Augmented reality navigation for liver surgery: an enhanced coherent point drift algorithm based hybrid optimization scheme

Ramesh Dhoju, Abeer Alsadoon, P. W.C. Prasad, Nedhal A. Al-Saiyd, Ahmad Alrubaie

Research output: Contribution to journalArticlepeer-review

Abstract

Augmented reality (AR) based bowel or liver surgery still has not been implemented successfully due to limitations of accurate and proper image registration of uterus and gallbladder during surgery. This research aims to improve target registration error, which helps to navigate through hidden uterus and gallbladder during surgery. Therefore, it will reduce risk of cutting uterus or common bile duct during surgery, which can be fatal and cause devastating effects on the patient. The proposed system integrates the enhanced Coherent Point Drift (CPD) Algorithm with hybrid optimization scheme that incorporates Nelder-Mead simplex and genetic algorithm, to optimize the obtained weight parameter, which in turns improves the target image registration error and processing time of image registration. The system has minimized the target registration error by 0.31 mm in average. It provides a substantial accuracy in terms of target registration error, where the root mean square error is enhanced from 1.28 ± 0.68 mm to 0.97 ± 0.41 mm and improves processing time from 16 ~ 18 ms/frame to 11 ~ 12 ms/frame. The proposed system is focused on improving the accuracy of deformable image registration accuracy of soft tissues and hidden organs, which then helps in proper navigation and localization of the uterus hidden behind bowel and gallbladder hidden behind liver.

Original languageEnglish
Pages (from-to)28179–28200
Number of pages22
JournalMultimedia Tools and Applications
Volume80
DOIs
Publication statusE-pub ahead of print - 31 May 2021

Fingerprint

Dive into the research topics of 'Augmented reality navigation for liver surgery: an enhanced coherent point drift algorithm based hybrid optimization scheme'. Together they form a unique fingerprint.

Cite this