1
|
Wang L, Kelly B, Lee EH, Wang H, Zheng J, Zhang W, Halabi S, Liu J, Tian Y, Han B, Huang C, Yeom KW, Deng K, Song J. Multi-classifier-based identification of COVID-19 from chest computed tomography using generalizable and interpretable radiomics features. Eur J Radiol 2021; 136:109552. [PMID: 33497881 PMCID: PMC7810032 DOI: 10.1016/j.ejrad.2021.109552] [Citation(s) in RCA: 17] [Impact Index Per Article: 5.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/04/2020] [Revised: 12/09/2020] [Accepted: 01/12/2021] [Indexed: 12/11/2022]
Abstract
PURPOSE To investigate the efficacy of radiomics in diagnosing patients with coronavirus disease (COVID-19) and other types of viral pneumonia with clinical symptoms and CT signs similar to those of COVID-19. METHODS Between 18 January 2020 and 20 May 2020, 110 SARS-CoV-2 positive and 108 SARS-CoV-2 negative patients were retrospectively recruited from three hospitals based on the inclusion criteria. Manual segmentation of pneumonia lesions on CT scans was performed by four radiologists. The latest version of Pyradiomics was used for feature extraction. Four classifiers (linear classifier, k-nearest neighbour, least absolute shrinkage and selection operator [LASSO], and random forest) were used to differentiate SARS-CoV-2 positive and SARS-CoV-2 negative patients. Comparison of the performance of the classifiers and radiologists was evaluated by ROC curve and Kappa score. RESULTS We manually segmented 16,053 CT slices, comprising 32,625 pneumonia lesions, from the CT scans of all patients. Using Pyradiomics, 120 radiomic features were extracted from each image. The key radiomic features screened by different classifiers varied and lead to significant differences in classification accuracy. The LASSO achieved the best performance (sensitivity: 72.2%, specificity: 75.1%, and AUC: 0.81) on the external validation dataset and attained excellent agreement (Kappa score: 0.89) with radiologists (average sensitivity: 75.6%, specificity: 78.2%, and AUC: 0.81). All classifiers indicated that "Original_Firstorder_RootMeanSquared" and "Original_Firstorder_Uniformity" were significant features for this task. CONCLUSIONS We identified radiomic features that were significantly associated with the classification of COVID-19 pneumonia using multiple classifiers. The quantifiable interpretation of the differences in features between the two groups extends our understanding of CT imaging characteristics of COVID-19 pneumonia.
Collapse
Affiliation(s)
- Lu Wang
- School of Medical Informatics, China Medical University Puhe Rd, Shenbei New District, Shenyang, Liaoning, 110122, China
| | - Brendan Kelly
- Department of Radiology, School of Medicine, Stanford University 725 Welch Rd MC 5654, Palo Alto, CA, 94305, United States
| | - Edward H. Lee
- Department of Radiology, School of Medicine, Stanford University 725 Welch Rd MC 5654, Palo Alto, CA, 94305, United States
| | - Hongmei Wang
- Department of Radiology, The First Affiliated Hospital of University of Science and Technology of China, No. 1 Swan Lake Road Hefei, Anhui, 230036, China
| | - Jimmy Zheng
- Department of Radiology, School of Medicine, Stanford University 725 Welch Rd MC 5654, Palo Alto, CA, 94305, United States
| | - Wei Zhang
- Department of Radiology, the Lu’an Affiliated Hospital, Anhui Medical University, No. 21 Wanxi Rd, Lu’an, Anhui, 237005, China
| | - Safwan Halabi
- Department of Radiology, School of Medicine, Stanford University 725 Welch Rd MC 5654, Palo Alto, CA, 94305, United States
| | - Jining Liu
- Bengbu Medical College, Department of Imaging Medicine, 2600 Donghai Avenue, Bengbu, Anhui, 233030, China
| | - Yulong Tian
- Wannan Medical College, Department of Imaging Medicine and Nuclear Medicine, 22 Wenchang West Rd, Higher Education Park, Wuhu, Anhui, 241002, China
| | - Baoqin Han
- Wannan Medical College, Department of Imaging Medicine and Nuclear Medicine, 22 Wenchang West Rd, Higher Education Park, Wuhu, Anhui, 241002, China
| | - Chuanbin Huang
- Wannan Medical College, Department of Imaging Medicine and Nuclear Medicine, 22 Wenchang West Rd, Higher Education Park, Wuhu, Anhui, 241002, China
| | - Kristen W. Yeom
- Department of Radiology, School of Medicine, Stanford University 725 Welch Rd MC 5654, Palo Alto, CA, 94305, United States
| | - Kexue Deng
- Department of Radiology, The First Affiliated Hospital of University of Science and Technology of China, No. 1 Swan Lake Road Hefei, Anhui, 230036, China,Corresponding author
| | - Jiangdian Song
- School of Medical Informatics, China Medical University Puhe Rd, Shenbei New District, Shenyang, Liaoning, 110122, China; Department of Radiology, School of Medicine, Stanford University 1201 Welch Rd, Lucas Center, Palo Alto, CA, 94305, United States.
