Human beings are experts at recognizing the identity of their fellow humans due to their ability to discriminate based on our varying shapes, colors, and sizes. Although most of the variation we use to recognize each other is found in external features of the body, such as height, weight, eye color, hair length, and skin tone, a large portion of human variation is contained in the skeleton. These variations originate from the person's age, geographic origin, sex, or simply the idiosyncrasies of an individual's genetics and environment.

Using current technology, a large collection of bone specimens can be easily analyzed and represented using machine learning and data mining algorithms. By digitizing bone samples, an opportunity exists to use computerized statistical techniques to enhance and simplify the osteological work of researchers. We envision a system in which the computer can help identify plausible anatomical landmarks on a recently digitized bone fragment to aid in its identification. Using shape analysis techniques, shape parameters can be fit to the bone fragment and then extrapolated to the entire skeleton. The skeletal model can easily be ported to a handheld device which can be used as a reference in the field.

In a medical context, the information contained within the model may be used to segment medical images using model-based techniques. Fitting the model to the scan of a patient would enhance surgery planning and simulation. Also, the skeletal model would be incredibly useful in an educational setting, where students could examine the effects that parameters, such as age, sex, and geographic location, have on the shape of the skeleton.

The goal of the Parametric Skeleton Project is to make this vision a reality. We intend to build a parametric skeletal model providing the greatest anatomical detail possible given current technology. Our model will contain data for both sexes and as many different sample ages and geographic locations as possible. The surface elements of our model will be tagged with anatomical landmark data, tying the domain knowledge to the geometry. Using this data, our system will be designed to allow the construction of further human models based on ours. In particular, the Parametric Skeleton Project is a component of the Parametric Human Project for digital, anatomically accurate ergonomics.

We have identified three major milestones to the completion of our goals. First is the data acquisition step, in which we will collect a wide variety of digital models of human bones and label them with anatomical landmark data. In the following model construction step, we will define and compute a number of geometric, statistical markers and tie them to the anatomical landmark information. Using this information, we will align the various bone samples and capture their shape variation using statistical techniques, thereby building a parametric model. Finally, we demonstrate the many uses of our model by building the several applications.