You are logged out.
Geology is a fascinating physical science that keeps inspiring new generations of students. Some of them are lucky enough to access some of the most unique geological formations where we can truly say that nature was confused...The Alps, The Andes, Hawaii, etc. However, nowadays, there is a growing demand for quantitative and automatic results excent of researcher bias, yet we place ourselves on an initial IT learning curve with very little collaboration with software engineers. This trend will certainly unlock a great number of broader social contributions (renewable energy, energy-critical elements exploration, geological risk assessment, climate change, etc.).
If you are a geologist, the Earth is your laboratory, it's your book of time. When studying a rock, unlike Maths, Chemistry, or Physics researchers, we place ourselves in a vast physical world. Works in a new area are not easy and follow an interative process with field excursions, sample preparation, analysis, interpretation, and downstream collaborative efforts (see video). Ground-truth veryfications cannot be done with scarse resources and experimental demonstrations are mostly unfeasible, unless doing precise time calibrations and radiometric dating, which means reducing the scale of observation. This lead to studies having a long maduration time (considering technological limitations) and tending to be academical to fill the gaps. Thus, research is barely responding to industry issues that require real-time integration of high-level interpretations from expensive but empirical data, thus relying on the expertise of followers of geoscientific gurus (e.g.: Mineral exploration: Richard Sillitoe, Sampling: Pierre Gy, Geostatistics: Georges Matheron, Geomechanics: Richard Bieniawski, etc.).
This project facilitates the advanced research process and empowers researchers without a long fieldwork record, allowing analysis of screened minerals and generation of representative outputs at the trace element level, for example penalty or bonus metals acompaigning precious and base metals in raw materials (Mining). The webpage presents an advanced virtual microscope (pilot version) for navigating and interrogating high-resolution rock sample slide images. It form parts of a growing branch of projects interested in image processing and aims to developing and bringing many of the IT innovations in microscopy from pioneering fields to geological applications.
We are looking forward a digital future with automation and greater research scalability. Whether an amateur observer, collector or computer scientist looking for new knowledge boundaries, curiosity and desire to develop more powerful observational and computational skills might have broght you here. The virtual microscope adapted to Geology (see video) will help you remotely showcasing samples and databases with your peers without necessarily accessing a high-tech microscope in a basement.
MBS-49 beds 4 baths 4,250 sqft. Feeder of Zn-Pb-rich lens. Sample showing recrystallized py +/- apy porphyroblast in a silicate matrix. Minor Pb-Zn minerals.
MBS-916 beds 4 baths 4,250 sqft. Zn-Pb lens "mount" showing intermediate-sulfidation assemblage overprinted by high-sulfidation veins (right)
MBS-576 beds 4 baths 4,250 sqft. 480 12th, Unit 14, San Francisco, CA
MBS-696 beds 4 baths 4,250 sqft. 480 12th, Unit 14, San Francisco, CA
MBS-126B6 beds 4 baths 4,250 sqft. 480 12th, Unit 14, San Francisco, CA
We locally work in well-implemented facilities that have circa 2017 state-of-the-art optical and mass spectrometry microscopy imaging systems. Mass spectrometry data, in its raw form, corresponds to a 2D plot of element or molecule masses against abundance or time. In this form (e.g.: spot analysis), it is devoid of the critical information of spatial context normally within rectangular areas. By contrast, mass spectrometry imaging (MSI) powerfully places atoms and molecules directly within their native context for ecological, biological, medical and geological studies.
Optical Microscopy with Polarized Light (Petrographic Microscope)
The oldest microscopy technique applied in Geology yet it has the greatest automation potential. It shows characteristic properties of minerals and delineates the texture at the sample surface. The acquisition produces high-resolution photographs that are overall superior to the other methods for WSI and can be used to do image segmentation works and ideally overcome the sub-grain problem once combined with MSI data. It uses incident visible light (VI) oscillating on the same plane (polarized) and colors depend on differential interactions of the pathway inside anisotropic mineral phases. A dedicated digital camera with a CMOS sensor automatically captures the light at the top end of the pathway generating a raster of tiles.