| |
Collapse
|
2
|
Wang F, Zhou L, Chen N, Li X. The effect of pretreatment BMI on the prognosis and serum immune cells in advanced LSCC patients who received ICI therapy. Medicine (Baltimore) 2021; 100:e24664. [PMID: 33663076 PMCID: PMC7909129 DOI: 10.1097/md.0000000000024664] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 08/12/2020] [Accepted: 01/19/2021] [Indexed: 01/05/2023] Open
Abstract
This study aims to evaluate the prognosis and serum immune cells of patients with different pretreatment body mass index (BMI) values. The data of 61 newly diagnosed patients with advanced lung squamous cell carcinoma (LSCC) who received immune checkpoint inhibitors (ICIs) combined with chemotherapy were obtained from the database of Rizhao People's Hospital (Rizhao, Shandong). According to the cutoff value of BMI (23.2 kg/m2), 32 patients had a high BMI and the remaining 29 patients had a low BMI. The effects of different BMIs on the prognosis and serum immune cells of patients were analyzed. The median progression-free survival (PFS) times were 7.72 months in the high BMI group and 4.83 months in the low BMI group [adjusted hazard ratio (HR), 0.23; 95% confidence interval (CI), 0.11-0.48; P < .001]. In terms of the overall survival (OS), the median times of the high BMI group and low BMI group were 18.10 and 13.90 months, respectively (adjusted HR, 0.15; 95% CI, 0.07-0.32; P < .001). After 4 cycles of ICI therapy combined with chemotherapy, the objective response rate was 59.4% for the high BMI group and 20.7% for the low BMI group (P = .002). In addition, the number of serum immune cells in patients with high BMI was significantly higher than that in patients with low BMI (all P < .001). There was a linear relationship between BMI value and the number of serum immune cells (all R2 > 0.7). The current results showed that high BMI is associated with better prognosis in LSCC patients who received ICIs, which may be related to higher levels of serum immune cells.
