Search Filters

  • Media Source
  • Presentation Format
  • Media Type
  • Media Year
  • Language
  • Diagnosis / Condition
  • Diagnosis Method
  • Patient Populations
  • Treatment / Technique

A Novel Quantitative Assessment of Bone Tendon Junction Healing in Patients after ACL Reconstruction by High Resolution Peripheral Computer Tomography: The Development of a Deep-Learning System

2023 Congress Paper Abstracts
myRating

A Novel Quantitative Assessment of Bone Tendon Junction Healing in Patients after ACL Reconstruction by High Resolution Peripheral Computer Tomography: The Development of a Deep-Learning System

Michael Ong, MBChB(UK), BSc(UK), MRCS(Edin), MSc(CUHK), FRCSEd, HONG KONG Patrick S. H. Yung, MBChB, FRCS(Orth), FHKCOS, FHKAM, FRCS, HONG KONG

The Chinese University of Hong Kong, Shatin, HONG KONG


2023 Congress   ePoster Presentation   2023 Congress   Not yet rated

 

Anatomic Location

Anatomic Structure

Diagnosis / Condition

Treatment / Technique

Ligaments

ACL

Patient Populations

Diagnosis Method

Sports Medicine


Summary: This study aimed to devise a novel quantitative assessment of peri-tunnel bone shell size and determined its correlation to functional recovery after ACLR


Introduction

Graft healing after ACL reconstruction is a major challenge determining the surgical outcome. Bone shell size reflects bone-tendon junction healing progress, which may contribute to functional recovery after ACLR. This study aimed to devise a novel quantitative assessment of peri-tunnel bone shell size and determined its correlation to functional recovery after ACLR.

Methods

24 patients received ACLR were imaged by high-resolution-CT. Knee functions were evaluated by International Knee Documentation Committee (IKDC) score. Quadriceps muscle atrophy and quadriceps elastic properties were evaluated by ultrasound imaging and shear wave elastography. U-Net was further employed for the development of an image-based machine-learning algorithm on detecting the tunnel size, using 625 CT images (500 for training and 125 for validation).

Results

Both tunnel size and bone shell formation altered along the depth of tunnel. Bone shell formation was associated with time post operation, and variations in femoral tunnel angle (r=0.567, p=0.018). Regression analysis showed that bone shell formation in femoral tunnel (standardized ß= 0.440, p=0.022) and quadricep atrophy (standardized ß=-0.400, p=0.036) were significantly associated with IKDC scores (adjusted R2 = 0.376). For the machine learning algorithm, the mean pixel accuracy and mean intersection of union values of the algorithm were 0.95 and 0.77, with the precision and recall 81.3% and 95% respectively.

Discussion

Bone shell formation at graft tunnel interface may indicate favourable tissue reactions inside bone tunnels after ACLR, with its positive correlation to functional recovery as shown by IKDC scores. A gradual increase in bone shell formation after ACLR was noticed but the variations were large at long time points post operation. The spatial changes of bone shell formation along femoral and tibial tunnels are different, probably due to variations in local mechanical environment as well as availability of progenitor cells. The extent of bone shell formation in femoral tunnel may be affected by tunnel angle and graft type. Referring to the findings in animal studies, the bone shell at graft tunnel interface reflected the extent of bone tendon junction healing, which promote the mechanical stability of the graft. The evaluation of mature graft incorporation by measuring bone shell formation will help to avoid premature return-to-play with higher risk of graft failure, or can be utilized as a monitoring tool to guide customized rehabilitation program. Biological modulation enhancing bone-tendon junction healing can also be. While for the novel machine learning model developed by us, it demonstrated high accuracy and predictability, comparable to consultant level assessment.

Conclusion

The measurement of bone shell formation by high-resolution-CT is suitable to assess the graft healing after ACLR. Based on our novel method, further machine learning model will provide an accurate imaging assessment and monitoring tool for ACL injury patients.


More 2023 ISAKOS Congress Content