Integration of Specim SisuROCK and machine learning advances mineral exploration workflows
The Helmholtz Institute Freiberg for Resource Technology (HIF) is currently utilising Specim hyperspectral imaging and the SisuROCK workstation to significantly accelerate geological decision-making processes. In minerals exploration the ability to determine drilling directions or blasting requirements requires feedback on the mineralogy and for this to occur quickly for efficient operations. Any delay in receiving mineralogical data can result in substantial financial costs and lost opportunities. Consequently, the development of reliable and rapid mineral characterisation is a priority for both mining research and for broader industry operations.
The Exploration Technology division at HIF, led by Dr Richard Gloaguen, focuses on creating technologies that provide faster and more objective geological exploration. The primary objective is to reduce the time spent on traditional mineral analysis of drill cores, which has historically been a bottleneck in the mining lifecycle.
Image 1. Real-time hyperspectral drill core scanning with the SisuROCK workstation at Helmholtz Institute Freiberg, enabling rapid mineral mapping for exploration decision-making. Image courtesy of Helmholtz Institute Freiberg.
By implementing the SisuROCK drill core hyperspectral scanning workstation along with machine learning and data fusion techniques the HIF team is able to generate mineralogical maps and quantitative information in real time. Dr Gloaguen emphasises that if drilling is underway, results are required quickly rather than weeks later. The current system delivers mineralogical information fast enough to directly support active decision-making while drilling.
Image 2. Dr. Richard Gloaguen, Head of Exploration Technology at the Helmholtz Institute Freiberg for Resource Technology, inspecting drill cores during hyperspectral scanning with the SisuROCK workstation. Image courtesy of Helmholtz Institute Freiberg.
The limitations of traditional drill core analysis
Traditional methods of drill core analysis are heavily dependent on manual geological logging, laboratory measurements, and expert interpretation. While these methods are well established, they present several operational challenges. Turnaround times for mineralogical results can take weeks or months, and the results themselves are often subjective as they depend on the expertise of individual geologists. Furthermore, manual logging is difficult to scale across large core archives, and the resulting datasets for mineralogy, petrophysics, and structure often remain disconnected.
These constraints limit how effectively drill core data can guide exploration. The HIF approach seeks to move away from these fragmented workflows toward a more integrated and automated system.
Image 3. Drill cores prepared for hyperspectral scanning as part of a high-throughput mineral mapping workflow. Image courtesy of Helmholtz Institute Freiberg.
Hyperspectral scanning and real-time processing
The HIF has implemented a comprehensive scanning setup based on Specim technology. The system integrates the SisuROCK system with its hyperspectral and RGB cameras, precisely co-registered spatially, which allows for the simultaneous capture of spectral, spatial, and structural information along the entire length of a drill core. The setup covers the spectral range from visible to short-wave infrared (VNIR–SWIR), and RGB enabling detailed mineral characterisation across a wide variety of geological materials.
Image 4. The SisuROCK hyperspectral drill core scanning station installed in a containerised laboratory, enabling rapid and consistent mineral analysis directly at exploration sites. Image courtesy of Helmholtz Institute Freiberg.
SisuROCK delivers very high throughput which is especially suited drilling operations or large mineral core archive logging and can scan over 2,000 metres of drill core in an eight-hour shift. To improve the accuracy of these mineral maps, the team integrates hyperspectral data with information from other sensors including the RGB camera in SisuROCK but also with XRF measurements, and high-resolution SEM-MLA mineralogical analysis. This multi-sensor data fusion is particularly useful for detecting low-abundance minerals and fine-scale features like alteration halos, which are critical for exploration targeting.
Machine learning and automated output
Handling the large volumes of data generated by continuous hyperspectral scanning requires advanced machine learning methods. The team employs generative AI-based clustering and dictionary learning to enable the automated interpretation of complex mineral assemblages.
A major milestone for the institute is the development of a real-time data processing and visualisation workflow. Using custom routines and open-source tools like the HyWiz platform, mineralogical results can be inspected almost immediately after a scan is completed. Instead of providing complex spectral data that requires specialist interpretation, the system outputs ready-to-use mineral maps. Dr Gloaguen notes that most mining companies need these maps to make informed decisions in the field without requiring dedicated hyperspectral specialists on-site.
Image 5. Hyperspectral mineral mapping of drill cores. Image courtesy of Helmholtz Institute Freiberg.
Future applications and benefits
The HIF aims to eventually integrate this hyperspectral and multi-sensor data into interactive 3D geological models. By identifying large-scale alteration patterns, geologists can better understand subsurface processes and vector toward new ore deposits. This workflow also ensures that data from both new and legacy drill cores in national archives is captured consistently and remains accessible for future research.
The benefits of this technological implementation include:
-
Delivery of real-time mineral maps instead of lengthy analysis cycles.
-
Objective and repeatable results that reduce dependence on subjective logging.
-
Improved decision-making during active drilling operations.
-
Lower analysis costs by reducing the requirement for extensive laboratory work.
The use of Specim’s hyperspectral imaging technology and the SisuROCK workstation provides a practical solution to the time constraints inherent in mineral exploration. This system supports the transition from research innovation to practical application in the mining industry.
This article is based on research and technical developments from Specim, in collaboration with Dr Richard Gloaguen and the Exploration Technology division at the Helmholtz Institute Freiberg for Resource Technology (HIF). Using the SisuROCK workstation, the HIF team demonstrated how hyperspectral imaging and machine learning can be integrated to accelerate mineral exploration. All images and technical data are courtesy of the Helmholtz Institute Freiberg and Specim.
Transform your drill core logging
The Specim SisuROCK delivers high-speed hyperspectral imaging for automated drill core logging and mineralogical analysis. Contact our specialists to discuss your geological mapping or mining exploration requirements.