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

OBJECTIVES: 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. MATERIALS AND METHODS: 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. RESULTS: 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. CONCLUSION: 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.

LN-Net: Perfusion Pattern-Guided Deep Learning for Lymph Node Metastasis Diagnosis Based on Contrast-Enhanced Ultrasound Videos.

OBJECTIVE: The blood flow in lymph nodes reflects important pathological features. However, most intelligent diagnosis based on contrast-enhanced ultrasound (CEUS) video focuses only on CEUS images, ignoring the process of extracting blood flow information. In the work described here, a parametric imaging method for describing blood perfusion pattern was proposed and a multimodal network (LN-Net) to predict lymph node metastasis was designed. METHODS: First, the commercially available artificial intelligence object detection model YOLOv5 was improved to detect the lymph node region. Then the correlation and inflection point matching algorithms were combined to calculate the parameters of the perfusion pattern. Finally, the Inception-V3 architecture was used to extract the image features of each modality, with the blood perfusion pattern taken as the guiding factor in fusing the features with CEUS by sub-network weighting. DISCUSSION: The average precision of the improved YOLOv5s algorithm compared with baseline was improved by 5.8%. LN-Net predicted lymph node metastasis with 84.9% accuracy, 83.7% precision and 80.3% recall. Compared with the model without blood flow feature guidance, accuracy was improved by 2.6%. The intelligent diagnosis method has good clinical interpretability. CONCLUSION: A static parametric imaging map could describe a dynamic blood flow perfusion pattern, and as a guiding factor, it could improve the classification ability of the model with respect to lymph node metastasis.

Establishment and Evaluation of a Rat Model of Medial Malleolar Fracture with Vascular Injury.

OBJECTIVE: A stable animal model was needed to study bone non-union caused by insufficient blood supply, the main object of this paper is to develop a medial malleolar fracture model with controllable arterial vascular injuries in rats for revealing the biochemical mechanism of non-union by insufficient blood supply. METHODS: A total of 18 rats were randomly divided into three equal groups: the Sham group, the Fracture group, and the Fracture + Vascular group. The animals were subjected to unilateral medial malleolar bone fracture and vascular injury using customized molding equipment. The fracture site was scanned by micro-CT, and vascular injury was evaluated by laser Doppler flowmetry (LDF) 24 h after modeling. Histological examination (HE), alkaline phosphatase (ALP) and tartrate-resistant acid phosphatase (TRAP) staining, immunohistochemistry and immunofluorescence were conducted on the medial malleolar fracture tissues of three rats randomly selected from each group 24 h after modeling. Subsequently, to further confirm the success of fracture modeling, the fracture sites of three other rats in each group underwent micro-CT scanning again 6 weeks after surgery. RESULTS: The results of a 24 h micro-CT showed that all rats used to create the fracture models showed controlled injury of the medial malleolus. The model was stable, and the satisfaction of the homemade equipment agreed with the expectation. LDF showed that the blood flow of rats in the Fracture + Vascular group decreased significantly 24 h after fracture injury, while collateral blood flow perfusion increased by 50% on average. The results of HE, ALP and TRAP staining in the medial malleolus showed that the number of osteoblasts (OBs) and osteoclasts (OCs) in the Fracture group increased significantly, but the number of OBs and OCs in the Fracture + Vascular group decreased sharply relative to the number in the Sham group 24 h later. Furthermore, immunohistochemistry and immunofluorescence results showed that the number of neovessels in the Fracture group was significantly increased, while the number of neovessels in the Fracture + Vascular group was significantly decreased, which was consistent with the above results. After 6 weeks of modeling, the micro-CT results showed that the fractures in the Fracture group had healed substantially, while those in the Fracture + Vascular group had not. CONCLUSION: This study provided a reproducible and stable experimental animal model for medial malleolar fractures with arterial injury.

Segmentation of the cervical lesion region in colposcopic images based on deep learning.

Background: Colposcopy is an important method in the diagnosis of cervical lesions. However, experienced colposcopists are lacking at present, and the training cycle is long. Therefore, the artificial intelligence-based colposcopy-assisted examination has great prospects. In this paper, a cervical lesion segmentation model (CLS-Model) was proposed for cervical lesion region segmentation from colposcopic post-acetic-acid images and accurate segmentation results could provide a good foundation for further research on the classification of the lesion and the selection of biopsy site. Methods: First, the improved Faster Region-convolutional neural network (R-CNN) was used to obtain the cervical region without interference from other tissues or instruments. Afterward, a deep convolutional neural network (CLS-Net) was proposed, which used EfficientNet-B3 to extract the features of the cervical region and used the redesigned atrous spatial pyramid pooling (ASPP) module according to the size of the lesion region and the feature map after subsampling to capture multiscale features. We also used cross-layer feature fusion to achieve fine segmentation of the lesion region. Finally, the segmentation result was mapped to the original image. Results: Experiments showed that on 5455 LSIL+ (including cervical intraepithelial neoplasia and cervical cancer) colposcopic post-acetic-acid images, the accuracy, specificity, sensitivity, and dice coefficient of the proposed model were 93.04%, 96.00%, 74.78%, and 73.71%, respectively, which were all higher than those of the mainstream segmentation model. Conclusion: The CLS-Model proposed in this paper has good performance in the segmentation of cervical lesions in colposcopic post-acetic-acid images and can better assist colposcopists in improving the diagnostic level.

Design of lung nodules segmentation and recognition algorithm based on deep learning.

BACKGROUND: Accurate segmentation and recognition algorithm of lung nodules has great important value of reference for early diagnosis of lung cancer. An algorithm is proposed for 3D CT sequence images in this paper based on 3D Res U-Net segmentation network and 3D ResNet50 classification network. The common convolutional layers in encoding and decoding paths of U-Net are replaced by residual units while the loss function is changed to Dice loss after using cross entropy loss to accelerate network convergence. Since the lung nodules are small and rich in 3D information, the ResNet50 is improved by replacing the 2D convolutional layers with 3D convolutional layers and reducing the sizes of some convolution kernels, 3D ResNet50 network is obtained for the diagnosis of benign and malignant lung nodules. RESULTS: 3D Res U-Net was trained and tested on 1044 CT subcases in the LIDC-IDRI database. The segmentation result shows that the Dice coefficient of 3D Res U-Net is above 0.8 for the segmentation of lung nodules larger than 10 mm in diameter. 3D ResNet50 was trained and tested on 2960 lung nodules in the LIDC-IDRI database. The classification result shows that the diagnostic accuracy of 3D ResNet50 is 87.3% and AUC is 0.907. CONCLUSION: The 3D Res U-Net module improves segmentation performance significantly with the comparison of 3D U-Net model based on residual learning mechanism. 3D Res U-Net can identify small nodules more effectively and improve its segmentation accuracy for large nodules. Compared with the original network, the classification performance of 3D ResNet50 is significantly improved, especially for small benign nodules.