SEGMENTATION OF ELECTRONIC PLANOGRAMS BY MEANS OF ARTIFICIAL INTELLIGENCE
Background. Computer plantography is one of the methods of diagnosing the condition of the feet. During the calculation of clinical and functional parameters during the study, one of the stages is to identify points lying on the border of the foot's contact zone with the support surface. In the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Penza State University Publishing House
2025-05-01
|
| Series: | Модели, системы, сети в экономике, технике, природе и обществе |
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Background. Computer plantography is one of the methods of diagnosing the condition of the feet. During the calculation of clinical and functional parameters during the study, one of the stages is to identify points lying on the border of the foot's contact zone with the support surface. In the form of active development of artificial intelligence technologies, an urgent task is to develop a model for segmentation of contact zones as one of the stages of automation of this research. The purpose of the work is to develop and evaluate a segmentation model of the foot support zone in computer plantography images using artificial intelligence. Materials and methods. The study used a dataset containing 500 images of computer plantography of different patients. Results. Based on the results of training the yolo11x-seg model (image segmentation model), high performance was achieved in detecting and segmenting contact zones in the anterior and middle parts of the foot and separately in the posterior part of the foot. The quality metrics of the model were: mAP50 0,9727, mAP50-95 0,8293, accuracy 0,9849, completeness 0,9684 in the segmented area detection task, and mAP50 0,9727, mAP50-95 0,8482, accuracy 0,9849, completeness 0,9688 in the semantic segmentation task. The obtained indicators reflect the ability of the model to effectively identify and segment the common contact area in the forefoot and middle part of the foot, as well as the contact area in the posterior part of the foot. Conclusions. The integration of this model into medical decision support systems will speed up the process of image analysis and reduce the labor costs of specialists, which will optimize research and improve the quality of medical services.
|
|---|---|
| ISSN: | 2227-8486 |