BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning
BackgroundA wide variety of benign and malignant processes can manifest as non-mass enhancement (NME) in breast MRI. Compared to mass lesions, there are no distinct features that can be used for differential diagnosis. The purpose is to use the BI-RADS descriptors and models developed using radiomic...
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2021-11-01
|
| Series: | Frontiers in Oncology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fonc.2021.728224/full |
| _version_ | 1818824915800817664 |
|---|---|
| author | Jiejie Zhou Jiejie Zhou Yan-Lin Liu Yang Zhang Yang Zhang Jeon-Hor Chen Jeon-Hor Chen Freddie J. Combs Ritesh Parajuli Rita S. Mehta Huiru Liu Zhongwei Chen Youfan Zhao Zhifang Pan Meihao Wang Risheng Yu Min-Ying Su Min-Ying Su |
| author_facet | Jiejie Zhou Jiejie Zhou Yan-Lin Liu Yang Zhang Yang Zhang Jeon-Hor Chen Jeon-Hor Chen Freddie J. Combs Ritesh Parajuli Rita S. Mehta Huiru Liu Zhongwei Chen Youfan Zhao Zhifang Pan Meihao Wang Risheng Yu Min-Ying Su Min-Ying Su |
| author_sort | Jiejie Zhou |
| collection | DOAJ |
| description | BackgroundA wide variety of benign and malignant processes can manifest as non-mass enhancement (NME) in breast MRI. Compared to mass lesions, there are no distinct features that can be used for differential diagnosis. The purpose is to use the BI-RADS descriptors and models developed using radiomics and deep learning to distinguish benign from malignant NME lesions.Materials and MethodsA total of 150 patients with 104 malignant and 46 benign NME were analyzed. Three radiologists performed reading for morphological distribution and internal enhancement using the 5th BI-RADS lexicon. For each case, the 3D tumor mask was generated using Fuzzy-C-Means segmentation. Three DCE parametric maps related to wash-in, maximum, and wash-out were generated, and PyRadiomics was applied to extract features. The radiomics model was built using five machine learning algorithms. ResNet50 was implemented using three parametric maps as input. Approximately 70% of earlier cases were used for training, and 30% of later cases were held out for testing.ResultsThe diagnostic BI-RADS in the original MRI report showed that 104/104 malignant and 36/46 benign lesions had a BI-RADS score of 4A–5. For category reading, the kappa coefficient was 0.83 for morphological distribution (excellent) and 0.52 for internal enhancement (moderate). Segmental and Regional distribution were the most prominent for the malignant group, and focal distribution for the benign group. Eight radiomics features were selected by support vector machine (SVM). Among the five machine learning algorithms, SVM yielded the highest accuracy of 80.4% in training and 77.5% in testing datasets. ResNet50 had a better diagnostic performance, 91.5% in training and 83.3% in testing datasets.ConclusionDiagnosis of NME was challenging, and the BI-RADS scores and descriptors showed a substantial overlap. Radiomics and deep learning may provide a useful CAD tool to aid in diagnosis. |
| first_indexed | 2024-12-19T00:03:28Z |
| format | Article |
| id | doaj.art-2f65c7c8b295401da0052da4b4502bc5 |
| institution | Directory Open Access Journal |
| issn | 2234-943X |
| language | English |
| last_indexed | 2024-12-19T00:03:28Z |
| publishDate | 2021-11-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Oncology |
| spelling | doaj.art-2f65c7c8b295401da0052da4b4502bc52022-12-21T20:46:21ZengFrontiers Media S.A.Frontiers in Oncology2234-943X2021-11-011110.3389/fonc.2021.728224728224BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep LearningJiejie Zhou0Jiejie Zhou1Yan-Lin Liu2Yang Zhang3Yang Zhang4Jeon-Hor Chen5Jeon-Hor Chen6Freddie J. Combs7Ritesh Parajuli8Rita S. Mehta9Huiru Liu10Zhongwei Chen11Youfan Zhao12Zhifang Pan13Meihao Wang14Risheng Yu15Min-Ying Su16Min-Ying Su17Department of Radiology, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, ChinaDepartment of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaDepartment of Radiological Sciences, University of California, Irvine, Irvine, CA, United StatesDepartment of Radiological Sciences, University of California, Irvine, Irvine, CA, United StatesDepartment of Radiation Oncology, Rutgers-Cancer Institute of New Jersey, Robert Wood Johnson Medical School, New Brunswick, NJ, United StatesDepartment of Radiological Sciences, University of California, Irvine, Irvine, CA, United StatesDepartment of Radiology, E-DA Hospital and I-Shou University, Kaohsiung, TaiwanDepartment of Radiological Sciences, University of California, Irvine, Irvine, CA, United StatesDepartment of Medicine, University of California, Irvine, Irvine, CA, United StatesDepartment of Medicine, University of California, Irvine, Irvine, CA, United StatesDepartment of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaDepartment of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaDepartment of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaZhejiang