Advancing musculoskeletal tumor diagnosis: Automated segmentation and predictive classification using deep learning and radiomics

• Published in: Computers in biology and medicine Vol. 175; p. 108502
• Main Authors: Wang, Shuo, Sun, Man, Sun, Jinglai, Wang, Qingsong, Wang, Guangpu, Wang, Xiaolin, Meng, Xianghong, Wang, Zhi, Yu, Hui
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.06.2024; Elsevier Limited
• Subjects: Adult; Aged; Algorithms; Automation; Bone cancer; Bone Neoplasms - classification; Bone Neoplasms - diagnostic imaging; Bone tumors; Brain cancer; Cancer therapies; Classification; Datasets; Deep Learning; Diagnosis; Feature maps; Female; Heterogeneity; Humans; Image contrast; Image enhancement; Image Interpretation, Computer-Assisted - methods; Image processing; Image reconstruction; Image segmentation; Internal Medicine; Lesions; Machine learning; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medical diagnosis; Middle Aged; Muscle Neoplasms - diagnostic imaging; Musculoskeletal tumors; Other; Patients; Radiomics; Segmentation; Soft tissues; Statistical analysis; Tumors; Deep learning; Musculoskeletal tumors; Segmentation; Classification; Radiomics; deep learning; radiomics; segmentation; classification
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2024.108502

Musculoskeletal (MSK) tumors, given their high mortality rate and heterogeneity, necessitate precise examination and diagnosis to guide clinical treatment effectively. Magnetic resonance imaging (MRI) is pivotal in detecting MSK tumors, as it offers exceptional image contrast between bone and soft tissue. This study aims to enhance the speed of detection and the diagnostic accuracy of MSK tumors through automated segmentation and grading utilizing MRI. The research included 170 patients (mean age, 58 years ±12 (standard deviation), 84 men) with MSK lesions, who underwent MRI scans from April 2021 to May 2023. We proposed a deep learning (DL) segmentation model MSAPN based on multi-scale attention and pixel-level reconstruction, and compared it with existing algorithms. Using MSAPN-segmented lesions to extract their radiomic features for the benign and malignant classification of tumors. Compared to the most advanced segmentation algorithms, MSAPN demonstrates better performance. The Dice similarity coefficients (DSC) are 0.871 and 0.815 in the testing set and independent validation set, respectively. The radiomics model for classifying benign and malignant lesions achieves an accuracy of 0.890. Moreover, there is no statistically significant difference between the radiomics model based on manual segmentation and MSAPN segmentation. This research contributes to the advancement of MSK tumor diagnosis through automated segmentation and predictive classification. The integration of DL algorithms and radiomics shows promising results, and the visualization analysis of feature maps enhances clinical interpretability. •We have designed a novel DL segmentation model based on multi-scale attention and pixel-level segmentation. Our method was compared with other state-of-the-art segmentation algorithms to demonstrate its segmentation accuracy.•We performed radiomics feature extraction on the segmented lesion regions, followed by feature selection, and applied a classification model for benign-malignant prediction. Furthermore, we compared four different classifiers and obtained high classification accuracy.•We analyzed feature maps of MSK tumors to demonstrate the effectiveness of our segmentation method. Additionally, we discussed some interesting findings, which may enhance the interpretability of AI in clinical medicine.