Miraculous: A Python Tool for Plantar Pressure Analysis

Abstract

Plantar pressure analysis tools are commonly used to characterize a variety of gait related pathologies and to track healing progress. Commercial plantar pressure software tools are often expensive and non-modifiable. This project presents Miraculous, an open-source Python tool which allows researchers and clinicians to view, modify, and add to stock data processing pipelines and metric calculations for plantar pressure analysis. The software utilizes a YOLOv8l object detection model that was trained and validated on over 20,000 individual footsteps for both localization and classification. The model achieved an accuracy of over 99% for classifying left and right steps. Incomplete steps were classified with an accuracy of 60.9%. Once steps are located and classified with the model in the software, individual steps and their metrics are processed and averaged so trial level statistics can be computed. Finally, a clinical report is automatically generated containing useful metrics and visualizations for plantar pressure analysis. The entire workflow from data import to clinical report generation is housed in a web-based application built entirely with Python, and users are encouraged to edit and add to the framework to suit their specific needs. Miraculous overcomes the limitations of commercial plantar pressure software packages by providing a platform the is both free and adaptable.

Video