11.1. Introduction
In this section, we introduce the idea of using simulations to understand the likely performance of a given Computer Assisted Surgery (CAS) system.
As we have seen on this course, a CAS system has several components. The end result is something that is used in a complex environment, by a range of people, and depends on many of the topics we have learnt:
What type of images are available? See the Imaging Overview section.
Is segmentation involved? See the Segmentation and Modelling section.
What devices are tracked? See the Tracking section.
How are they calibrated? See the Calibration section.
What is displayed on screen? See the Graphics section.
How does the user interact? See the Human Computer Interaction (HCI) section.
So, a complete system can take a long time to develop.
Typically, the system design starts with an idea, and progresses through the following stages:
Algorithm design, software only testing, or lab testing, lab evaluation. [Kang2014], [Thompson2015], [Hu2016].
Animal testing. [Kang2014], [Thompson2015]
Human evaluation. [Edwards2000], [Prevost2019], [Hu2016].
and the timeframe for each stage is often measured in years.
As we have covered, at lot of current research is based around the geometry of a system, and involves research into registration methods and calibration methods.
In this chapter we explore the idea that it would be wise to initially model the geometry of a system so we understand the relative importance of each component part. For example, in a liver surgery system ([Thompson2015]), we have
video intrinsic calibration
video extrinsic (stereo) calibration
tracking
hand-eye calibration
registration
In this scenario, which component has the most effect on the performance of the system? If we had a good understanding of the geometric parameters, we can determine clear specifications for when a component is ‘good enough’, and spend time wisely.
From an ethical point of view, performing simulations and lab work would also save costly animal experiments, and reduce the risk of deploying systems that are un-trustworthy into patient studies.