During my PhD, my research was focused on using statistical shape modeling techniques to improve the outcome of Total Knee Arthroplasty (TKA).

Knee is the biggest and one of the most complicated joints in the human body. It has to be capable of supporting 2.5 times body weight and at the same time it has to have 6 degrees of freedom. The knee has the highest probability of injury, and osteoarthritis (OA) is the most common disease for this complex joint.

Osteoarthritis is a degenerative disease that is characterized by loss or damage in articular cartilage, which causes the functional disability of the knee. Approximately 27 million Americans suffer from OA and it is predicted that by the end of 2030, 25% of the U.S. adult population will suffer from OA. The symptoms of this disease are pain, stiffness, and restricted motion.

To treat advanced osteoarthritis disease, Total Knee Arthroplasty (TKA) has been proposed by Shiers in 1954. In this surgery, diseased parts of the femur and tibiae are trimmed and replaced with implants. This prosthesis consists of three (in case of patellar resurfacing, otherwise two) components: femoral component, tibial insert, tibial part, and patellar component. Tibial and femoral components are made of stainless steel and the others are made of UHMWPE (Ultra high molecular weight polyethylene).

One open question in TKA is whether to resurface the patella or not. Considering the exponential increase in the number of TKA surgeries in past few years, and increasing number of younger patients who need TKA, it is significantly important to develop a method for predicting the outcome of TKA for each specific patient. Sometimes after TKA with no PC, the patient complains about anterior knee pain. In these cases, surgeon suggests a second surgery to resurface the patella. In contrast, the probability of complication after TKA with patella resurfacing is quite low, but in case of any problem the side effects of this surgery are much more severe than no-PC. The main probable complications for the patient after patellar resurfacing are osteonecrosis, patella fracture, implant failure, polyethylene wear damage, component loosening, and component dissociation. To predict the best patella resurfacing option for each patient and avoid such complications, an accurate model of the knee is needed.

One of the major problems to build the patient-specific model concerns the extraction of the biomechanical parameters from pre-operative images. This task is usually time consuming and some of the mechanical parameters could not be extracted from clinical images. For instance, bone mechanical properties can be precisely derived from μCT dataset, however this modality is not available in clinical methods. The idea of this project is to employ statistical information obtained on a pre-existing population to derive the patient-specific FE model.