TL;DR
This paper introduces a novel, automated method for measuring the foot arch index using a flexible membrane pressure sensor. By combining image processing techniques with pressure data, the system accurately distinguishes between high arch, normal, and flat feet, offering a faster and more convenient alternative to traditional manual methods.
What Problem Does This Paper Address?
The foot arch index is a crucial indicator for assessing foot health and biomechanics. However, existing manual measurement methods are often inefficient, time-consuming, and dependent on the skill of the operator, leading to poor reliability and repeatability. While some automated methods exist, they can be computationally expensive or still require manual intervention. This paper aims to bridge this gap by developing an automatic, accurate, fast, and cost-effective foot arch index measurement system using flexible membrane pressure sensors.
Methods
The proposed method leverages a flexible membrane pressure sensor to capture plantar pressure distribution. This raw data is then processed through several key stages:
- Data Acquisition and Visualization: Pressure data from the sensor is converted into a 44x52 matrix and then mapped to a grayscale image (0-255) where brighter pixels indicate higher pressure.
- Noise Reduction (8-Neighborhood Correlation Pixel Method): A novel method is introduced to remove isolated noise points. It identifies pixels whose 8 neighbors are all zero (indicating no pressure) and sets them to zero, preserving important foot arch data. This is presented as an improvement over traditional median, Gaussian, or mean filters, which can blur the image or remove valuable arch information.
- Toe Data Removal (Row Element Association Algorithm): To accurately calculate the arch index based on the foot sole area, the data corresponding to the toes must be removed. This algorithm identifies and zeros out toe data by analyzing the number of consecutive effective pixel points in specific rows, distinguishing them from the more continuous foot sole pressure.
- Binarization: The effective foot sole pressure data is binarized, converting pressure values greater than zero to 255 and zero values to 0. This simplifies subsequent calculations without losing essential information about the pressure area.
- Foot Endpoint Location: The algorithm automatically identifies the start and end points of the left and right foot soles by traversing the binarized image. This is crucial for determining foot length.
- Arch Index Calculation: The foot arch index is calculated as the ratio of the effective arch pressure area to the total effective foot sole pressure area (excluding toes). This is a common definition of the arch index.
A prototype system was developed and tested to validate the method.
Datasets
The paper does not explicitly mention or introduce new benchmark datasets. Instead, it relies on data collected from subjects during experimental validation. The experimental setup involved three subjects with different foot arch types (high arch, normal, flat foot), each undergoing 10 manual and 10 automatic measurements.
Key Insights and Ideas
- Novel Noise Reduction: The 8-neighborhood correlation pixel method appears to be a significant contribution, offering a way to remove noise without blurring the image or losing critical foot arch data, which is a common challenge in pressure-based measurements.
- Efficient Toe Removal: The row element association algorithm provides an automated and effective way to segment the foot sole from the toes, a necessary step for accurate arch index calculation.
- Integrated System: The paper demonstrates the successful integration of pressure sensing, image processing, and algorithmic analysis into a cohesive system for automatic foot arch index measurement.
- Clinical Relevance: The method's ability to accurately distinguish between different foot arch types holds significant potential for clinical applications, aiding in the diagnosis and management of foot-related conditions.
Key Results
The study's experimental results, comparing manual and automatic measurements, indicate good agreement.
- Accuracy and Repeatability: The average absolute deviation (AVEDEV) and standard deviation (STDEV) for the automatic measurement were slightly lower than for the manual method (AVEDEV: 0.0069 vs. 0.0071; STDEV: 0.0090 vs. 0.0093).
- Coefficient of Variation (CV): The average CV for the automatic method (0.0408) was also lower than for the manual method (0.0440), suggesting better precision and repeatability.
- Distinguishing Foot Types: The results, visualized in Figure 13, show that the automatic method effectively distinguishes between high arch, normal, and flat feet, aligning well with manual measurements.
Limitations and Future Work
- Subject Stance: The paper acknowledges that subjects must stand naturally and remain still during measurement for accurate results. Any tilting or movement can lead to significant deviations.
- Generalizability: While the prototype was validated, further testing with a larger and more diverse population would be beneficial to confirm its generalizability across different demographics and foot conditions.
- Extension to Medical Diagnosis: The authors suggest that the designed prototype is "worth being extended to actual foot arch index medical diagnosis," implying that further development and clinical validation are needed for widespread medical adoption.
- Dynamic Measurement: The current method appears to focus on static measurements. Future work could explore extending this to dynamic analysis during walking or other activities.
Summary updated