Bone reconstruction from X-ray images using statistical shape models
Research type
Research Study
Full title
Lower limb bone shape reconstruction from planar images using statistical shape and appearance models
IRAS ID
263732
Contact name
Anthony MJ Bull
Contact email
Sponsor organisation
Imperial College
Duration of Study in the UK
0 years, 10 months, 15 days
Research summary
Summary of Research
Orthopaedic reconstruction after trauma requires knowledge of the bone shape in a healthy state to guarantee best outcomes of the intervention. Typically, the contralateral side is used to create a geometric template. The three dimensional (3D) shape of the contralateral side is typically obtained using X-ray computed tomography (CT) which exposes the patient to ionising radiation. This could be avoided if other methods for accurate reconstruction of the bone shape would be available. Further, this technique can only be used if the contralateral side is intact and does not show any deformations that would indicate an asymmetry. Therefore, in cases of bilateral fractures or musculoskeletal conditions leading to asymmetry in the limbs, the contralateral side cannot be used as a template.
We have developed techniques to predict the original shape of a bone when it has been disrupted by extensive fracture or partial amputation from a simple X-ray. These techniques use statistical shape and appearance models, which we tested on cadaveric CT scans from the publicly available Digital Korean dataset. The statistical shape and appearance model represents the average shape and a compact representation of the main variations of the underlying dataset.
To use the technique in clinical practice, the method should be tested using in-vivo medical images. Therefore, we aim to obtain CT and planar X-ray images from the records of the Department of Radiology of the Imperial College NHS Trust to create new statistical shape and appearance models which represents the patient group considered for reconstruction. The model will then be used to reconstruct the 3D shape of the intact bone. The reconstructed shape will be compared to the 3D shape of the contralateral side, if available. Alternatively, the accuracy will be evaluated by comparing the projection of the shape model to X-ray images of the contralateral side.Summary of Results
Surgery to bones after severe injury seeks to recreate the intact shape of the bone. Therefore, the surgeon needs to know what the original, pre-injury, shape is. This is hard, because in some injuries, the bone is missing or severely deformed. If the bone injured is of the arm or leg, then the shape of the bone on the other side is used to estimate the pre-injury shape. This is obtained using three-dimensional (3D) medical imaging, such as from X-ray computed tomography (CT) scans. These scans use radiation and could be avoided if other methods for accurate reconstruction of the bone shape would be available. Also, this can only be done if the other side is intact and doesn't have any injury. Therefore, in cases where there is an injury on both sides, the other side cannot be used as a guide.We have developed methods to predict the original shape of a bone from a simple X-ray of the injured bone. In this study, we tested these methods using medical images from patients, and the bone defects were simulated by removing a portion of the bone from the lower end of the femur.
This study shows that the reconstructions from partly available planar images using our developed method, statistical shape and appearance models, had an accuracy which would support their potential use in orthopaedic reconstruction after trauma, in examples such as a template for the reconstruction using AR systems, creating personalised instrumented guides for joint reconstruction and construction of patient-specific implants.REC name
Wales REC 5
REC reference
20/WA/0073
Date of REC Opinion
2 Mar 2020
REC opinion
Favourable Opinion