|
ИСТИНА |
Войти в систему Регистрация |
ИСТИНА ИНХС РАН |
||
Gender identification of Drosophila melanogaster based on morphological analysis of microscopic imagesстатья
- Авторы: Mestetskiy L.M., Guru D.S., Benifa J.V.Bibal, Nagendraswamy H.S., Chola Channabasava
- Журнал: Visual Computer
- Том: 38
- Номер: 4
- Год издания: 2022
- Издательство: Springer Verlag
- Местоположение издательства: Germany
- DOI: 10.1007/s00371-022-02447-9
- Аннотация: Drosophila melanogaster (D. melanogaster) is an imperative genomic model organism that is employed widely in healthcare and biological research works. Roughly 61% of recognized human genes have a perceptible similarity with the genetic code of D. melanogaster flies, besides 50% of its protein structures have mammalian equivalents. In recent times, numerous studies have been done in D. melanogaster to investigate the functions of particular genes that are available in its central nervous system, including the major organs like the heart, liver and kidney. The findings of these research works through D. melanogaster are utilized as a key mechanism to explore human interrelated diseases. However, it is essential to recognize the male and female Drosophila flies for the better understanding of human disease related studies, and it is a tricky job. This paper describes a unique programmed system to categorize the gender of D. melanogaster from the ventral view portraits captured through microscope. The proposed method includes image segmentation of the body of D. melanogaster in the form of a binary image and the construction of a continuous morphological model based on its skeleton. An analysis of the skeleton makes it possible to assess the sharpness of the caudal end of the D. melanogaster abdomen through a detailed assessment of the curvature. Based on this assessment, a Drosophila melanogaster Gender (DMG) classifier is constructed for the gender determination of D. melanogaster flies. The accuracy of the DMG classifier is about 98% in proportion to the existing state-of-the-art shape-based classifiers with optimal computing time.
- Добавил в систему: Местецкий Леонид Моисеевич