Scanning Electron Microscopy with Energy Dispersive Spectrometry and Back-scattered Electron detectors
A well stablished technique used since the 30s that has the best spatial resolution providing major and minor element compositions on well polished and C-coated samples. It uses a column (FEG) that accelerates an electron beam that incides in a sample on a nanometric-scale spot creating numerous interactions (elastic or inelastic) withing the constituting elements (electronic transitions). Then, a set of detectors clustered around the chamber capture incident secondary electrons, backscattered electrons and X-ray radiations producing a signal that is digitalized, processed, and constantly refreshed in online PC screens (see video). In fact, acquired images represents a volumetric sight near the sample surface from the interaction volume.
Laser Ablation Inductively-coupled Plasma Mass Spectrometry with Time Of Flying
A ground-greaking technique that is fastly developing with the greatest elemental sensibility (spectral resolution) but the lowest spatial resolution. It has the potential to provide 80 analytes concentrations down to parts per trillion (ppt) and 10-min LA mapping experiments. It employs a LA system (UV light) that erodes the sample surface and introduces particles to an elemental analyzer (MS) where they are nebulized, atomized, and ionized before a magnetic field separates them by they m/q+ ratio for TOF measuments. With proper synchronization, signals can be recorded at discrete positions to generate elemental images.
The techniques outlined above must be applied sequentially from left to right (like the old days) and not without a backing field campaign and comprehensive study that justify their application. The inverse approach is sometimes used for simultaneous testing of mature hypotheses and it is most probably a waste of money. The field to laboratory work proportion equilibrium should be around 50/50 for solving real-world problems. Moreover, the basic toolkit for knowing what to do is having a periodic table and understanding the electromagnetic spectrum.
Our New Methods
We focus on acquiring high-quality image/map information of cm2 areas of rocks and statistical analysis. This approach is still undeveloped using scientific tools, versus using automatic mineralogy systems, but has demonstrated high application potential benefiting from the great pixel population that can be interrogated for pulling geochemical data.
The following routines can be applied to your thin section:
- Whole-slide image acquisition at high resolutions (spectral and spatial) in OM and SEM.
- Analytical data and metadata management (MatLab).
- Image stack pyramiding, registration and interrogation (QGIS).
- LA-ICP-MS maps image registration (MatLab).
- Image analysis with HCA of PCA (MatLab).
Join & Contribute
In multidisciplinary projects collaboration has always been a key for success. We are looking forward to develop and learn from innovative IT technologies looking forward geological applications. For this we have required filling methodological gaps for boosting analysis quantification, speed, and representativeness. You can also contribute pilot testing this platform and inspiring us to develop more scripts. Your will also benefit enriching our rock sample collection to make it public. You might already be working on your own or like the challenge of fixing one of the following issues.
Image Registration & Tile Distortion
Geology requires open-source image alignment implementations for rigid/non-rigid transformation models for generating better image stacks (QGIS workaround) and correct image tile distortion (previous rectification step). This would enable correlation microscopy works.Learn More
Developing scripts for registering a large tile collection (batch processing) and creating large image mosaics provides many more possibilities and produces less artifacts with respect to using state-of-the-art proprietary software.Learn More
Front-end Web Development
Separation of Concerns (SoC)
This is a basic design principle will be followed in the web developing process for creating 2 different processing streams for a "navigational" layer stack (rendered) and a "compositional" array (hidden from platform viewport). The main motivation is circunventing slow image pyramid interrogation.Learn More
Back-end Web Development
Server-side implementations will be required to interrogate and perform image analysis of whole-slides (Node.js, TensorFlow for AI). This computationally expensive routines should seamlessly produce statistical analysis of mineral-element association recognition.Learn More
Fast and accurate SEM-EDX spectral processing
Proprietary software uses an iterative process for calculating the pixel composition of maps. This is extremely computationally expensive and operates as a "black box". We propose a workaround that will have to be tested.Learn More