Mask, Train, Repeat! Artificial Intelligence for Quantitative Wood Anatomy

Abstract

The recent developments in artificial intelligence have the potential to facilitate new research methods in ecology. Especially Deep Convolutional Neural Networks (DCNNs) have been shown to outperform other approaches in automatic image analyses. Here we apply a DCNN to facilitate quantitative wood anatomical (QWA) analyses, where the main challenges reside in the detection of a high number of cells, in the intrinsic variability of wood anatomical features, and in the sample quality. To properly classify and interpret features within the images, DCNNs need to undergo a training stage. We performed the training with images from transversal wood anatomical sections, together with manually created optimal outputs of the target cell areas. The target species included an example for the most common wood anatomical structures: four conifer species; a diffuse-porous species, black alder (Alnus glutinosa L.); a diffuse to semi-diffuse-porous species, European beech (Fagus sylvatica L.); and a ring-porous species, sessile oak (Quercus petraea Liebl.). The DCNN was created in Python with Pytorch, and relies on a Mask-RCNN architecture. The developed algorithm detects and segments cells, and provides information on the measurement accuracy. To evaluate the performance of this tool we compared our Mask-RCNN outputs with U-Net, a model architecture employed in a similar study, and with ROXAS, a program based on traditional image analysis techniques. First, we evaluated how many target cells were correctly recognized. Next, we assessed the cell measurement accuracy by evaluating the number of pixels that were correctly assigned to each target cell. Overall, the “learning process” defining artificial intelligence plays a key role in overcoming the issues that are usually manually solved in QWA analyses. Mask-RCNN is the model that better detects which are the features characterizing a target cell when these issues occur. In general, U-Net did not attain the other algorithms’ performance, while ROXAS performed best for conifers, and Mask-RCNN showed the highest accuracy in detecting target cells and segmenting lumen areas of angiosperms. Our research demonstrates that future software tools for QWA analyses would greatly benefit from using DCNNs, saving time during the analysis phase, and providing a flexible approach that allows model retraining.

Automation of tree‐ring detection and measurements using deep learning

“…Though trained only on Picea abies , the consistent structure of conifer wood resulted in similar precision and recall values in other conifer species. These findings are consistent with other studies (Fabijańska et al, 2017; Martinez‐Garcia et al, 2021; Resente et al, 2021). These results illustrate our CNN's out‐of‐the‐box utility while suggesting results can be improved for specific applications with the re‐trainer tool.…”

“…Increased use of imaged increment cores (Griffin et al, 2021; Larsson, 2003; Rademacher et al, 2021) provides an opportunity to automate detection and subsequent measurement of annual ring widths. Previous attempts at identifying ring boundaries have had variable and sometimes impressive success (Conner et al, 1998; Fabijańska et al, 2017; Fabijańska & Danek, 2018; Martinez‐Garcia et al, 2021; Resente et al, 2021) but the only deployable method is limited to three‐dimensional images (Martinez‐Garcia et al, 2021). Here, we present a new supervised Mask R‐CNN and processing pipeline available as a Docker container.…”

Improving ecological insights from dendroecological studies of Arctic shrub dynamics: Research gaps and potential solutions

“…The recent advancements in methods and standard protocols (Gärtner and Schweingruber, 2013;von Arx et al, 2016), automated image-analysis systems (von Arx et al, 2013;von Arx and Carrer, 2014;von Arx and Dietz, 2005) and processing using artificial intelligence (Resente et al, 2021) allow to significantly increase the number of measured anatomical features, while also reducing the time required for the analysis. Nevertheless, the establishment of the new quantitative wood anatomy network (Q-NET) reflects the need of a growing community to share knowledge and experience, but also to advance methodologies (Resente et al, 2021) and to profit from interdisciplinary collaborations to tackle broader ecological questions at multiple temporal and spatial scales (von Arx et al, 2021).…”

Automation of tree-ring detection and measurements using deep learning

“…This is a powerful tool that can greatly cut down on the processing time to generate ring-width measurements while maintaining the benefits of image based measurements (Griffin et al, 2021; Maxwell & Larsson, 2021). Though trained only on Picea abies , the consistent structure of conifer wood means it should be easily deployable across other conifer species (Fabijańska et al, 2017; Martinez-Garcia et al, 2021; Resente et al, 2021). We consider this CNN to be a valuable tool for efficiently generating data that can reduce analysis time considerably, especially in large datasets or for potential reanalysis of old unmeasured datasets.Though our CNN performed well with 95% of its detections correct and correctly identifying 99% of true rings, these errors still prohibit an unsupervised use of the generated data and requires an experienced dendrochronologist to oversee crossdating (Holmes et al, 1986).…”

“…Increased use of imaged increment cores (Griffin et al, 2021;Larsson, 2003;Rademacher et al, 2021) provides an opportunity to automate detection and subsequent measurement of annual ring widths. Previous attempts at identifying ring boundaries have had variable and sometimes impressive AUTOMATED TREE-RING DETECTION AND MEASUREMENT success (Conner et al, 1998;Fabijańska et al, 2017;Fabijańska & Danek, 2018;Martinez-Garcia et al, 2021;Resente et al, 2021) but the only deployable method is limited to three-dimensional images (Martinez-Garcia et al, 2021). Here we present a new supervised Mask R-CNN and processing pipeline available as a Docker container.…”