Optical Coherence Tomography as a Non-Invasive Tool for Plant Material Characterization in Agriculture: A Review

Abstract

Characterizing plant material is crucial in terms of early disease detection, pest control, physiological assessments, and growth monitoring, which are essential parameters to increase production in agriculture and prevent unnecessary economic losses. The conventional methods employed to assess the aforementioned parameters have several limitations, such as invasive inspection, complexity, high time consumption, and costly features. In recent years, optical coherence tomography (OCT), which is an ultra-high resolution, non-invasive, and real-time unique image-based approach has been widely utilized as a significant and potential tool for assessing plant materials in numerous aspects. The obtained OCT cross-sections and volumetrics, as well as the amplitude signals of plant materials, have the capability to reveal vital information in both axial and lateral directions owing to the high resolution of the imaging system. This review discusses recent technological trends and advanced applications of OCT, which have been potentially adapted for numerous agricultural applications, such as non-invasive disease screening, optical signals-based growth speed detection, the structural analysis of plant materials, and microbiological discoveries. Therefore, this review offers a comprehensive exploration of recent advanced OCT technological approaches for agricultural applications, which provides insights into their potential to incorporate OCT technology into numerous industries.

Keywords: agriculture; disease detection; image analysis; image processing; optical coherence tomography; optical imaging.

Figure 1

The schematic diagram represents an application overview of OCT in agriculture. (a) System schematic of optical coherence tomography. (b) OCT applications in agriculture.

Figure 2

A-scan analysis for assessing OCT images. (a) The cross-sectional OCT image acquisition process. (b) Single A-scan intensity profiling. (c) Average A-scan of an image. (d) Average A-scan of multiple images.

Figure 3

Comparison of healthy and CGMMV-infected cucumber seeds through OCT images and A-scan profiles: (i) 3D and en-face images (figure source [55]). (i(A,B)) three dimensional OCT images, and (i(C,D)) is the XY plane images of the healthy and infected cucumber seeds. (ii) OCT en-face images of cucumber seed and the corresponding A-scan profile (figure source [55]). (iii) The difference between healthy and CGMMV-infected seeds after heat-drying (A) or water immersion (B). SC, seed coat; ISG, infected seed gap; HSG, healthy seed gap; IE, infected seed endosperm; HE, healthy seed endosperm; GD, gap distance between healthy and CGMMV-infected seeds (figure source [55]).

Figure 4

A 3D illustration of the changes in seed morphology as a result of germination: (a–c) 3D, en-face, and cross-sectional OCT images of a seed treated with SDW; (d–f) 3D, en-face, and cross-sectional OCT images of a seed treated with butanediol; (g–i) 3D, en-face, and cross-sectional OCT images of a seed treated with 1-hexadecene. C, cotyledons; E, endosperm; EM, embryo; H, hypocotyl; ME, micropylar endosperm; R, radicle; SC, storage cotyledons; SP, sprout; T, testa (seed coat). The horizontal and vertical scale bars are 700 μm and 200 μm, respectively (figure source [59]).

Figure 5

OCT and A-scan profiles for revealing the different phases of the germination at an interval of 3 h. (i) OCT images of the germination phases. Images (i(A–D)), (i(E–I)), and (i(J–L)) show phases I and III, and their corresponding A-scan profiles (i(a–d)), (i(e–i)), and (i(j–l)), obtained from positions 1, 2, and 3, are marked by the red, blue, and green colors, respectively (figure source [61]). The cotyledon layer becomes visible in (i(C)) marked by red circle; the radicle first observed in (i(E)) marked by blue circle; the seed coat cracked in (i(J)) marked by yellow circle. (ii) Boxplot for the thickness of the seed coat at different germination phases, where (ii(a–l)) represent the corresponding data for the A-scan shown in (i(a–l)) (figure source [61]).

Figure 6

Comparison of healthy, apparently healthy, and infected leaves examined using OCT cross-sectional images and A-scan profiles. (i) The 2D cross-sectional images of healthy, apparently healthy, and infected persimmon and apple leaves. (i(a–c)) OCT cross-sectional images of persimmon leaves. (i(d–f)) OCT cross-sectional images of apple leaves (figure source [64]). (ii) Depth profiles of healthy, apparently healthy, and infected apple leaves; (ii(a,b)), (ii(c,d)), and (ii(e,f)) show the depth profiles of healthy, apparently healthy, and infected apple leaves, respectively. (ii(a,c,e)) Depth intensity profiles of three regions of interest (ROIs) from a single leaf. (ii(b,d,f)) The averaged depth intensity profiles of three ROIs from a single leaf (figure source [64]).

Figure 7

Comparison of healthy, apparently healthy, and infected persimmon leaves using A-scan profiles. (a–i) A-scan profiles of healthy, apparently healthy, and infected leaves, respectively. (a,d,g), (b,e,h), and (c,f,i) A-scan profiles of four ROIs, average depth profiles of four ROIs, and curve-fitted depth profiles of four ROIs of a single persimmon leaf, respectively. UE: upper epidermis, PP: palisade parenchyma, SP: spongy parenchyma (figure source [64]).

Figure 8

Illustration of the structural differences in fruit specimens through histology, OCT cross-section, and en-face images with the progression of the disease. (i) Histological validation of the 2D OCT images obtained from healthy and naturally infected fruit specimens. (i(a,b)) Morphology of healthy fruit. (i(c,d)) Morphology of apparently healthy fruit. (i(e,f)) Morphology of entirely infected fruit (figure source [49]). (ii) Illustration of the inner structure of healthy fruit and the changes in its structure at the depth direction, using en-face images. (ii(a–d)) The 3D and en-face images of healthy specimens at depth direction. Images (ii(e–l)) illustrate the morphological changes in depth direction with disease progression (figure source [49]).

Figure 9

Illustration of the development of progressive rind breakdown (RBD) in mandarin fruit: (i(a)) unaffected fruit, (i(b)) mildly affected, (i(c)) moderately affected, and (i(d)) severely affected (figure source [72]). (ii) Oil glands in 3D representation with (a) no RBD, (b) moderate RBD, and (c) severe RBD (figure source [72]).

Figure 10

Compact, portable, and wearable OCT imaging modality configuration. (i) Schematic diagram of the OCT system. BLS, broadband laser source; C, collimator; CB, capture button; DG, diffraction grating; FC, fiber coupler; GS, Galvano scanner; L, lens; LCD, liquid crystal display; LSC, line scanning camera; M, mirror; PC, polarization controller; S, sample (figure source [50]). (ii) Wearable imaging modality appearance. (ii(a)) complete system, (ii(b)) wearable OCT with its operator, (ii(c)) wearable OCT in action, and (ii(d)) display of in vivo and real-time imaging (figure source [50]).

Figure 11

Assessment of biofilm thickness using OCT imaging: (i(A,C)) dripper system, (i(B)) test bench (figure source [78]) The test bench was composed of 1. a tank (60 l); 2. a water pump; 3. a 0.13 mm mesh screen filter; 4. a pressure reducer; 5. a pressure gauge; 6. the drip line with an emitter system located at 10-cm intervals; 7. a collector; 8. a gutter. (ii(A)) Measurement of biofilm thickness at the inlet of drippers. (ii(B)) Measurement of biofilm thickness in the return of drippers. (ii(C)) The return areas of drippers were measured after 1 and 4 months (figure source [78]).