Surgical planning of pelvic tumor using multi-view CNN with relation-context representation learning

Limb salvage surgery of malignant pelvic tumors is the most challenging procedure in musculoskeletal oncology due to the complex anatomy of the pelvic bones and soft tissues. It is crucial to accurately resect the pelvic tumors with appropriate margins in this procedure. However, there is still a lack of efficient and repetitive image planning methods for tumor identification and segmentation in many hospitals. In this paper, we present a novel deep learning-based method to accurately segment pelvic bone tumors in MRI. Our method uses a multi-view fusion network to extract pseudo-3D information from two scans in different directions and improves the feature representation by learning a relational context. In this way, it can fully utilize spatial information in thick MRI scans and reduce over-fitting when learning from a small dataset. Our proposed method was evaluated on two independent datasets collected from 90 and 15 patients, respectively. The segmentation accuracy of our method was superior to several comparing methods and comparable to the expert annotation, while the average time consumed decreased about 100 times from 1820.3 seconds to 19.2 seconds. In addition, we incorporate our method into an efficient workflow to improve the surgical planning process. Our workflow took only 15 minutes to complete surgical planning in a phantom study, which is a dramatic acceleration compared with the 2-day time span in a traditional workflow.

BMSMM-Net: A Bone Metastasis Segmentation Framework Based on Mamba and Multiperspective Extraction

RATIONALE AND OBJECTIVES

Metastatic bone tumors significantly reduce patients' quality of life and expedite cancer spread. Traditional diagnostic methods rely on time-consuming manual annotations by radiologists, which are prone to subjectivity. Employing deep learning for rapid, precise segmentation of bone metastases can greatly improve patient outcomes and survival. However, accurate segmentation remains challenging due to the diverse and complex nature of osteoblastic, osteolytic, or mixed lesions.

MATERIALS AND METHODS

In this study, we presented a novel segmentation framework, termed BMSMM-Net, tailored specifically for the detection of bone metastases. The framework integrates our newly proposed Bottleneck Gating Mamba layer (BGM) into the network backbone, enhancing the long-range dependencies in the depth feature maps. Additionally, we designed a Skip-Mamba (SKM) module on the skip connections to facilitate long-range modeling during multi-scale feature fusion. Furthermore, a Multi-Perspective Extraction (MPE) module was employed in the feature extraction phase, utilizing three different sizes of convolutional kernels to enhance sensitivity to bone metastases.

RESULTS

Our framework was evaluated on the BM-Seg dataset through comparative and ablation studies. It achieved F1 scores of 91.07% and 95.17% for segmenting bone metastases and bone regions, respectively, along with mIoU scores of 83.60% and 90.78%, BMSMM-Net provides high-performance segmentation of bone metastases. Additionally, it maintains good computational efficiency compared to existing models.

CONCLUSION

The BMSMM-Net framework, integrating BGM, SKM, and MPE modules, effectively addresses the segmentation challenges of bone metastases. It significantly enhances accuracy, outperforms advanced existing methods, and maintains lower complexity, making it suitable for clinical application.