Automated Generation of Patient Specific Models for Visual and Haptic Simulation of Hip Fracture Surgery
The goal of this project is an auto-generated patient specific model for haptic and visual simulation of hip fracture surgery. Osteoporotic fractures constitute a problem of increasing clinical importance. A problem with the cervical type of hip fracture is the great risk of complications. A patient-specific simulation model would enable the surgeon to perform simulated surgery on the patient. Instead of discussing alternative techniques using plain X-ray films, the surgeon would have the chance to test several operative approaches, resulting in a safer and more rapid real operation. In addition, these models would be useful in the training of surgeons and development of new techniques. The first step in the generation of the model is segmentation of the bone in a CT-volume. Then, the local bone density will be estimated from the CT-data. The resulting information will then be converted to fit models suitable for visual and haptic simulation.
Magnus Borga , Assoc. Prof.
Johanna Pettersson , Tek. lic.
Hans Knutsson , Prof.
Örjan Smedby , Prof.
Ola Whalström , Assoc. prof, MD
Bo Tillander , PhD, MD
- Former Staff:
- Project Description:
Background and motivation
Osteoporotic fractures constitute a problem of increasing clinical importance. In parallel with the aging population and increasing prevalence of osteoporosis, the incidence of hip fractures is expected to double in the next 20 years. An additional problem with the cervical type of hip fracture is the greater risk of complications, which often make prosthetic replacement necessary, resulting in greater risks and increased costs, for both acute treatment and future care. Thus, every improvement in the surgical technique will be valuable.
Based on diagnostic information including history, physical examination and plain radiography, the surgeon decides whether and how surgery should be performed. This clinical decision, based on previous experience, is often difficult due to the great variation in anatomy and fracture extension.
A patient specific simulation model, incorporating pre-operative information on the individual patient, would enable the surgeon to perform simulated surgery on the patient. Instead of discussing alternative techniques using plain X-ray films, the surgeon would have the chance to test several operative approaches, resulting in a safer and more rapid operation. In addition, these models would be useful in the training of surgeons and development of new techniques.
This research project aims at generating these patient specific models using an automatic segmentation technique, i.e. a method where no, or at least very little, user interaction is required. The motivation for developing such a method is that this method is beneficial, and also essential, in many aspects. For a surgeon to be able to perform pre-operative planning on individual patients in the simulator we must be able to generate a model of the specific patient's anatomy quite rapidly. Automatic generation of the virtual models are advantageous also in other cases. Here are some suggestions:
* Easy to build patient libraries, which reduces costs and opens up for a more versatile training environment.
* Facilitates validation of simulator systems.
* Facilitates population studies for research purposes.
* Makes it easier for manufacturers to try out new equipment such as implants.
20 patients with cervical hip fractures will go through a computer tomography (CT) examination in addition to the conventional X-ray examination in order to generate a database for the project. 20 cases are considered enough to obtain a sufficient variation of patient anatomies and fracture types in order for the method to be applicable on a larger amount of patients further on.
The concrete result of the project is a framework for automatic generation of patient specific model for haptic and visual simulation. The research focuses on methods for extracting properties, such as shape, structure and density, from CT data of the femur region, and from these properties generate models for the simulator.
Method for automatic segmentation
The segmentation involves two issues; separating the bone from the surrounding soft tissue and separating different parts of the bone from each other. The goal with the segmentation process is a surface model representing the anatomy of a specific patient that can be incorporated into the simulator system and used for practice or pre-operative planning.
The method that has been employed for the automatic segmentation process is the Morphon method [1-3]. This is a general non-rigid registration method that has been developed at our department. The method takes a prototype image/volume and deforms it until it fits the target image/volume. The registration process is an iterative algorithm where each repetition passes through the following steps:
* Displacement estimation
* Deformation field accumulation and regularisation
* Prototype deformation
The algorithm is initiated on a coarse resolution scale to catch large, global displacements, and continues to finer resolution scales until an optimal match between the two datasets is obtained. For a more detailed description of this method the reader is referred to references [1-3].
As mentioned above the algorithm works on both 2D and 3D datasets. In this project we work with 3D CT volumes collected from patients with this type of fracture. The prototype is a volume where each voxel is labeled such that it belongs either to the pelvis, the femur or the background. By deforming this segmented representation of the hip area until it fits the corresponding structure in a specific patient we obtain a volume with a labeled representation of that specific patient's anatomy. From the labeled volume it is easy to generate surface models of the pelvis and femur as two separate objects.
