Interactive medical image segmentation software for preoperative planning of complex cancer surgeries

点击翻译按钮获取中文摘要

The visualisation of anatomy in 3D has been shown to improve the ability to localizestructures when compared with traditional 2D imaging slices. This has allowed imaging tomove from a largely diagnostic tool to one that can be used for both diagnosis and operativeplanning.To allow for this segmentation of the images need to be performed, where each voxel islabeled by its nature i.e. extracting anatomical structures such as bones, lungs and otherorgans. This allows for better 3D visualisation and consequently facilitates understanding ofthe anatomy. Currently segmentation of medical images is not widely used in clinical practice,because of the time required to learn and use existing software. It can take several hours ofwork, even for a trained biomedical engineer, to perform a segmentation of anatomicalstructures using current available software. Also setting up the software for a few cases can beexpensive, and clinicians often do not have the computational resources to run thesealgorithms.The objective during the proof of concept study is to:a) Develop a novel algorithm to allow for an interactive segmentation of medical images.b) Make the interaction with the software web based, using the latest web technologies.c) Test the proof of concept by performing segmentations for a clinical study for patientsundergoing kidney cancer surgery.