Related Experiment Video
Updated: Jan 7, 2026
07:43
In Vivo Quantification of Hip Arthrokinematics during Dynamic Weight-bearing Activities using Dual Fluoroscopy
Published on: July 2, 2021
3.4K
MPLDM: Multi-modal prosthetic loosening diagnostic model for total hip arthroplasty
Xiao Chen1, Pang Lyu2, Wencheng Han3
1Department of Orthopedics, People's Hospital of Zhengzhou University, Henan Provincial People's Hospital, People's Hospital of Henan University, Zhengzhou, 450003, Henan, China.
Medical Image Analysis
|December 30, 2025
Summary
This study introduces a Multi-modal Prosthetic Loosening Diagnostic Model (MPLDM) for diagnosing complications after total hip arthroplasty (THA). The model accurately identifies aseptic loosening and periprosthetic joint infection (PJI) using integrated patient data.
Area of Science:
- Orthopedic surgery
- Medical imaging
- Artificial intelligence in medicine
Background:
- Total hip arthroplasty (THA) is a common procedure with potential complications like aseptic loosening and periprosthetic joint infection (PJI).
- Accurate diagnosis of these complications is challenging and requires integrating diverse clinical data sources.
- Existing diagnostic methods may not fully leverage multi-modal patient information.
Purpose of the Study:
- To develop and validate a novel Multi-modal Prosthetic Loosening Diagnostic Model (MPLDM) for improved diagnosis of THA complications.
- To enhance the utilization of integrated clinical data, including imaging, medical records, and lab results.
- To create a robust diagnostic tool that can handle missing data modalities.
Main Methods:
- Developed the MPLDM, utilizing four independent encoders for different data types (X-ray, CT, medical records, lab tests).
- Incorporated attention mechanisms to improve information interaction between modalities.
- Implemented a multi-modal fusion module for robust prediction and designed the model for missing data scenarios.
- Created and utilized a comprehensive multi-modal dataset for training and evaluation.
Main Results:
- The MPLDM achieved excellent diagnostic performance with mean precision, recall, and F1-scores of 0.9140, 0.8353, and 0.8645, respectively.
- Demonstrated superior accuracy and robustness compared to single-modal and other multi-modal baseline models.
- Validated the model's effectiveness in diagnosing both aseptic loosening and PJI-induced loosening.
Conclusions:
- The MPLDM offers a significant advancement in diagnosing aseptic loosening and PJI after THA.
- The model's ability to integrate multi-modal data and handle missing inputs enhances its clinical applicability.
- This approach holds promise for improving patient outcomes through more accurate and timely diagnosis of THA complications.
