DCES-PA: Deformation-controllable elastic shape model for 3D bone proliferation analysis using hand HR-pQCT images.

Area of Science:

• Medical imaging analysis
• Rheumatology
• Biomedical engineering

Background:

• Bone proliferation is a key feature in inflammatory rheumatic diseases.
• Current high-resolution peripheral quantitative computed tomography (HR-pQCT) analysis is limited by subjective variations and slice inconsistencies.
• Effective automated tools for bone proliferation detection are scarce.

Purpose of the Study:

• To introduce an unsupervised method for 3D bone proliferation analysis (DCES-PA).
• To overcome limitations of existing shape analysis methods in proliferation detection.
• To improve the accuracy and sensitivity of proliferation annotation in medical imaging.

Main Methods:

• Developed the Deformation-Controllable Elastic Shape model (DCES-PA) integrating pathological knowledge and statistical shape analysis.
• Regularized first and second-order derivatives of the displacement field, decomposing vector fields by deformation types (shearing, scaling, bending).
• Incorporated intrinsic and extrinsic surface curvatures to control high-curvature regions and used varifold distances for correspondence-free analysis.

Main Results:

• DCES-PA demonstrated improved accuracy compared to state-of-the-art shape-based methods in proliferation analysis.
• The method achieved highly sensitive proliferation annotations on simulated and real clinical datasets.
• DCES-PA effectively assists physicians in analyzing bone proliferation in rheumatic diseases.

Conclusions:

• DCES-PA offers a robust and accurate unsupervised approach for 3D bone proliferation analysis.
• The method addresses limitations in current HR-pQCT annotation by reducing subjectivity and improving consistency.
• DCES-PA has the potential to significantly aid in the diagnosis and management of inflammatory rheumatic diseases.