Dataset Information

Automated graptolite identification at high taxonomic resolution using residual networks.

ABSTRACT: Graptolites, fossils significant for evolutionary studies and shale gas exploration, are traditionally identified visually by taxonomists due to their intricate morphologies and preservation challenges. Artificial intelligence (AI) holds great promise for transforming such meticulous tasks. In this paper, we demonstrate that graptolites can be identified with taxonomist accuracy using a deep learning model. We construct the most sophisticated and largest professional single organisms image dataset to date, which is composed of >34,000 images of 113 graptolite species annotated at pixel-level resolution to train the model, develop, and evaluate deep learning networks to classify graptolites. The model's performance surpassed taxonomists in accuracy, time, and generalization, achieving 86% and 81% accuracy in identifying graptolite genus and species, respectively. This AI-based method, capable of recognizing minute morphological details better than taxonomists, can be integrated into web and mobile apps, extending graptolite identification beyond research institutes and enhancing shale gas exploration efficiency.

SUBMITTER: Niu ZB

PROVIDER: S-EPMC10783601 | biostudies-literature | 2024 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Automated graptolite identification at high taxonomic resolution using residual networks.

Niu Zhi-Bin ZB Jia Si-Yuan SY Xu Hong-He HH

iScience 20231123 1

Graptolites, fossils significant for evolutionary studies and shale gas exploration, are traditionally identified visually by taxonomists due to their intricate morphologies and preservation challenges. Artificial intelligence (AI) holds great promise for transforming such meticulous tasks. In this paper, we demonstrate that graptolites can be identified with taxonomist accuracy using a deep learning model. We construct the most sophisticated and largest professional single organisms image datas ...[more]

PMID: 38213629

Similar Datasets

Project description:Rapid and reliable identification of insects is important in many contexts, from the detection of disease vectors and invasive species to the sorting of material from biodiversity inventories. Because of the shortage of adequate expertise, there has long been an interest in developing automated systems for this task. Previous attempts have been based on laborious and complex handcrafted extraction of image features, but in recent years it has been shown that sophisticated convolutional neural networks (CNNs) can learn to extract relevant features automatically, without human intervention. Unfortunately, reaching expert-level accuracy in CNN identifications requires substantial computational power and huge training data sets, which are often not available for taxonomic tasks. This can be addressed using feature transfer: a CNN that has been pretrained on a generic image classification task is exposed to the taxonomic images of interest, and information about its perception of those images is used in training a simpler, dedicated identification system. Here, we develop an effective method of CNN feature transfer, which achieves expert-level accuracy in taxonomic identification of insects with training sets of 100 images or less per category, depending on the nature of data set. Specifically, we extract rich representations of intermediate to high-level image features from the CNN architecture VGG16 pretrained on the ImageNet data set. This information is submitted to a linear support vector machine classifier, which is trained on the target problem. We tested the performance of our approach on two types of challenging taxonomic tasks: 1) identifying insects to higher groups when they are likely to belong to subgroups that have not been seen previously and 2) identifying visually similar species that are difficult to separate even for experts. For the first task, our approach reached $CDATA[$CDATA[$>$$92% accuracy on one data set (884 face images of 11 families of Diptera, all specimens representing unique species), and $CDATA[$CDATA[$>$$96% accuracy on another (2936 dorsal habitus images of 14 families of Coleoptera, over 90% of specimens belonging to unique species). For the second task, our approach outperformed a leading taxonomic expert on one data set (339 images of three species of the Coleoptera genus Oxythyrea; 97% accuracy), and both humans and traditional automated identification systems on another data set (3845 images of nine species of Plecoptera larvae; 98.6 % accuracy). Reanalyzing several biological image identification tasks studied in the recent literature, we show that our approach is broadly applicable and provides significant improvements over previous methods, whether based on dedicated CNNs, CNN feature transfer, or more traditional techniques. Thus, our method, which is easy to apply, can be highly successful in developing automated taxonomic identification systems even when training data sets are small and computational budgets limited. We conclude by briefly discussing some promising CNN-based research directions in morphological systematics opened up by the success of these techniques in providing accurate diagnostic tools.

