Introducing Fiji and ICY image processing techniques in ichnological research as a 2 tool for sedimentary basin analysis

Abstract In recent years, image treatment has been appraised as a very powerful tool to facilitate ichnological analysis, especially in marine cores of modern sediments, supporting the determination of certain ichnological features. However, it is still a new approach and detailed research is necessary to encounter a faster and more efficient method. The present study focuses on two image processing techniques, Fiji and ICY, and their comparison with a refined version of the well-established high-resolution image treatment. Strengths and weaknesses of the methodologies for the determination of three main features were explored: i) visibility of trace fossils; ii) quantification of the percentage of bioturbated surface, and iii) penetration depth estimation. Refined high-resolution image treatment gives the best results for enhanced visibility of trace fossils, whereas Fiji is found to be a sound and rapid option. One disadvantage shared by Fiji and ICY is the binary character of the produced images, which may impede later ichnotaxonomical differentiation. Both Fiji and ICY (+ Fiji) are rapid alternatives for quantifying the bulk amount of bioturbated surface. The Magic Wand Method (+ RefineEdge), based on high-resolution image treatment, provides good results regardless of the contrast of the images, and it additionally allows for a more detailed quantification. The semi-automatic character of ICY favors quick estimation of penetration depth and facilitates differentiation between distinct tracemaker communities, based on a rapid quantification of pixel values. Thus, Fiji and ICY methods offer good results and are much less time-consuming than high-resolution image treatment. They are proposed as faster alternatives for the estimation of ichnological features, especially useful at the beginning stages of research, when a large number of samples must be analyzed.

resolution image treatment gives the best results for enhanced visibility of trace fossils, 23 whereas Fiji is found to be a sound and rapid option. One disadvantage shared by Fiji and 24 ICY is the binary character of the produced images, which may impede later  developments (e.g., Rodríguez-Tovar and Dorador, 2014;Rodríguez-Tovar et al., 2015a, 66 b; Dorador and Rodríguez-Tovar, 2016a;Dorador et al., 2016) and provides information 67 for sedimentary basin research (e.g., Alonso et al., 2016;Dorador and Rodríguez-Tovar, 68 2016b; Takashimizu et al., 2016), supported by the ichnological approach. So far, these 69 techniques have been mainly applied on marine cores from modern deposits. However, 70 Dorador and Rodríguez-Tovar (2018) recently showed that image treatment might be an  research (e.g., Grove and Jerran, 2011;Goldstein et al., 2017). ImageJ has been 81 punctually used in ichnological analysis to enhance visibility of certain image attributes, 82 to estimate bioturbated surface and to make shape/length measurements (Francus, 2001;83 Nicolo et al., 2010;Lauridsen et al., 2011;Curth et al., 2014). Fiji has already been 84 applied in neo-ichnological analysis for format conversion in reconstructed volumes 85 obtained by computed tomography of invertebrate burrow systems (Hale et al., 2015), 86 and recently to estimate the degree of bioturbation 87 2019). This method holds great potential for ichnological studies because its toolbox 88 features diverse tools and filters for the measurement of specific paleontological 5 89 parameters (i.e., shape/length of bones; Iepure et al., 2012;Jarrett, 2016;O'Connor et al., 90 2018).

91
Whereas ICY software has been applied in cell biology studies for tracking, 92 extracting or actively delimiting particles/cells (e.g., de Chaumont et al., 2011;Meijering 93 et al., 2012;Montagnac et al., 2013), it has not been previously applied to geological 94 research. The present research marks the first documented testing of its usefulness in the 95 realm of ichnological aspects.

96
The aim of this paper is to examine two image processing techniques, Fiji and  Fiji software offers a wide range of plugins for enhancing image visibility, similar 116 to the image adjustments of Adobe Photoshop (e.g., brightness, hue, saturation, levels).

