Room: ePoster Forums
Purpose: To explore how to establish the classification and selection method of universal vaginal cylinder based on the data of 3D printed individual cylinder, in order to improve the doseconformity of vaginal application.
Methods: Retrospective analysis was performed on 60 cases of 3d-printed individual vaginal cylinders. After simply classificate them according to appearance, the contour of each type of vaginal cylinders was matched according to the coordinate system. The average parameters of each type of vaginal cylinders were obtained through the analysis of equidistant cross-section distribution and the length and length of axis under the same coordinate system.MATLAB software was used to build the vaginal model classification system. The core algorithm using in this research is the nearest neighbor algorithm, and the average parameters were taken as the reference for the auto-classification program. Automatic-matching was implemented by the program, and the suitable universal cylinder was selected according to the three-dimensional vaginal contour.
Results: Individualized cylinder data were divided into five types: A(Standard columnar type), B(Flat-wide type), C(Cylinder type), D(Horn-type), E(Lateral-curvature type), and F(Special type), The proportions of each are 36.7%, 8.3%, 16.7%, 18.3%, 11.67%, and 8.3%, respectively.Based on the model automatic classification of the software system, the matching degree of the automatic classification of A, B and C types of vaginal cylinders is more than 95%, but the matching results of the other three types are notsatisfactory.
Conclusion: Five types of applicators were obtained by preliminary analysis of the existing individualized vaginal cylinders. By identifying and classifying the contour of the self-made software, the compromise choice between the traditional single cylinder applicator and the individualized vaginal 3d printed cylinder applicator could be obtained to improve dose conformity and reduce the economic burden of patients.However,the individualized cylinder data is limited, results of statistics and analysis need to be further improved.