Shit in, shit out. [Pardon my French]
The title of this blog refers to a question I once asked a video editor. I was disappointed about the quality of my movies and wanted to have the same crisp high resolution video as my friend, who owned the latest HD camera. Obviously there is no way to make the same beautiful movies with a cheap old camera. There is no magical black box that can convert gritty low resolution video into high definition video. If you make “shit” quality in video, you can only produce “shit” quality movies. The same holds true for 3D-DXA. It’s not a magic black box that can convert a gritty low resolution DXA image into a beautiful detailed 3D model.
You do not need a lot of technical knowledge to understand this fact. You can already come a long way with just a bit of common sense. In promotional material, DXA device manufacturers often show a nice smooth high resolution image of the femur. This is quite misleading. Dual-energy X-ray Absorptiometry, as the name already says, uses two images of a different X-ray intensity, and combines them into a single image of the bone density. Combining these two images also combines the noise, and the actual DXA images is in fact much noisier than a single X-ray image. Also the number of pixels is much less than what the smooth illustrations do suggest.
If you have access to a DXA workstation, open a DXA image of the hip and have a look at the femoral neck. Tell me, can see the cortex of the femoral neck? Can you measure its thickness? What about the density of only the cortical bone? What about the thickness and density of the cortical bone in the middle of the image? Of course not! Do you really think there is a magical computer algorithm that can extract these measurements from this single blurry picture? In computed tomography we use hundreds of projections and even then it is extremely difficult to measure the cortical parameters. It is impossible from devices by GE or Hologic, but god forbid if you try it with an image from a Stratos device by DMS-Apelem. It is actually very hard to find an example DXA image of the Stratos device on the internet (hiding something DMS-Apelem?), but I found one from a scientific publication (it’s the one on the right):
I apologize to the authors for using their illustration without their permission, but this is all for the good of mankind!