Techniques: Volumetric Visualization

You could try to assemble a group of people more diverse in individual priorities, goals, and thought processes, than a team comprised of software developers, engineers, radiologists, medical clerks, surgeons, managers, technicians, university staff, and so on, destined to implement some form of surgical planning or navigation.

Generally, doctors would like to see patients become healthy again, the sooner the better. Managers prefer to see consistent progress made on the project starting now. Radiologists / radiographers stick to their usual protocols and 2-D slices. Mechanical engineers can’t wait for some drawings or drafts to start to work on. The other staff are already overwhelmed with numerous tasks, meanwhile the software engineers won’t know what to do with themselves without their precious product backlog items already there, ready to tackle.

Good leadership with strict time management could - and should - steer the ship in the right direction, but it’s easier said than done, as they say. Instead of reading several hundred articles at once, a large number of book chapters and tutorials on several complex topics, e.g. each anatomical region and surgical procedure, there’s another way the software-related members of the team could get the ball rolling.

That is to use the only common language among these vastly different individuals: the visualization of the problem, or at least the problematic area first.

One could start with the classic (or maybe other variants of the traditional) marching cubes algorithm, by trying to show off what those different thresholds actually display regarding the patient’s anatomy, and that is perfectly fine. However, even before defining any premature CT or MRI scan protocols, or even worrying about the positioning of the patient, etc. it probably would be best to ask for any - with reasonably low slice thickness and relatively high resolution - relevant CT or MRI scan of a patient, and explore the options of visualizing the whole scan.

As with almost any other software project, the way forward is creating small, simple proof of concepts (PoCs), that will spark the interest of all involved, and encourage those much needed discussions. The goal here is to make the eyes light up, and establish common ground.

So a step one could very well be, for almost every situation or case, a volumetric visualization of the entire scan volume.

Whether the surgeon is searching for potential osteophytes, cysts, malunion fractures, maybe examining nearby soft tissue or just looking for any deformities or irregularities, volume rendering is extremely useful for exploring the space. Dynamic window/level functions and cut planes help adjusting the focus on the area of interest, hiding everything that is obstructing the view necessary for investigation. Only then a productive discourse will commence, that could end in high-level technical details, or smaller user stories, or it could take any other useful direction.

Since segmentation resulting in anatomically correct - especially 3-D printable - mesh models is a very expensive step, polygonal meshes really need to be created only for selected "objects" in this space that will be part of the ongoing planning process. The simple example below reveals a display of a malunion on the distal end of the radius, using an implementation in VTK, the Visualization Toolkit.