, Straus CM, Li F, Armato SG. Deep learning-based segmentation of malignant pleural mesothelioma tumor on computed tomography scans: application to scans demonstrating pleural effusion. J Med Imaging (Bellingham) 2020; 7:012705. [PMID: 32016133 DOI: 10.1117/1.jmi.7.1.012705] [Citation(s) in RCA: 7] [Impact Index Per Article: 1.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/23/2019] [Accepted: 12/24/2019] [Indexed: 12/29/2022] Open
Abstract
Tumor volume is a topic of interest for the prognostic assessment, treatment response evaluation, and staging of malignant pleural mesothelioma. Many mesothelioma patients present with, or develop, pleural fluid, which may complicate the segmentation of this disease. Deep convolutional neural networks (CNNs) of the two-dimensional U-Net architecture were trained for segmentation of tumor in the left and right hemithoraces, with the networks initialized through layers pretrained on ImageNet. Networks were trained on a dataset of 5230 axial sections from 154 CT scans of 126 mesothelioma patients. A test set of 94 CT sections from 34 patients, who all presented with both tumor and pleural effusion, in addition to a more general test set of 130 CT sections from 43 patients, were used to evaluate segmentation performance of the deep CNNs. The Dice similarity coefficient (DSC), average Hausdorff distance, and bias in predicted tumor area were calculated through comparisons with radiologist-provided tumor segmentations on the test sets. The present method achieved a median DSC of 0.690 on the tumor and effusion test set and achieved significantly higher performance on both test sets when compared with a previous deep learning-based segmentation method for mesothelioma.
Collapse
Affiliation(s)
- Eyjolfur Gudmundsson
- The University of Chicago, Department of Radiology, Chicago, Illinois, United States
| | - Christopher M Straus
- The University of Chicago, Department of Radiology, Chicago, Illinois, United States
| | - Feng Li
- The University of Chicago, Department of Radiology, Chicago, Illinois, United States
| | - Samuel G Armato
- The University of Chicago, Department of Radiology, Chicago, Illinois, United States
| |
Collapse
|
8
|
Armato SG, Francis RJ, Katz SI, Ak G, Opitz I, Gudmundsson E, Blyth KG, Gupta A. Imaging in pleural mesothelioma: A review of the 14th International Conference of the International Mesothelioma Interest Group. Lung Cancer 2018; 130:108-114. [PMID: 30885330 DOI: 10.1016/j.lungcan.2018.11.033] [Citation(s) in RCA: 11] [Impact Index Per Article: 1.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/19/2018] [Accepted: 11/25/2018] [Indexed: 01/02/2023]
Abstract
Mesothelioma patients rely on the information their clinical team obtains from medical imaging. Whether x-ray-based computed tomography (CT) or magnetic resonance imaging (MRI) based on local magnetic fields within a patient's tissues, different modalities generate images with uniquely different appearances and information content due to the physical differences of the image-acquisition process. Researchers are developing sophisticated ways to extract a greater amount of the information contained within these images. This paper summarizes the imaging-based research presented orally at the 2018 International Conference of the International Mesothelioma Interest Group (iMig) in Ottawa, Ontario, Canada, held May 2-5, 2018. Presented topics included advances in the imaging of preclinical mesothelioma models to inform clinical therapeutic strategies, optimization of the time delay between contrast administration and image acquisition for maximized enhancement of mesothelioma tumor on CT, an investigation of image-based criteria for clinical tumor and nodal staging of mesothelioma by contrast-enhanced CT, an investigation of methods for the extraction of mesothelioma tumor volume from MRI and the association of volume with patient survival, the use of deep learning for mesothelioma tumor segmentation in CT, and an evaluation of CT-based radiomics for the prognosis of mesothelioma patient survival.
Collapse
Affiliation(s)
- Samuel G Armato
- Department of Radiology, The University of Chicago, Chicago, Illinois, USA.
| | - Roslyn J Francis
- Department of Nuclear Medicine, Sir Charles Gairdner Hospital, Perth, Western Australia, Australia; Faculty of Health and Medical Sciences, University of Western Australia Medical School, Australia
| | - Sharyn I Katz
- Department of Radiology, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA
| | - Guntulu Ak
- Lung and Pleural Cancers Research and Clinical Center, Eskisehir Osmangazi University, Eskisehir, Turkey
| | - Isabelle Opitz
- Division of Thoracic Surgery, University Hospital Zurich, Zurich, Switzerland
| | | | - Kevin G Blyth
- Glasgow Pleural Disease Unit, Queen Elizabeth University Hospital, Glasgow, UK; Institute of Infection, Immunity & Inflammation, University of Glasgow, UK
| | - Ashish Gupta
- Department of Radiology, The Ottawa Hospital, Ottawa, Ontario, Canada
| |
Collapse
|