QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps
Background and Purpose: Chronic active multiple sclerosis (MS) lesions are characterized by a paramagnetic rim at the edge of the lesion and are associated with increased disability in patients. Quantitative susceptibility mapping (QSM) is an MRI technique that is sensitive to chronic active lesions...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-01-01
|
| Series: | NeuroImage: Clinical |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2213158222000444 |
| _version_ | 1811233826738798592 |
|---|---|
| author | Hang Zhang Thanh D. Nguyen Jinwei Zhang Melanie Marcille Pascal Spincemaille Yi Wang Susan A. Gauthier Elizabeth M. Sweeney |
| author_facet | Hang Zhang Thanh D. Nguyen Jinwei Zhang Melanie Marcille Pascal Spincemaille Yi Wang Susan A. Gauthier Elizabeth M. Sweeney |
| author_sort | Hang Zhang |
| collection | DOAJ |
| description | Background and Purpose: Chronic active multiple sclerosis (MS) lesions are characterized by a paramagnetic rim at the edge of the lesion and are associated with increased disability in patients. Quantitative susceptibility mapping (QSM) is an MRI technique that is sensitive to chronic active lesions, termed rim + lesions on the QSM. We present QSMRim-Net, a data imbalance-aware deep neural network that fuses lesion-level radiomic and convolutional image features for automated identification of rim + lesions on QSM. Methods: QSM and T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) MRI of the brain were collected at 3 T for 172 MS patients. Rim + lesions were manually annotated by two human experts, followed by consensus from a third expert, for a total of 177 rim + and 3986 rim negative (rim−) lesions. Our automated rim + detection algorithm, QSMRim-Net, consists of a two-branch feature extraction network and a synthetic minority oversampling network to classify rim + lesions. The first network branch is for image feature extraction from the QSM and T2-FLAIR, and the second network branch is a fully connected network for QSM lesion-level radiomic feature extraction. The oversampling network is designed to increase classification performance with imbalanced data. Results: On a lesion-level, in a five-fold cross validation framework, the proposed QSMRim-Net detected rim + lesions with a partial area under the receiver operating characteristic curve (pROC AUC) of 0.760, where clinically relevant false positive rates of less than 0.1 were considered. The method attained an area under the precision recall curve (PR AUC) of 0.704. QSMRim-Net out-performed other state-of-the-art methods applied to the QSM on both pROC AUC and PR AUC. On a subject-level, comparing the predicted rim + lesion count and the human expert annotated count, QSMRim-Net achieved the lowest mean square error of 0.98 and the highest correlation of 0.89 (95% CI: 0.86, 0.92). Conclusion: This study develops a novel automated deep neural network for rim + MS lesion identification using T2-FLAIR and QSM images. |
| first_indexed | 2024-04-12T11:25:50Z |
| format | Article |
| id | doaj.art-d597666120be446e8da665e52b7b8ac5 |
| institution | Directory Open Access Journal |
| issn | 2213-1582 |
| language | English |
| last_indexed | 2024-04-12T11:25:50Z |
| publishDate | 2022-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage: Clinical |
| spelling | doaj.art-d597666120be446e8da665e52b7b8ac52022-12-22T03:35:14ZengElsevierNeuroImage: Clinical2213-15822022-01-0134102979QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility mapsHang Zhang0Thanh D. Nguyen1Jinwei Zhang2Melanie Marcille3Pascal Spincemaille4Yi Wang5Susan A. Gauthier6Elizabeth M. Sweeney7Department of Electrical and Computer Engineering, Cornell University, Ithaca, NY, USA; Department of Radiology, Weill Cornell Medicine, New York, NY, USADepartment of Radiology, Weill Cornell Medicine, New York, NY, USADepartment of Radiology, Weill Cornell Medicine, New York, NY, USA; Department of Biomedical Engineering, Cornell University, Ithaca, NY, USADepartment of Radiology, Weill Cornell Medicine, New York, NY, USADepartment of Radiology, Weill Cornell Medicine, New York, NY, USADepartment of Electrical and Computer Engineering, Cornell University, Ithaca, NY, USA; Department of Radiology, Weill Cornell Medicine, New York, NY, USA; Department of Biomedical Engineering, Cornell