Users are able to interact with the prototype software (a key feature, identified in Section 2.6) in the following ways:
Click on the PMP View - The user may click on the PMP. Every PMP point represents a tumour surface point. Selecting a point updates the planar views to intersect at the selected tumour surface point as well as updating the plane shown in the 3-D view and the boundary shown in the PMP.
Click on a Slice View - The user may click on any of the three planar views. This updates the plane shown in the 3-D view and the boundary shown in the PMP.
§3.5 Interaction 41
Press the Up or Down Arrow Keys - The user may press the up or down arrow keys on the keyboard. Doing so navigates up or down in the stack of images on the selected axis. Changing the selected slice also updates the position of the plane in the 3-D view and the position of the boundary in the PMP.
Press the Enter Key - The user may press the enter key to start or finish a task.
The three upper images in Figure 3.3 show 2-D MR slices on three axes (often axial, coronal and sagittal) of the relevant part of the human anatomy (the brain in this case). In the general application of the PMP technique, there are two surfaces of interest. The inner surface represents the boundary of the focal therapy volume, i.e. the volume of tissue that is estimated to have been heated beyond a specified temperature. Such a treatment volume is cumulatively tracked in order to predict a volume of tissue that could be expected to eventually die. As noted in Section 2.6, cell death does not necessarily occur during the therapy itself. The outer surface could be the surface of a tumour or a “safety” volume, beyond which it is desirable not to extend focal therapy treatment. The inner volume will be referred to as the “treatment region" and the outer surface as the “tumour boundary". For the purpose of this discussion, these two volumes are considered to be singular masses with no internal holes. The case of re-entrant tumour boundaries would typically be subdivided into simpler surfaces, so that treatments could be applied in multiple stages, perhaps with several repositionings of the therapeutic device.
The lower right diagram of Figure 3.3 shows an interactive PMP visualisation. The contour maps show a Mercator-like projection (noted in Section 2.6.1) of colour- coded inter-surface proximities between the outer tumour boundary and the inner treatment volume. The green, dotted line represents the tumour boundary in this projection. Proximities are shown from the most distant internalproximity in black (the treatment boundary furthest internalto the tumour boundary) to the most dis- tant, externalproximity in red (the treatment boundary protruding furthest outside of the tumour boundary).
When a user selects a PMP point, the three MRI slices update to intersect at the corresponding tumour surface point. This interaction (a feature noted in Section 2.6) is illustrated in Figure 3.5. These slices are displayed on the 3-D plot of the tumour (lower-left). A user is also able to navigate using the 2-D MRI views. Clicking on one of the upper three views selects it as the focus of keyboard events. Pressing up and down arrow keys steps through the MRI slices along that particular axis.
3.6
Summary
This chapter described PMP, a new interactive visualisation technique and prototype developed to address the focal therapy challenges described in Chapter 2. Design decisions made in this software implementation of PMP are summarised in Table 3.1. PMP addresses the problem of monitoring the proximity of two 3-D surfaces, which is useful for a surgeon treating a tumour with a carefully grown heat-treatment volume. Chapter 4 describes the first in a series of user studies that explore PMP, and assess its potential to address the challenges described in Chapter 2.
The studies described in Chapters 4 to 7 used between 7 and 24 participants. HCI research commonly uses smaller sample sizes and power ratings, as compared to other research fields. This is often the case when testing interface design options that result in large effect sizes. Coe [98], and Sullivan and Feinn [99], discuss effect sizes. Lazar et al. [5] and Müller et al. [100] discuss sample sizes in HCI.
Medical Images
Proximity Map Projection
Figure 3.5: PMP Schematic - Three points are shown on a PMP (right), each linked to another point on an image slice within a stack (left).
§3.6 Summary 43
Table 3.1: PMP Design Decisions
Design Issue Decision
Occlusion 3-D to 2-D mapping chosen over volume rendering because of the greater importance of surface, compared to interior, data Projection
Method
Mercator-like projection chosen for its computational simplic- ity and visual familiarity with users
Spatial Data KD-Tree data structure chosen because of its simplicity, and performing sufficiently well to achieve contemporary refresh rates
Colour Encoding of Proximities
Blue chosen for non-protruding proximities, and red for pro- truding, because red is commonly associated with heat or dan- ger
Colour Scale Logarithmic scale was chosen to emphasise smaller proximity differences nearer to the surface
Programming Language
IDL chosen because of its rapid development features and vi- sualisation orientation
Anatomical Labels
Axial, coronal, and sagittal axis labelled “X”, “Y”, and “Z” because they are familiar to non-medical study participants Tumour and
Treatment Colour
Red chosen to represent the heat treatment region due to red’s common association with heat, while blue chosen for the tu- mour region to contrast the treatment region
Detail Views 2-D axial, coronal and sagittal views chosen for detail view as they are familiar to medical users and common in contempo- rary interfaces
PMP – Detail Link
PMP points programatically linked to detail view slices to fa- cilitate efficient updating of view “stack” positions
Alternative Stack Navigation
Up and down arrow keys made available to navigate through detail slices to provide a familiar and simple alternative in case PMP usage was found to be uncomfortable or difficult for users
Chapter4
Navigating Proximity Data - PMP
in the Static Scenario
4.1
Introduction
This chapter describes the first in a series of user studies that explore the PMP tech- nique and assess its potential to address the challenges described in Chapter 2. This study tested hypotheses 1, 2, 3, and 4, described in Section 1.2, and explored as- pects of monitoring and predicting when and where two 3-D surfaces approach each other, in a "static scenario". The results of this study informed the design of sub- sequent investigations of PMP’s application to the specific needs of surgeons. In a static scenario, the surfaces, and the proximities between them, do not change. The user study described in this chapter considered whether participants could more quickly and accurately select 2-D “slices” of 3-D data, using PMP, compared with a typical current technique. The results of that study[9] are described in Section 4.3 and discussed further in Section 4.4, with conclusions being drawn in Section 4.5.