Principal Warps: Thin-Plate Splines and the Decomposition of Deformations
IEEE Transactions on Pattern Analysis and Machine Intelligence
Active shape models—their training and application
Computer Vision and Image Understanding
An Adaptive-Focus Deformable Model Using Statistical and Geometric Information
IEEE Transactions on Pattern Analysis and Machine Intelligence
Comparison of Prostate Biopsy Protocols using 3-D Computer Simulation
PACMEDTEK '98 Proceedings of the Symposium on Pacific Medical Technology
| Hi-index | 0.00 |
Predefined biopsy plans are important to assist urologists in prostate cancer diagnosis and selection of biopsy sites. It is challenging to map a predefined biopsy plan onto current transrectal ultrasound (TRUS) prostate volume due to the wide range of shape variation of prostates, and the spatial difference between the prostate model where plan is defined and actual prostate imaged in the ultrasound scan. In this paper, we present a novel method to map a predefined biopsy plan onto the current prostate volume. A statistical prostate mean shape trained from 38 prostate volumes is used as the model, and plans are defined on the mean shape. For each ultrasound scan, the prostate is segmented out first, using a fully automated segmentation method, then the shape model is registered to the segmented prostate using a robust surface-based registration technique, and correspondence is determined. Finally, predefined plans on the mean shape are mapped onto the current prostate volume using interpolation from surface correspondence. We tested our method using biopsy plans with various number of cores, including 6 (sextant), 10, 12, and 16 cores. The mapped plans fitted well into current volume by observation. The system is developed on Microsoft Windows XP using VC++.