University, Ithaca, NY, USADepartment of Radiology, Weill Cornell Medicine, New York, NY, USA; Department of Neurology, Weill Cornell Medicine, New York, NY, USA; Feil Family Brain and Mind Institute, Weill Cornell Medicine, New York, NY, USAPenn Statistics in Imaging and Visualization Endeavor (PennSIVE) Center, Department of Biostatistics, Epidemiology, and Informatics, University of Pennsylvania, Philadelphia, PA, USA; Corresponding author.Background and Purpose: Chronic active multiple sclerosis (MS) lesions are characterized by a paramagnetic rim at the edge of the lesion and are associated with increased disability in patients. Quantitative susceptibility mapping (QSM) is an MRI technique that is sensitive to chronic active lesions, termed rim + lesions on the QSM. We present QSMRim-Net, a data imbalance-aware deep neural network that fuses lesion-level radiomic and convolutional image features for automated identification of rim + lesions on QSM. Methods: QSM and T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) MRI of the brain were collected at 3 T for 172 MS patients. Rim + lesions were manually annotated by two human experts, followed by consensus from a third expert, for a total of 177 rim + and 3986 rim negative (rim−) lesions. Our automated rim + detection algorithm, QSMRim-Net, consists of a two-branch feature extraction network and a synthetic minority oversampling network to classify rim + lesions. The first network branch is for image feature extraction from the QSM and T2-FLAIR, and the second network branch is a fully connected network for QSM lesion-level radiomic feature extraction. The oversampling network is designed to increase classification performance with imbalanced data. Results: On a lesion-level, in a five-fold cross validation framework, the proposed QSMRim-Net detected rim + lesions with a partial area under the receiver operating characteristic curve (pROC AUC) of 0.760, where clinically relevant false positive rates of less than 0.1 were considered. The method attained an area under the precision recall curve (PR AUC) of 0.704. QSMRim-Net out-performed other state-of-the-art methods applied to the QSM on both pROC AUC and PR AUC. On a subject-level, comparing the predicted rim + lesion count and the human expert annotated count, QSMRim-Net achieved the lowest mean square error of 0.98 and the highest correlation of 0.89 (95% CI: 0.86, 0.92). Conclusion: This study develops a novel automated deep neural network for rim + MS lesion identification using T2-FLAIR and QSM images.http://www.sciencedirect.com/science/article/pii/S2213158222000444Multiple sclerosisQuantitative susceptibility mappingChronic active lesionsConvolutional neural networkRadiomic features |
| spellingShingle | Hang Zhang Thanh D. Nguyen Jinwei Zhang Melanie Marcille Pascal Spincemaille Yi Wang Susan A. Gauthier Elizabeth M. Sweeney QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps NeuroImage: Clinical Multiple sclerosis Quantitative susceptibility mapping Chronic active lesions Convolutional neural network Radiomic features |
| title | QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| title_full | QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| title_fullStr | QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| title_full_unstemmed | QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| title_short | QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| title_sort | qsmrim net imbalance aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps |
| topic | Multiple sclerosis Quantitative susceptibility mapping Chronic active lesions Convolutional neural network Radiomic features |
| url | http://www.sciencedirect.com/science/article/pii/S2213158222000444 |
| work_keys_str_mv | AT hangzhang qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT thanhdnguyen qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT jinweizhang qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT melaniemarcille qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT pascalspincemaille qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT yiwang qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT susanagauthier qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps AT elizabethmsweeney qsmrimnetimbalanceawarelearningforidentificationofchronicactivemultiplesclerosislesionsonquantitativesusceptibilitymaps |