Microscopic Image Dataset of Fungal Spores from Grapevine
Creators
-
1.
Center for Scientific Research and Higher Education at Ensenada
Description
Development of a high-quality fungal spore microscope image dataset.
To build the fungal spore microscope image dataset, Lasiodiplodia species initially obtained from grapevines showing Botryosphaeria dieback were recovered from the fungal collection of the laboratory of plant pathology of CICESE. The samples collected belong to 4 different species of fungi, which are: L. brasiliensis, L. crassispora, L. exigua, and L. gilanensis. Isolates are properly prepared so that only a single species is found in a specific culture plate. To induce the production of spores (conidiospores), inside pycnidia, each fungus was grown onto Water Agar (15 g/L) supplemented with sterile pine needles (5 % w/v). Inoculated Petri dishes were incubated at room temperature under 12 h near ultraviolet electromagnetic radiation light and 12 h darkness for 15 days. Formed pycnidia were recovered, transferred to a slide with a drop of water, and pushed gently to squeeze spores, to be observed under a brighfield microscope.
Methods (English)
To increase the richness of the dataset, several representative samples of different biological strains from different vineyards in the northwest part of Mexico were used, particularly from the states of Baja California and Sonora. This was done to help improve the robustness and generalization capability of the trained deep learning models. Between two and four different strains were used for each Lasiodiplodia fungi.
A total of 1320 microscope images were acquired using a Nikon Eclipse E200 optical microscope. Three slides were prepared for each of the fungal strains, the biological samples were suspended in water and mounted on each microscope slide, and the spores were visualized using 100x magnification with immersion oil under brightfield illumination. Two IDS camera models, UI-3370CP-M and UI-1240SE-C, with monochromatic and color sensors, respectively, were used to capture grayscale and color images in order to observe the relevance of color information in the subsequent analysis. A total of 20 images were taken for each slide to capture as many different morphological features as possible.
Files
fungal_data.zip
Files
(1.8 GB)
| Name | Size | Download all |
|---|---|---|
|
md5:20db79f7b6afad73bb326a61f97e23ce
|
1.8 GB | Preview Download |