Design, Performance, and Analysis of an Automated Tool for Neuropathy Assessment on the Plantar Surface

Abstract

Diabetic peripheral neuropathy results in the loss of sensation in the hands and feet. Other symptoms such as numbness, burning, shooting pain, and electrical sensation have been reported. Individuals with this disease are at a greater risk of requiring amputations, as a result of unknowingly puncturing the soles of their feet leading to infection and ulceration. Tuning forks, electrodiagnostic equipment, and other novel inventions have been used to detect for neuropathy on the plantar surface, but the Semmes-Weinstein monofilament remains the most common tool. The monofilament is pressed against the plantar surface until it buckles, at which point it theoretically produces a maximum contact force, often expressed using grams of force. The most popular monofilament examines for the loss of protective sensation, designated at the threshold sensitivity of 10.0 grams of force. However, this tool’s accuracy is subject to factors such as insertion rate, angle of insertion, diameter, length, human skin material properties, temperature, humidity, and even material fatigue. As such, the purpose of this dissertation is to demonstrate the practical concerns of the hand-applied Semmes-Weinstein monofilament assessment technique, present a novel diagnostic tool which automates the assessment protocol, and implement it in a clinical study to ascertain subjects’ current degree of threshold sensitivity on the plantar surface.

Relevant background information is provided in Chapter 1, while Chapter 2 of this dissertation details a theoretical contact mechanics analysis of a hand-applied Semmes-Weinstein monofilament in contact with a human skin sample. Theoretical equations and finite element analysis are both used to explore the influence that monofilament diameter, insertion depth, and material skin properties have on the contact force, and corresponding normal stress at the center of contact. Both a homogeneous isotropic and composite isotropic model are considered, the later includes the epidermis, dermis, and subcutaneous fat layers of human skin. The study resulted in 188 finite element analysis simulations, which after linear regression analysis led to the derivation of empirical equations. The equations relate contact force and normal stress at the center of contact to insertion depth, the epidermis stiffness, and dermis stiffness. The conclusions of this study were that small amounts of insertion depth can have a substantial impact on the contact force produced, and that the material properties of the epidermis and dermis layers are also impactful. These findings suggest that attention to application technique is recommended when interpreting the results while using the hand-applied monofilament for neuropathy assessment on the plantar surface. Furthermore the necessity for a force feedback loop used to measure the contact force during application is established.

After understanding the practical concerns of the hand-applied Semmes-Weinstein monofilament assessment, examined in a non-buckling simulation study, and the dependencies on external factors on the tool’s accuracy, the development of an automated tool is presented in Chapter 3. This automated tool took the current commercially available 10.0-gram force Semmes-Weinstein monofilament and placed it in a robotic CNC device. The monofilament was moved into position with the use of belts and pulleys and was attached to an innovative probe subassembly. The probe subassembly used a stepper motor load cell feedback loop to constantly measure the contact force during insertion, until a prescribed value was achieved. The device used a new methodology which evaluated 13 locations per foot, grouped within three regions: toes, ball, and heel. The device was used at each location until the threshold sensitivity was determined, based on the following force classifications: 0.35, 0.70, 2.0, 4.0, 6.0, 8.0, 10.0, and >10.0 grams of force. The methodology featured randomization, false positive checks, and documentation.

The automated tool was used in a clinical study that is presented in Chapters 4 and 5. Chapter 4 encompasses the device’s force producing accuracy, a comparison to a hand-applied Semmes-Weinstein monofilament assessment, and linear regression analysis between subject’s age, body mass index (BMI), ankle brachial index (ABI), fasting blood sugar (FBS), HbA1c and threshold sensitivity. The subjects examined in Chapter 4 were healthy control subjects, without type 2 diabetes mellitus. A Threshold Sensitivity Index (TSI) was calculated for each region, in addition to a TSI Norm. TSI Norm was representative of a subject’s entire threshold sensitivity across 13 locations per foot. The automated tool’s force producing accuracy was determined to be associated with the region the locations were within. The maximum average absolute errors were 0.5, 1.2, and 0.9 grams of force at the toe, ball, and heel locations, respectively. The toes demonstrated an average absolute error less than or equal to 0.4 grams of force at 98% of the locations evaluated, while the ball locations and heel locations were 84% and 60%, respectively. The 10.0 grams of force hand-applied monofilament was underdiagnosing 21% of the locations when compared to the automated device. Linear regression analysis found that the healthy control subjects attributed their threshold sensitivity (TSI Norm) to their age via significant linear regression (R2=0.3422, P=0.004), while BMI, ABI, fasting blood sugar, and HbA1c were uncorrelated. Chapter 5 repeated the linear regression analysis between TSI Norm, age, BMI, ABI, fasting blood sugar, and HbA1c, for subjects with type 2 diabetes mellitus, with and without neuropathy symptoms. The subjects from Chapter 4 and 5 had ages, BMIs, and ABIs that where not significantly different when evaluated using ANOVAs. These subjects had significantly different fasting blood sugars and HbA1c levels. Nonetheless, the study did not find a significant difference between each group’s TSI Norm. The groups with type 2 diabetes did not correlate their threshold sensitivity to their age, BMI, ABI, fasting blood sugar, or HbA1c. This finding encourages the use of the automated tool for follow up screenings to monitor neuropathy disease progression on the plantar surface.

Chapter 6 details the development of the second iteration of the automated tool. Although the second prototype retained a similar gantry system and probe subassembly, improvements enhanced its structural rigidity and usability. Chapter 7 presents future studies that could be conducted in concert with the automated tool. Topics such as the effects of sex and time of year are encouraged, as well as determining locations that yield a better potential for earlier diagnosis. Using the automated tool for treatment monitoring would also be beneficial. Improvements to the control systems and the code that runs the automated tool are encouraged.

This dissertation presents the development and functionality of an automated tool for neuropathy assessment on the plantar surface. It was used in a clinical study, which concluded that threshold sensitivity is challenging to predict using age, BMI, ABI, fasting blood sugar, or HbA1c alone. With the automated tool threshold sensitivity can be documented and studied over time, which could provide both clinicians and their patients insights into the efficacy of treatments and disease progression.