Search

Many species of plants produce leaves with distinct teeth around their margins. The presence and nature of these teeth can often help botanists to identify species. Moreover, it has long been known that more species native to colder regions have teeth than species native to warmer regions. It has therefore been suggested that fossilized remains of leaves can be used as a proxy for ancient climate reconstruction. Similar studies on living plants can help our understanding of the relationships. The required analysis of leaves typically involves considerable manual effort, which in practice limits the number of leaves that are analyzed, potentially reducing the power of the results. In this work, we describe a novel algorithm to automate the marginal tooth analysis of leaves found in digital images. We demonstrate our methods on a large set of images of whole herbarium specimens collected from Tilia trees (also known as lime, linden or basswood). We chose the genus Tilia as its constituent species have toothed leaves of varied size and shape. In a previous study we extracted c.1600 leaves automatically from a set of c.1100 images. Our new algorithm locates teeth on the margins of such leaves and extracts features such as each tooth's area, perimeter and internal angles, as well as counting them. We evaluate an implementation of our algorithm's performance against a manually analyzed subset of the images. We found that the algorithm achieves an accuracy of 85% for counting teeth and 75% for estimating tooth area. We also demonstrate that the automatically extracted features are sufficient to identify different species of Tilia using a simple linear discriminant analysis, and that the features relating to teeth are the most useful.

Motivation: Converting the vast quantity of free-format text found in journals into a concise, structured format makes the researcher's quest for information easier. Recently, several information extraction systems have been developed that attempt to simplify the retrieval and analysis of biological and medical data. Most of this work has used the abstract alone, owing to the convenience of access and the quality of data. Abstracts are generally available through central collections with easy direct access (e.g. PubMed). The full-text papers contain more information, but are distributed across many locations (e.g. publishers' web sites, journal web sites and local repositories), making access more difficult. In this paper, we present BioRAT, a new information extraction (IE) tool, specifically designed to perform biomedical IE, and which is able to locate and analyse both abstracts and full-length papers. BioRAT is a Biological Research Assistant for Text mining, and incorporates a document search ability with domain-specific IE., Results: We show first, that BioRAT performs as well as existing systems, when applied to abstracts; and second, that significantly more information is available to BioRAT through the full-length papers than via the abstracts alone. Typically, less than half of the available information is extracted from the abstract, with the majority coming from the body of each paper. Overall, BioRAT recalled 20.31% of the target facts from the abstracts with 55.07% precision, and achieved 43.6% recall with 51.25% precision on full-length papers.