Foot Abnormality Mapping using Statistical Shape Modelling

Abstract

About 20% of the population suffer from disabling foot or ankle pain that require the use of foot orthotics. Traditionally, those foot orthotics are designed manually, but digital procedures are desired to provide a faster, more objective, and more reliable workflow. In this study, we introduce a method for detecting shape abnormalities in feet for the purposes of pathology diagnosis and orthotic design. The proposed method consists of two phases. In the training phase, a statistical 3D foot model (based on 42 healthy subjects) is built. In the test phase, the landmarks of a new 3D foot scan are compared to the trained model. A landmark is detected as an outlier if it is in the extreme ranges. This testing process is repeated at all landmarks to identify all abnormal foot regions. Preliminary results show that, when testing a foot of a known pathology (hallux valgus, heel spur, foot pronation), we are able to detect abnormal regions accurately. We also examined the effect of using rigid or similarity-based alignment during 3D model building and abnormality detection. We show that our proposed method is a faster and a more objective approach than traditional approaches for abnormality detection of the foot. As such, this method may prove useful in the medical diagnosis of foot pathologies and in automated orthotic design.

Three‐dimensional quantitative analysis of healthy foot shape: a proof of concept study

“…The above procedure is followed for each individual. To avoid reference bias, the whole approach is iterated three times, where in each iteration, the population average calculated from the previous iteration is used as the reference foot [31].…”

Subject-specific identification of three dimensional foot shape deviations using statistical shape analysis

“…The first, and the one used here, is to build a statistical model of what is considered normal. This model can then be compared to using established statistical tests in order to find outliers [52,53,56]. By taking this approach, our system has a strong theoretical foundation for justifying why an exemplar is an outlier [66].…”

“…In each iteration, the population mean calculated from the previous iteration is used as the reference foot mesh [56]. Convergence is reached if the average distance between corresponding points on the reference mesh from the previous iteration and the reference mesh from the current iteration is less than ε = 0.001 mm.…”

PAPPI: Personalized analysis of plantar pressure images using statistical modelling and parametric mapping

“…Like SPM, PAPPI is based on the idea of bringing plantar pressure measurements into anatomical alignment, then performing statistics at each pixel. Unlike SPM, PAPPI employs a statistical outlier detection algorithm to classify plantar pressure abnormalities pixel-by-pixel [31,32]. This outlier detection involves the pixel-bypixel modelling of plantar pressures from a healthy population as well as the relationship between those pressures and demographic factors such as age, weight, and gender.…”

“…The personalized evaluation of a patient's peak pressure image follows the workflow shown in Fig 3 and is similar to outlier detection techniques seen elsewhere [31,32]. The patient's demographic factors (age, sex, weight, height, shoe size) are inputted into our linear regression model to predict their peak pressure image.…”