MULTISPECTRAL IMAGING A NEW SEED ANALYSIS TECHNOLOGY? UNIVERSITY
OUTLINE Multispectral imaging Seed health Seed germination Seed purity Conclusions
MULTISPECTRAL IMAGING ultraviolet (UV) near-infrared (NIR) 200 300 400 500 600 700 800 900 1000 nm
MULTISPECTRAL IMAGING LED Wavelength (nm) Color Compound / application example 1 375 UVA Fluorescence, mycorradicin (Klingner et al., 1995) 2 405 Violet Melanins (Nosanchuk et al., 1998) 3 435 Chlorophyll A (absorption and excitation) (Salisbury Indigo and Ross, 1992) Riboflavin, Chlorophyll B, β-caroten (Salisbury and 4 450 Blue Ross, 1992)... 18 940 NIR Fat (Song et al., 2009) 19 970 NIR Water (Penuelas et al., 1993) LEDs: 19 different high-resolution bands acquired sequentially Combined with backlight and fluorescence filters
Structure
Chemical information
RED CLOVER SEED
THE RED CLOVER SEED - 19 IMAGES
MULTISPECTRAL IMAGING High resolution images - spectra of values of 19 wavelengths Each pixel in the image contains values expressing the refleced light from the surface of an object across different wavelengths Each wavelength provides specific information of surface compounds The spectra are related to the structure and chemistry of the surface
SEED HEALTH
SEED HEALTH Test methods Inspection of the dry seed Washing test Blotter method Agar plate method Embryo-count method Seedling symptom test PCR
SEED HEALTH IN SPINACH Experimental set-up Four commercial seed lots 16 x 25 seeds subjected to the freeze-blotter test Visual inspection Single-infected seeds used for propagation of inoculation material Sterile seeds were artificially infected Inspected my multispectral imaging
SEED HEALTH Six groups of 3 seeds (un-infected control, Stemphylium botryosum, Verticillium spp., Fusarium spp., Alternaria alternata, and Cladosporium spp.) analysed in visual and near infrared light. VIS 450nm NIR 870nm
SEED HEALTH 140 120 100 Alternaria Cladosporium Fusarium Stemphylium Uninfected Verticillium Mean intensity 80 60 40 20 Stemphylium 0 395 430 450 470 505 565 590 630 645 660 700 850 870 890 910 920 940 950 970 Wavelength (nm) Alternaria Fusarium Cladosporium Verticillium
HISTOGRAM OF PIXEL VALUE Alternaria spp. Fusarium spp. Stemphylium spp. Cladosporium spp. Verticillium spp. Uninfected Number of pixels 450nm 870nm Pixel intensity Seed Science and Technology Olesen et al., 2011
SEED HEALTH Multispectral imaging could distinguish between uninfected and infected seed and between fungi Using the blotter test Higher throughput if dry seeds can be analysed
MULTISPECTRAL IMAGE BARLEY, FUNGAL INFECTION Natural fungal infection G. zeae (red) M. tassia (yelow) A. infectoria (green) Dothidomycetes (white) G. Tricincta - the rest Identified by Next Generation Sequencing
FUNGAL INFECTION (ncda SCORING OF PIXELS)
ncda CLASSIFICATION
DETECTION OF FUSARIUM IN MALTING BARLEY Seeds with Fusarium spp. showing red signatures were identified in a 500 g sample with a standard deviation of 0,13 - with grey signatures - a standard deviation of 0,78
SEED GERMINATION
VIABILITY PREDICTION OF RICINUS CUMMUNIS L. SEEDS USING MULTISPECTRAL IMAGING Viable seeds were distinguished from dead seeds with 92% accuracy Olesen et al., 2015 Sensors doi:10.3390/s150204592 23
SEED GERMINATION Germination test: Normal seedlings Abnormal seedlings Hard seeds Fresh ungerminated Dead seeds
NORMAL / ABNORMAL SEEDLINGS Evaluate root:shoot ratio
NORMAL / ABNORMAL SEEDLINGS Identification of transition point
RADICLE EMERGENCE (RE) TEST Accepted vigour test since 2012 Does early counts of radicle emergence correlate with the number of normal seedlings?
RADICLE EMERGENCE COMMERCIAL GRASS SEED d02 h08 d02 h 14 d03 h 14 RE: 11 RE: 11+6 =17 RE: 17+4 =21 Information on normal seedlings (22) provided by the seed company laboratory
MULTISPECTRAL IMAGING - RADICLE EMERGENCE Caryopsis: 5,5/8,1 = 0,67 Radicle
GERMINATION TEST The evaluation of normal and abnormal seedlings is very difficult Multispectral imaging is useful to evaluate discoloration, necrosis etc. Multispectral imaging is useful for radicle emergence test
SEED PURITY
THE PURITY ANALYSIS The objective of the purity analysis is to determine The percentage composition by weight of the sample being tested The identity of the various species of seeds and inert particles constituting the sample.
SEED PURITY IN SPINACH Polygonum convolvulus Galeopsis tetrahit Galium aparine
PURITY ANALYSIS OF SPINACH SAMPLES Class Example Images Features Sensitivity Spinach Color, shape, 99.9% texture Cleavers / Galium aparine Primarily shape and texture 99.5% Black bindweed/ Polygonum convolvulus Texture, shape and color for 99.7%
SECOND LEVEL ATTRIBUTES - SEED CLASSIFICATION Classifier performance on test set with 57.115 seeds
COLOR FEATURES - DISCRIMINATE SPINACH AND RADISH SEEDS BY CDA TRANSFORMATION RGB view of radish seeds RGB view of spinach seeds ncda transform on multispectral image enhances color contrast between spinach and radish.
DISCRIMINATE SPINACH AND RADISH SEEDS BY CDA TRANSFORMATION Spectra for spinach seeds (green) and radish seeds (red) used to create CDA transform
CONCLUSIONS Multispectral imaging has potential for seed quality assessment Seed health testing: For diseases with signatures on the seed surface, blotter and agar plate method The standard germination test - difficult evaluation. Radicle emergence test may be a way forward Purity testing: Multispectral imaging has great potential, training with a diversity of samples required
ACKNOWLEDGEMENTS Ass. prof. Johannes R. Jørgensen Ass. prof. René Gislum Post doc. Merete Halkjær Olesen (The Food Administration agency) Post doc. Muhammad Imran (Yara International) Post doc. Santosh Shresta PhD-student Zahra Salimi Jens Michael Carstensen, Videometer Karsten Hartelius, Videometer
THE FUNDING
Thank you for your attention. UNIVERSITY