Collapse
|
3
|
Guillo E, Bedmar Gomez I, Dangeard S, Bennani S, Saab I, Tordjman M, Jilet L, Chassagnon G, Revel MP. COVID-19 pneumonia: Diagnostic and prognostic role of CT based on a retrospective analysis of 214 consecutive patients from Paris, France. Eur J Radiol 2020; 131:109209. [PMID: 32810701 PMCID: PMC7414360 DOI: 10.1016/j.ejrad.2020.109209] [Citation(s) in RCA: 34] [Impact Index Per Article: 8.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/08/2020] [Revised: 07/16/2020] [Accepted: 08/03/2020] [Indexed: 02/07/2023]
Abstract
OBJECTIVES To evaluate the diagnostic and prognostic performance of CT in patients referred for COVID19 suspicion to a French university hospital, depending on symptoms and date of onset. METHODS From March 1st to March 28th, 214 patients having both chest CT scan and reverse transcriptase polymerase chain reaction (RT- PCT) within 24 h were retrospectively evaluated. Sensitivity, specificity, negative and positive predictive values of first and expert readings were calculated together with inter reader agreement, with results of RT-PCR as standard of reference and according to symptoms and onset date. Patient characteristics and disease extent on CT were correlated to short-term outcome (death or intubation at 3 weeks follow-up). RESULTS Of the 214 patients (119 men, mean age 59 ± 19 years), 129 had at least one positive RT-PCR result. Sensitivity, specificity, negative and positive predictive values were 79 % (95 % CI: 71-86 %), 84 %(74-91 %), 72 %(63-81 %) and 88 % (81-93 %) for initial CT reading and 81 %(74-88 %), 91 % (82-96 %), 76 % (67-84 %) and 93 % (87-97 %), for expert reading, with strong inter-reader agreement (kappa index: 0.89). Considering the 123 patients with symptoms for more than 5 days, the corresponding figures were 90 %, 78 %, 80 % and 89 % for initial reading and 93 %, 88 %, 86 % and 94 % for the expert. Disease extent exceeded 25 % for 68 % and 26 % of severe and non-severe patients, respectively (p < 0.001). CONCLUSION CT sensitivity increased after 5 days of symptoms. A disease extent > 25 % was associated with poorer outcome.
Collapse
Affiliation(s)
- Enora Guillo
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France
| | - Ines Bedmar Gomez
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France
| | - Severine Dangeard
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France
| | - Souhail Bennani
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France
| | - Ines Saab
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France; Université de Paris, Descartes-Paris 5, 75006, Paris, France
| | - Mickael Tordjman
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France
| | - Lea Jilet
- Unité de Recherche Clinique Centre d'Investigation Clinique, Paris Descartes Necker/Cochin, Hôpital Tarnier, 75014, Paris, France
| | - Guillaume Chassagnon
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France; Université de Paris, Descartes-Paris 5, 75006, Paris, France
| | - Marie-Pierre Revel
- Department of Radiology, Cochin Hospital, AP-HP. Centre, 75014, Paris, France; Université de Paris, Descartes-Paris 5, 75006, Paris, France.
| |
Collapse
|
4
|
Deep learning: definition and perspectives for thoracic imaging. Eur Radiol 2019; 30:2021-2030. [PMID: 31811431 DOI: 10.1007/s00330-019-06564-3] [Citation(s) in RCA: 31] [Impact Index Per Article: 6.2] [Reference Citation Analysis] [Abstract] [Key Words] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/10/2019] [Revised: 10/22/2019] [Accepted: 10/30/2019] [Indexed: 12/19/2022]
Abstract
Relevance and penetration of machine learning in clinical practice is a recent phenomenon with multiple applications being currently under development. Deep learning-and especially convolutional neural networks (CNNs)-is a subset of machine learning, which has recently entered the field of thoracic imaging. The structure of neural networks, organized in multiple layers, allows them to address complex tasks. For several clinical situations, CNNs have demonstrated superior performance as compared with classical machine learning algorithms and in some cases achieved comparable or better performance than clinical experts. Chest radiography, a high-volume procedure, is a natural application domain because of the large amount of stored images and reports facilitating the training of deep learning algorithms. Several algorithms for automated reporting have been developed. The training of deep learning algorithm CT images is more complex due to the dimension, variability, and complexity of the 3D signal. The role of these methods is likely to increase in clinical practice as a complement of the radiologist's expertise. The objective of this review is to provide definitions for understanding the methods and their potential applications for thoracic imaging. KEY POINTS: • Deep learning outperforms other machine learning techniques for number of tasks in radiology. • Convolutional neural network is the most popular deep learning architecture in medical imaging. • Numerous deep learning algorithms are being currently developed; some of them may become part of clinical routine in the near future.
Collapse
|