Project description:BackgroundLipids are ubiquitous and serve numerous biological functions; thus lipids have been shown to have great potential as candidates for elucidating biomarkers and pathway perturbations associated with disease. Methods expanding coverage of the lipidome increase the likelihood of biomarker discovery and could lead to more comprehensive understanding of disease etiology.ResultsWe introduce LipidMatch, an R-based tool for lipid identification for liquid chromatography tandem mass spectrometry workflows. LipidMatch currently has over 250,000 lipid species spanning 56 lipid types contained in in silico fragmentation libraries. Unique fragmentation libraries, compared to other open source software, include oxidized lipids, bile acids, sphingosines, and previously uncharacterized adducts, including ammoniated cardiolipins. LipidMatch uses rule-based identification. For each lipid type, the user can select which fragments must be observed for identification. Rule-based identification allows for correct annotation of lipids based on the fragments observed, unlike typical identification based solely on spectral similarity scores, where over-reporting structural details that are not conferred by fragmentation data is common. Another unique feature of LipidMatch is ranking lipid identifications for a given feature by the sum of fragment intensities. For each lipid candidate, the intensities of experimental fragments with exact mass matches to expected in silico fragments are summed. The lipid identifications with the greatest summed intensity using this ranking algorithm were comparable to other lipid identification software annotations, MS-DIAL and Greazy. For example, for features with identifications from all 3 software, 92% of LipidMatch identifications by fatty acyl constituents were corroborated by at least one other software in positive mode and 98% in negative ion mode.ConclusionsLipidMatch allows users to annotate lipids across a wide range of high resolution tandem mass spectrometry experiments, including imaging experiments, direct infusion experiments, and experiments employing liquid chromatography. LipidMatch leverages the most extensive in silico fragmentation libraries of freely available software. When integrated into a larger lipidomics workflow, LipidMatch may increase the probability of finding lipid-based biomarkers and determining etiology of disease by covering a greater portion of the lipidome and using annotation which does not over-report biologically relevant structural details of identified lipid molecules.

Project description:Background and aimsHigh-resolution microendoscopy (HRME) is an optical biopsy technology that provides subcellular imaging of esophageal mucosa but requires expert interpretation of these histopathology-like images. We compared endoscopists with an automated software algorithm for detection of esophageal squamous cell neoplasia (ESCN) and evaluated the endoscopists' accuracy with and without input from the software algorithm.MethodsThirteen endoscopists (6 experts, 7 novices) were trained and tested on 218 post-hoc HRME images from 130 consecutive patients undergoing ESCN screening/surveillance. The automated software algorithm interpreted all images as neoplastic (high-grade dysplasia, ESCN) or non-neoplastic. All endoscopists provided their interpretation (neoplastic or non-neoplastic) and confidence level (high or low) without and with knowledge of the software overlay highlighting abnormal nuclei and software interpretation. The criterion standard was histopathology consensus diagnosis by 2 pathologists.ResultsThe endoscopists had a higher mean sensitivity (84.3%, standard deviation [SD] 8.0% vs 76.3%, P = .004), lower specificity (75.0%, SD 5.2% vs 85.3%, P < .001) but no significant difference in accuracy (81.1%, SD 5.2% vs 79.4%, P = .26) of ESCN detection compared with the automated software algorithm. With knowledge of the software algorithm, the specificity of the endoscopists increased significantly (75.0% to 80.1%, P = .002) but not the sensitivity (84.3% to 84.8%, P = .75) or accuracy (81.1% to 83.1%, P = .13). The increase in specificity was among novices (P = .008) but not experts (P = .11).ConclusionsThe software algorithm had lower sensitivity but higher specificity for ESCN detection than endoscopists. Using computer-assisted diagnosis, the endoscopists maintained high sensitivity while increasing their specificity and accuracy compared with their initial diagnosis. Automated HRME interpretation would facilitate widespread usage in resource-poor areas where this portable, low-cost technology is needed.

Project description:BackgroundThe digitization of biodiversity data is leading to the widespread application of taxon names that are superfluous, ambiguous or incorrect, resulting in mismatched records and inflated species numbers. The ultimate consequences of misspelled names and bad taxonomy are erroneous scientific conclusions and faulty policy decisions. The lack of tools for correcting this 'names problem' has become a fundamental obstacle to integrating disparate data sources and advancing the progress of biodiversity science.ResultsThe TNRS, or Taxonomic Name Resolution Service, is an online application for automated and user-supervised standardization of plant scientific names. The TNRS builds upon and extends existing open-source applications for name parsing and fuzzy matching. Names are standardized against multiple reference taxonomies, including the Missouri Botanical Garden's Tropicos database. Capable of processing thousands of names in a single operation, the TNRS parses and corrects misspelled names and authorities, standardizes variant spellings, and converts nomenclatural synonyms to accepted names. Family names can be included to increase match accuracy and resolve many types of homonyms. Partial matching of higher taxa combined with extraction of annotations, accession numbers and morphospecies allows the TNRS to standardize taxonomy across a broad range of active and legacy datasets.ConclusionsWe show how the TNRS can resolve many forms of taxonomic semantic heterogeneity, correct spelling errors and eliminate spurious names. As a result, the TNRS can aid the integration of disparate biological datasets. Although the TNRS was developed to aid in standardizing plant names, its underlying algorithms and design can be extended to all organisms and nomenclatural codes. The TNRS is accessible via a web interface at http://tnrs.iplantcollaborative.org/ and as a RESTful web service and application programming interface. Source code is available at https://github.com/iPlantCollaborativeOpenSource/TNRS/.

Dataset Information

Automated graptolite identification at high taxonomic resolution using residual networks.

Publications

Automated graptolite identification at high taxonomic resolution using residual networks.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets