A complete-pelvis segmentation framework for image-free total hip arthroplasty (THA): methodology and clinical study

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Orthopedic Surgery

Background:

• Complete-pelvis segmentation in antero-posterior pelvic radiographs is essential for creating patient-specific 3D models.
• These models are critical for surgical planning and postoperative assessment in image-free navigation for total hip arthroplasty (THA).

Purpose of the Study:

• To develop and present a fast, robust, and accurate framework for complete pelvis segmentation.
• To enable precise 3D model creation for image-free navigation in total hip arthroplasty.

Main Methods:

• A two-module framework utilizing a three-stage method for hemi-pelvis delineation based on statistical appearance and shape models.
• Incorporation of anatomy-specific information processing techniques to manage complex pelvic structures.
• Reflection of the delineated left hemi-pelvis to initialize and segment the right hemi-pelvis using the same three-stage method.

Main Results:

• Achieved a mean segmentation accuracy of 1.61±0.68 mm across two experimental datasets (143 and 40 AP radiographs).
• Clinical study demonstrated average accuracy of 1.2°±0.9° for anteversion and 1.6°±1.4° for inclination in postoperative acetabular cup orientation assessment.
• Segmentation of each radiograph was completed in under one minute.

Conclusions:

• The proposed framework shows promising clinical applicability for image-free total hip arthroplasty (THA).
• Preliminary results suggest high accuracy in pelvis segmentation and postoperative assessment.
• Further validation is recommended to fully establish its clinical utility.