MODELING OF PATELLA HEIGHT WITH DISTAL FEMUR AND PROXIMAL TIBIA REFERENCE POINTS WITH ARTIFICIAL NEURAL NETWORK
Abstract
The patellofemoral joint is one of the parts of the knee extension mechanism that plays a role in
the stability of the knee by enlarging the force arm of the quadriceps muscle and changing the
direction of the muscle strength. For the entire knee joint to perform its task painlessly and
functionally, the positions and strength of the muscles, the strength of the ligaments, and their
reaction to movement must be compatible. The Insall–Salvati (Ins-Sal) index is useful for
showing changes in patellar height produced by repositioning the tibial plateau, in other words,
showing changes in patellar tendon length. Patella height is an important value to be taken into
account in knee prosthesis surgery, tibial osteotomy, and anterior cruciate ligament reconstruction. The morphometric relationship between the reference measurements of the distal
femur and proximal tibia and the position of the patella will be useful in determining the natural
anatomy. In this study, we aimed to determine the relationship between patella height and
distal femur and proximal tibia reference areas by using the arti¯cial neural network method as
an alternative approach method. In order to assess the performance of the estimation of the InsSal index, the four ANN model with six input combinations which included age, gender and the
reference measurements for the right and left sides have been constructed and tested. The MSE
and r values are calculated for every four models for the training and test phase. The results
show that the proposed approach for modeling of relation between reference measurements and
the Ins-Sal index is a powerful approach.