Deep Learning for Classification of Bone Lesions on Routine MRI
Feyisope R. Eweje, SB,
Bingting Bao, MS,
Jing Wu, MD,
Deepa Dalal, MBBS,
Wei-hua Liao, MD,
Yu He, MD,
Yongheng Luo, MD,
Shaolei Lu, MD, PhD,
Paul Zhang, MD,
Xianjing Peng, MD,
Ronnie Sebro, MD, PhD,
Harrison X. Bai, MD,
Lisa States, MD
Affiliations
Feyisope R. Eweje, SB
Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, PA, 19104, USA; Perelman School of Medicine at University of Pennsylvania, Philadelphia, PA, 19104, USA
Bingting Bao, MS
Department of Radiology, Second Xiangya Hospital of Central South University, Changsha, Hunan, 410011, China
Jing Wu, MD
Department of Radiology, Second Xiangya Hospital of Central South University, Changsha, Hunan, 410011, China
Deepa Dalal, MBBS
Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, PA, 19104, USA
Wei-hua Liao, MD
Department of Radiology, Xiangya Hospital of Central South University, Changsha, Hunan, 410008, China
Yu He, MD
Department of Radiology, Second Xiangya Hospital of Central South University, Changsha, Hunan, 410011, China
Yongheng Luo, MD
Department of Radiology, Second Xiangya Hospital of Central South University, Changsha, Hunan, 410011, China
Shaolei Lu, MD, PhD
Department of Pathology and Laboratory Medicine, Warren Alpert Medical School of Brown University, Providence, RI, 02903, USA
Paul Zhang, MD
Department of Pathology and Laboratory Medicine, Hospital of the University of Pennsylvania, Philadelphia, PA, 19104, USA
Xianjing Peng, MD
Department of Radiology, Xiangya Hospital of Central South University, Changsha, Hunan, 410008, China; Xianjing Peng, Department of Radiology, Xiangya Hospital of Central South University, Changsha, Hunan, 410008, China.
Ronnie Sebro, MD, PhD
Mayo Clinic Radiology, Jacksonville, FL, 32224, USA
Harrison X. Bai, MD
Department of Diagnostic Imaging, Rhode Island Hospital, Providence, RI, 02903, USA; Harrison X. Bai, Department of Diagnostic Imaging, Warren Alpert Medical School of Brown University, Providence, Rhode Island 02912, USA. Phone: (401)793- 4480; Fax: (401)793-4444.
Lisa States, MD
Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, PA, 19104, USA; Corresponding authors: Lisa States, Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, PA, 19104, USA. Phone: (267)425-7146; Fax: (267)425-7068.
Background: Radiologists have difficulty distinguishing benign from malignant bone lesions because these lesions may have similar imaging appearances. The purpose of this study was to develop a deep learning algorithm that can differentiate benign and malignant bone lesions using routine magnetic resonance imaging (MRI) and patient demographics. Methods: 1,060 histologically confirmed bone lesions with T1- and T2-weighted pre-operative MRI were retrospectively identified and included, with lesions from 4 institutions used for model development and internal validation, and data from a fifth institution used for external validation. Image-based models were generated using the EfficientNet-B0 architecture and a logistic regression model was trained using patient age, sex, and lesion location. A voting ensemble was created as the final model. The performance of the model was compared to classification performance by radiology experts. Findings: The cohort had a mean age of 30±23 years and was 58.3% male, with 582 benign lesions and 478 malignant. Compared to a contrived expert committee result, the ensemble deep learning model achieved (ensemble vs. experts): similar accuracy (0·76 vs. 0·73, p=0·7), sensitivity (0·79 vs. 0·81, p=1·0) and specificity (0·75 vs. 0·66, p=0·48), with a ROC AUC of 0·82. On external testing, the model achieved ROC AUC of 0·79. Interpretation: Deep learning can be used to distinguish benign and malignant bone lesions on par with experts. These findings could aid in the development of computer-aided diagnostic tools to reduce unnecessary referrals to specialized centers from community clinics and limit unnecessary biopsies. Funding: This work was funded by a Radiological Society of North America Research Medical Student Grant (#RMS2013) and supported by the Amazon Web Services Diagnostic Development Initiative.
