| dc.description.abstract | Passive laxity envelopes can provide understanding of the knee’s comprehensive constraints and range of motion as well as insights not practically obtained in vivo. This research was to study variations in passive knee envelope predictions and provide recommendations for better data collection. The Experimental Joint Biomechanics Research Laboratory at the University of Kansas uses a radial basis function to generate a unified passive knee envelope and then a sequential down sampling process to study the variation in knee envelope predictions. This study had three objectives: analyze areas of high envelope variation from 13 knee predictions to better understand those variations; develop an envelope quality score to determine if the quantity and distribution of an existing data set will generate a low variation, high confidence knee envelope for a target isoload; and develop a new investigator feedback system to generate a higher quality envelope prediction. For objective 1, data sets from unified passive laxity envelopes for 13 cadaver specimens were analyzed by taking the average and standard deviation of the variations and mapping them onto the average envelope shape to compare locations of higher variation. No consistent location of variation was found, indicating the variations were not likely due to investigator instructions. This analysis did show the consistent collection of data at lower isoloads (2000 Nmm) and the under collection of data points at higher isoloads (8000-10000 Nmm), where the ligaments have a greater influence on the knee’s motion. Objective 2 was to develop an envelope quality score to evaluate if an existing data set will generate a confident envelope prediction for a given isoload. Recognizing no consistent pattern in the location of variations and that the radial basis function used to interpolate data sets into a unified envelope prediction weighs data closest to the target isoload more heavily, led to the analysis of a given data set relative to a target isoload. Based on these findings, an envelope quality scoring calculation was developed based on the distance, quantity, and distribution of data points relative to a given isoload. The data points were totaled if points were located within the Euclidean norm distance of 0.25 from each vertex in the envelope isoload surface. The resulting score helps investigators assess if an existing data set will produce an envelope prediction with less variation and greater confidence. Objective 3 was to create a new investigator feedback system based on tailoring data collection to the RBF process and providing real time guidance for data collection. This was done using a MATLAB program that calculates how collected data are distributed around the target isoload and graphically displays data points on a color-coded isoload and an RBF prediction feed for the developing envelope displayed on the feedback screen. Color coding shows the investigator areas where sufficient data around the target isoload has been collected, and areas where additional data points are still needed. Displaying this in near real time (updated in 1 second increments) gives the investigator opportunity to make needed adjustments during initial data collection. Eliminating possible sources of data collection inaccuracies like manipulation motions, differences in knee kinematics or demographics could further improve envelopes. Future work might determine best isoloads for specific studies, add the envelope quality score to the investigator feedback screen, or allow the investigator to change the target isoload as they collect data. Finally, increasing program efficiency for quicker computation may increase its potential use in clinical settings. | |