Engineering Research Center of Intelligent Medicine, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaDepartment of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, ChinaDepartment of Radiology, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, ChinaDepartment of Radiological Sciences, University of California, Irvine, Irvine, CA, United StatesDepartment of Medical Imaging and Radiological Sciences, Kaohsiung Medical University, Kaohsiung, TaiwanBackgroundA wide variety of benign and malignant processes can manifest as non-mass enhancement (NME) in breast MRI. Compared to mass lesions, there are no distinct features that can be used for differential diagnosis. The purpose is to use the BI-RADS descriptors and models developed using radiomics and deep learning to distinguish benign from malignant NME lesions.Materials and MethodsA total of 150 patients with 104 malignant and 46 benign NME were analyzed. Three radiologists performed reading for morphological distribution and internal enhancement using the 5th BI-RADS lexicon. For each case, the 3D tumor mask was generated using Fuzzy-C-Means segmentation. Three DCE parametric maps related to wash-in, maximum, and wash-out were generated, and PyRadiomics was applied to extract features. The radiomics model was built using five machine learning algorithms. ResNet50 was implemented using three parametric maps as input. Approximately 70% of earlier cases were used for training, and 30% of later cases were held out for testing.ResultsThe diagnostic BI-RADS in the original MRI report showed that 104/104 malignant and 36/46 benign lesions had a BI-RADS score of 4A–5. For category reading, the kappa coefficient was 0.83 for morphological distribution (excellent) and 0.52 for internal enhancement (moderate). Segmental and Regional distribution were the most prominent for the malignant group, and focal distribution for the benign group. Eight radiomics features were selected by support vector machine (SVM). Among the five machine learning algorithms, SVM yielded the highest accuracy of 80.4% in training and 77.5% in testing datasets. ResNet50 had a better diagnostic performance, 91.5% in training and 83.3% in testing datasets.ConclusionDiagnosis of NME was challenging, and the BI-RADS scores and descriptors showed a substantial overlap. Radiomics and deep learning may provide a useful CAD tool to aid in diagnosis.https://www.frontiersin.org/articles/10.3389/fonc.2021.728224/fullbreast neoplasmscomputer-assisted diagnosisdeep learningmachine learningmagnetic resonance imaging |
| spellingShingle | Jiejie Zhou Jiejie Zhou Yan-Lin Liu Yang Zhang Yang Zhang Jeon-Hor Chen Jeon-Hor Chen Freddie J. Combs Ritesh Parajuli Rita S. Mehta Huiru Liu Zhongwei Chen Youfan Zhao Zhifang Pan Meihao Wang Risheng Yu Min-Ying Su Min-Ying Su BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning Frontiers in Oncology breast neoplasms computer-assisted diagnosis deep learning machine learning magnetic resonance imaging |
| title | BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning |
| title_full | BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning |
| title_fullStr | BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning |
| title_full_unstemmed | BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning |
| title_short | BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning |
| title_sort | bi rads reading of non mass lesions on dce mri and differential diagnosis performed by radiomics and deep learning |
| topic | breast neoplasms computer-assisted diagnosis deep learning machine learning magnetic resonance imaging |
| url | https://www.frontiersin.org/articles/10.3389/fonc.2021.728224/full |
| work_keys_str_mv | AT jiejiezhou biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT jiejiezhou biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT yanlinliu biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT yangzhang biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT yangzhang biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT jeonhorchen biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT jeonhorchen biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT freddiejcombs biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT riteshparajuli biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT ritasmehta biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT huiruliu biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT zhongweichen biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT youfanzhao biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT zhifangpan biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT meihaowang biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT rishengyu biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT minyingsu biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning AT minyingsu biradsreadingofnonmasslesionsondcemrianddifferentialdiagnosisperformedbyradiomicsanddeeplearning |