Study Reveals Inconsistency in Computer-aided Socket Manufacturing

Manufacturing sockets using computer-aided technologies is popular
practice for central fabrication facilities, but there is currently no level of
standards for facilities that utilize this technology. This often results in
inconsistent socket quality among facilities, which researchers at the
University of Washington were surprised to discover.

“We had done a couple of studies on aspects of computer
manufacturing and had sent socket files to central fabrication facilities to
make sockets for us. About one third of them were really good, and the other
others were not so good,” Joan E. Sanders, PhD, lead author from
the university’s Department of Bioengineering, told O&P Business
. “That was interesting, but we didn’t know what the numbers
meant. We didn’t know what was clinically relevant or how much error was

Socket evaluation

The researchers designed another study to further investigate this
trend. The study included 10 unilateral transtibial amputees, seven male and
three female, who were either K3 or K4 level ambulators.

The researchers then chose six central fabrication facilities used in
the previous studies and separated them into two groups of three. Each group
contained a facility that had fabricated well-matched sockets and two
facilities that fabricated moderately- or poorly-matched sockets in the
previous studies.

Study Reveals Inconsistency in Computer-aided Socket Manufacturing

© 2012


A file was made of each participant’s socket and sent to the three
facilities in one of the groups for fabrication. A total of 33 sockets were
made, three for each test subject, except participant seven who tested one
socket from each of the six facilities.

First, the researchers tested the sockets computationally using a
shape-sensing machine that they had developed. These measurements of the test
sockets were compared with the subjects’ original socket shape to
determine socket volume error, regional socket volume error and local socket
shape error.

A second evaluation was then conducted by a prosthetist. The patient was
instructed to sit for 10 minutes with their normal prosthesis on and the
prosthetic foot supported on the floor to achieve a homeostatic condition
before test fitting. The patient then removed their prosthesis and donned a
test prosthesis for the clinical evaluation, which involved bearing weight on a
fitting stool, assessment of distal end bearing and pressure points in the
residual limb and socket, and visual inspection of skin color after socket
removal to assess tissue response. This process was then repeated for the
remaining two test sockets.

Room for improvement

Based on the computational analyses, the researchers determined that of
the 33 sockets tested, 13 had a mean radial error greater than 0.25 mm and
needed to be resized. Of the remaining 20 sockets, five had an interquartile
range greater than 0.40 mm and were globally or regionally oversized, and of
the remaining 15, five more needed shape modifications at closed contour
locations because they had closed contours of surface normal angle error
greater than 4.0·.

Only 10 were clinically acceptable and did not need to be modified.

When the researchers compared their computational measurements with the
practitioner’s clinical evaluation, they found that their assessments were

“We compared the assessments by the practitioner to the
quantitative information, and it was interesting because the quantitative
metrics that we used really matched well with the practitioner’s
assessment,” Sanders said. “So when we calculated the volume error on
the quantitative side, it matched the practitioner’s assessment that the
socket was too big. This is a good sign, because it will eventually allow us to
identify those sockets early before a patient comes in for a fitting.”

The researchers also found that the well-matched sockets in this study
were fabricated by the same facilities that had demonstrated well-matched
shapes in the previous studies, suggesting that inaccuracy in computer
manufacturing does not affect the entire industry.

“The sockets that fit well were made by the same facilities that
had done a good job in the prior study,” Sanders said. “So part of
the interpretation of that is there is not a universal error in computer socket
manufacturing. Some facilities just practice the art better than others.”

The researchers hope that their results will help create a level of
standards for computer manufacturing of sockets, which could improve the entire
fitting process.

“We would love to see companies, especially those that make
computer manufacturing equipment, incorporate our algorithms into their
technology and allow facilities that utilize computer manufacturing to evaluate
their sockets after they have been fabricated to determine if they match the
computer file,” Sanders said.

However, according to Sanders, in order to achieve standardized
production, central fabrication facilities first need to have a tool to test
their quality of work.

“When we published our first study in 2007, we got calls from
central fabrication facilities asking if we could test their sockets because
they had no way to know what level of error they were producing,” Sanders
said. “So a commercial product needs to be made so they can test the
quality of their shape manufacturing in their own facility, and then the
standards should follow.” — by Megan Gilbride


Sanders JE, Severance MR, Allyn KJ. Computer-socket manufacturing error:
How much before it is clinically apparent? J Rehabil Res Dev.

Disclosure: Sanders has no relevant financial disclosures.

Leave a Reply

Your email address will not be published.