Development of natural 3D interfaces

		A template comparing different contact area estimations. The contact area is estimated based on the stress calculated on the cartilage surfaces. A stress threshold was defined in A2 and used to determine the contact area on the animation in B2. Other views were derived (C2 and D2) and a graph of the contact area history is shown in E2. The row 2 was then copy-pasted into rows 3 to 5. Finally, different thresholds are chosen and the contact area changes propagate into the remaining cells.

Medical visualisation systems are usually based on the general flow chart paradigm. Flow chart visualisation systems are typically used for data visualisation and widely available as commercial products. In those systems, users select processing modules and wire them together into a pipeline. A common problem with these systems is that they spend the most part of their screen on operators and their interconnections. To deal with this problem another class of visualisation systems was introduced: spreadsheet-like interfaces. Spreadsheet-like interface are a generalisation of conventional spreadsheets where cells can contain graphical objects such as images, volumes, or animations or even widgets to interact with data. In this class of visualisation systems, screen space is spent on operands rather than operators, which are usually more interesting to the end user. They also benefit from the fundamental properties of spreadsheets where it is easy to organise, compare, analyse and perform operation on data makes up natural interfaces.

Our approach to medical visualisation is based on a spreadsheet framework, which consists basically of the following elements: cells, operators and dependencies. Cells are the basic data elements. They can contain numbers, images (2D or 3D), curves, vectors or matrices. They can also contain widgets for interactive cells (for example, sliders to control opacities for volumetric data). The cells are organised in a tabular layout, which makes them easy to browse. Operators are applied to cells or ranges of cells and results stored in cells. These operators define the dependencies between the cells. A firing algorithm controls dependencies as in conventional spreadsheets. This algorithm keeps track of dependencies between cells and automatically updates the cells to reflect changes.

In the context of this project, the explored data come from the biomechanical simulation of the soft tissues in the hip joint. Separated software, based on a conceptual joint model is used for hip simulation. The model is based on a hybrid approach. A kinematical component defines the bony rigid motion from measures on the static and dynamic MRI, while a biomechanical component computes soft connective tissues deformation, and allows estimating force exchange and consequent stress on those soft structures. The Figure shows an example of use case: we performed is 90° of flexion plus total internal rotation, a key motion in Orthopaedics.

Besides all the advantages known about conventional spreadsheet interfaces (creating analysis templates, applying operations in parallel), we can mention some that have shown to be very relevant in the context of medical applications. Clinicians do not care about the data operations, they are more interested on the data itself, and that is the heart of spreadsheet interfaces. Medical data being multidimensional are easily understood due to the tabular organisation of the interface, which reduces data dimensionality. It helps in creating a coherent mental model. Comparing different visual representations and modifying the parameters that make them different at the same time allows clinicians to isolate and focus on the interesting features and discard less useful views, as in the case study. That allows clinicians to gain time on diagnosis.

Conclusions can also be drawn about our hip joint model development. The possibility of easy comparison and parameterisation offered by the spreadsheet highlight the weak points of the model allowing us to correct them. It also aids in identifying the biomechanical features playing a key role on the medical problematic, guiding the researcher on the task of simplifying the model while keeping it medically useful.