117
For the purposes of this study, to provide quick image visibility improvement, the 118 Contrast Limited Adaptive Histogram Equalization (CLAHE) method was selected (Fig.   119   1A). This provides better visibility of ichnological features in a short time through the 120 local contrast of an image based on modification of two main parameters: i) block size, 121 controlling the size of the local region around a pixel for which the histogram is equalized; 122 and ii) histogram bins, defining the number of bins used for histogram equalization, which 123 should be smaller than the number of pixels in a block.

124
For the quantification of bioturbated surface, after applying CLAHE, the obtained 125 image must be converted to an 8-bit grayscale image. Then, the threshold has to be 126 defined to establish a differentiation range for black and white by means of the Image 127 menu. The determination of this parameter is supported by a preview screen, where the 128 potential image is observed, and the value can be modified to obtain a more appropriate 129 bioturbated differentiation (black pixels were assigned to bioturbation). The enhanced 130 black and white binary image can be used to quantify the amount of bioturbated surface 131 ( Fig. 1B left). The Fiji process menu offers a wide range of easy-to-use binary tools and 132 filters (e.g., dilate, erode, fill holes and skeletonize, among others). After testing, the 133 following ones were selected as the most suitable for ichnological analysis: a) Erode: it 134 removes erratic black pixels from the edges of bioturbation, proving extremely useful to 135 delete minor compounds of black pixels that do not correspond to trace fossils, and 136 thereby more precisely discern the shape of the burrows; b) Fill holes: it fills white pixels 137 located inside the trace fossils that are surrounded by black ones; and c) 138 Minimum/Maximum filters: they perform binary erosion by replacing each pixel in the 139 image with the smallest/largest pixel value in that pixel's neighborhood, helping one to 7 140 remove small black pixels that conform background noise (Fig. 1B right). All these tools 141 and filters present a preview option, which makes it easy, quick and intuitive to select the 142 size of the neighborhood.

143
Finally, manual corrections can be conducted using eraser or painting tools, 144 controlling that all the selected pixels belong to trace fossils. Afterwards, output 145 measurement results are quickly obtained with the Fiji analyze menu, whose program  (Fig. 2B left). The selection may also be obtained automatically using the K-Means 158 tool, whose algorithm calculates the threshold value after defining the number of "classes" 159 (i.e., areas in the histogram) to be differentiated (value 2 for binary images). In the case  2). Finally, application on the low contrast example gives more variable values, from 19% 271 to 47%; estimations obtained with CRSM and MWM represent a low index of 272 bioturbation (BI = 2), while that from SPSM corresponds to a higher index (BI = 3).

273
In conjunction with Fiji and ICY methods, the amount of bioturbated surface 274 corresponding to discrete trace fossils was calculated before and after filtering binary higher contrast image corresponding to U1385A-5H-CC, and a lower contrast image to 296 U1385A-8H-5A (Fig. 6).

297
In general, the results of the two methods were similar, but some noteworthy 298 differences should be pointed out (Fig. 6). For the higher contrast example, the high-299 resolution image treatment improved visualization of three Thalassinoides cross-sections 300 and the quantitative pixel analysis allowed the identification of corresponding penetration 301 depths of 2.8, 3.6 and 9.8 cm (Fig. 6A left). The IP-ICY shows two intensity value 302 packages that can be linked to Thalassinoides; the first records penetration depths of 2.6, 303 3.5 and 5.5 cm, and the second shows lower intensity values, giving an estimated 304 penetration depth of 9.5 cm (Fig. 6A right). For the low contrast core interval, according 305 to quantitative pixel analysis, four Thalassinoides cross-sections showed penetration 306 depths of 1.8, 6.3 and 10.0 (in two cases) cm (Fig. 6B left). Using IP-ICY it is more 307 difficult to distinguish between the fill of trace fossils and the host sediment, although 308 four different intensity value packages might be associated with the four Thalassinoides 309 cross-sections, having penetration depths of 2, 9, 10 and 11 cm.  (Table 1).

318
In terms of enhanced trace fossil visualization -facilitating trace fossil 319 differentiation and therefore ichnotaxonomical classification-the three methods provide 320 good results for high contrast images (Fig. 4)  The ichnofabric approach has been successfully used as a tool in a range of Earth