Some more results
The following images shows views from the simulator system. The top left image is the generic model that is originally incorporated in the system, the middle column shows two models generated from two different patients using the method described above, and the right column shows snapshots of a person working in the simulator on these models.
The models that we have shown here represents bone without fractures. The goal within the project is, however, to automatically segment femurs with cervical fractures and create models of these for the simulator. This work is in progress and results will be shown here later on.
Demonstration of the Morphon method. (~4Mb)
This is a small movie that demonstrates how the Morphon method can be used for segmentation purposes. A 2D prototype image consisting of a very simple object, a black circle on white background, is registered to the heart wall in an image from an ultrasound sequence.
Animation of hip area. (~6Mb)
This is a small animation demonstrating the above concept. The pelvis is separated and removed from the femur in order to be able to distinguish between the two objects and to give the surgeon the possibility to look at the fracture area without the obscuring pelvic bone. This dataset has been manually segmented slice by slice. This introduces some artifacts due to the difficulty in being consistent when determining which pixels belong to the pelvic bone and the femoral bone respectively. Also note that this dataset does not contain any fractures.
|||H. Knutsson, M.Andersson, "Morphons: Segmentation using Elastic Canvas and Paint on Priors", in Proceedings of the IEEE International Conference on Image Processing, Genova, Italy, September 2005|
|||A.Wrangsjö, J.Pettersson, H.Knutsson, "Non-Rigid Registration using Morphons", in Proceedings of the 14th Scandinavian conference on image analysis, Joensuu, Finland, June 2005|
|||J.Pettersson, H.Knutsson, M.Borga, "Generation of Patient Specific Bone Models From Volume Data Using Morphons", in Proceedings of the 13th Nordic-Baltic conference on biomedical engineering and medical physics, Umeå, Sweden, June 2005|
Pettersson J, Palmerius KL, Knutsson H, Wahlström O,
Tillander B, Borga M. Simulation of patient specific cervical hip
fracture surgery with a volume haptic interface. IEEE Transactions on
Biomedical Engineering 2008;55(4):1255-65.Rydell J, Knutsson H, Pettersson J, Johansson A,
Farnebäck G, Dahlqvist O, Lundberg P, Nyström F, Borga M. Phase
sensitive reconstruction for water/fat separation in MR imaging using
inverse gradient. In International Conference on Medical Image
Computing and Computer-Assisted Intervention (MICCAI'07). Brisbane,
Pettersson J, Knutsson H, Borga M. Segmentation and registration with the morphon method, four different applications. In Proceedings of the SSBA Symposium on Image Analysis. Linköping, Sweden, 2007;.Pettersson J, Knutsson H, Borga M. Non-rigid registration for automatic fracture segmentation. In Proceedings of the IEEE International Conference on Image Processing. Atlanta, USA: IEEE, 2006;.
Pettersson J, Knutsson H, Borga M. Automatic hip bone segmentation using non-rigid registration. In Proceedings of the IEEE International Conference on Pattern Recognition. Hong Kong, China: IEEE, 2006;.
Pettersson J, Knutsson H, Nordqvist P, Borga M. A hip surgery simulator based on patient specific models generated by automatic segmentation. In Proceedings of the Medicine Meets Virtual Reality Conference (MMVR'06). Long Beach, California, USA, 2006;431-436.Pettersson J, Knutsson H, Borga M. Generation of patient specific bone models from volume data using morphons. In Proceedings of the 13th Nordic-Baltic conference on biomedical engineering and medical physics (NBC'05). Umeå: NBC, 2005;.
Wrangsjö A, Pettersson J, Knutsson H. Non-rigid registration using morphons. In Proceedings of the 14th Scandinavian conference on image analysis (SCIA'05). Joensuu, 2005;
Pettersson J, Borga M, Andersson M, Knutsson H. Volume morphing for segmentation of bone from 3d data. In Proceedings of the SSBA Symposium on Image Analysis. Malmö, Sweden: SSBA, 2005;.Pettersson J, Borga M, Knutsson H. Some issues on the segmentation of the femur in CT data. In Proceedings of the SSBA Symposium on Image Analysis. 2004;158 - 161.