Models, systems, networks in economics, engineering, nature and society
Background. At the current stage of high-tech medicine systems development, colonoscopy is the most effective way of early diagnosis of colorectal cancer. The most typical anomaly leading to a high risk of cancer is the appearance and development of colorectal polyps – neoplasms several millimeters...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Penza State University Publishing House
2025-02-01
|
| Series: | Модели, системы, сети в экономике, технике, природе и обществе |
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850037124026335232 |
|---|---|
| author | V.V. Khryashchev A.L. Priorov N.V. Kotov K.I. Malygin |
| author_facet | V.V. Khryashchev A.L. Priorov N.V. Kotov K.I. Malygin |
| author_sort | V.V. Khryashchev |
| collection | DOAJ |
| description | Background. At the current stage of high-tech medicine systems development, colonoscopy is the most effective way of early diagnosis of colorectal cancer. The most typical anomaly leading to a high risk of cancer is the appearance and development of colorectal polyps – neoplasms several millimeters in size. Their detection during a colonoscopic examination is an extremely important task for an endoscopist. The aim of the work is to further improve the algorithms for automatic detection of polyps using digital image processing methods for their use in real-time medical decision support systems. Materials and methods. A study was conducted to evaluate the efficiency of using a number of classical and neural network image segmentation algorithms. For training and testing deep machine learning algorithms, a standard open database of polyp images – Kvasir-SEG, as well as our own database of images obtained and labeled in the endoscopic department of the Yaroslavl Regional Clinical Oncology Hospital, were used. Results. A significant superiority of neural network segmentation algorithms over classical approaches and algorithms was demonstrated. According to the Dice coefficient metric, the best result was shown by the Meta-Polyp neural network architecture, reaching a coefficient value of 0.933 when using a combined database of test polyp images. Conclusions. The considered algorithms will serve as the basis for developing a hardware and software complex with an algorithmic module for detecting pathologies in colonoscopic images. |
| format | Article |
| id | doaj-art-5d5b3f8b5ac342ce8cc204b316184e13 |
| institution | DOAJ |
| issn | 2227-8486 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Penza State University Publishing House |
| record_format | Article |
| series | Модели, системы, сети в экономике, технике, природе и обществе |
| spelling | doaj-art-5d5b3f8b5ac342ce8cc204b316184e132025-08-20T02:56:58ZengPenza State University Publishing HouseМодели, системы, сети в экономике, технике, природе и обществе2227-84862025-02-014869710.21685/2227-8486-2024-4-7Models, systems, networks in economics, engineering, nature and societyV.V. Khryashchev0A.L. Priorov1N.V. Kotov2K.I. Malygin3P.G. Demidov Yaroslavl State UniversityP.G. Demidov Yaroslavl State UniversityP.G. Demidov Yaroslavl State UniversityP.G. Demidov Yaroslavl State UniversityBackground. At the current stage of high-tech medicine systems development, colonoscopy is the most effective way of early diagnosis of colorectal cancer. The most typical anomaly leading to a high risk of cancer is the appearance and development of colorectal polyps – neoplasms several millimeters in size. Their detection during a colonoscopic examination is an extremely important task for an endoscopist. The aim of the work is to further improve the algorithms for automatic detection of polyps using digital image processing methods for their use in real-time medical decision support systems. Materials and methods. A study was conducted to evaluate the efficiency of using a number of classical and neural network image segmentation algorithms. For training and testing deep machine learning algorithms, a standard open database of polyp images – Kvasir-SEG, as well as our own database of images obtained and labeled in the endoscopic department of the Yaroslavl Regional Clinical Oncology Hospital, were used. Results. A significant superiority of neural network segmentation algorithms over classical approaches and algorithms was demonstrated. According to the Dice coefficient metric, the best result was shown by the Meta-Polyp neural network architecture, reaching a coefficient value of 0.933 when using a combined database of test polyp images. Conclusions. The considered algorithms will serve as the basis for developing a hardware and software complex with an algorithmic module for detecting pathologies in colonoscopic images. digital image processingobject detectionobject segmentationneural networksmachine learningcolonoscopycolorectal polyps |
| spellingShingle | V.V. Khryashchev A.L. Priorov N.V. Kotov K.I. Malygin Models, systems, networks in economics, engineering, nature and society Модели, системы, сети в экономике, технике, природе и обществе digital image processing object detection object segmentation neural networks machine learning colonoscopy colorectal polyps |
| title | Models, systems, networks in economics, engineering, nature and society |
| title_full | Models, systems, networks in economics, engineering, nature and society |
| title_fullStr | Models, systems, networks in economics, engineering, nature and society |
| title_full_unstemmed | Models, systems, networks in economics, engineering, nature and society |
| title_short | Models, systems, networks in economics, engineering, nature and society |
| title_sort | models systems networks in economics engineering nature and society |
| topic | digital image processing object detection object segmentation neural networks machine learning colonoscopy colorectal polyps |
| work_keys_str_mv | AT vvkhryashchev modelssystemsnetworksineconomicsengineeringnatureandsociety AT alpriorov modelssystemsnetworksineconomicsengineeringnatureandsociety AT nvkotov modelssystemsnetworksineconomicsengineeringnatureandsociety AT kimalygin modelssystemsnetworksineconomicsengineeringnatureandsociety |