Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem
The application of three-dimensional (3D) imaging techniques, such as X-ray tomography and focussed ion beam scanning electron microscopy (FIB-SEM), is increasingly widespread in microstructural analysis of natural materials. However, our ability to collect high-resolution tomographic datasets, each...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-07-01
|
| Series: | Next Materials |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2949822825001819 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850182221261963264 |
|---|---|
| author | Po-Yen Tung Richard J. Harrison |
| author_facet | Po-Yen Tung Richard J. Harrison |
| author_sort | Po-Yen Tung |
| collection | DOAJ |
| description | The application of three-dimensional (3D) imaging techniques, such as X-ray tomography and focussed ion beam scanning electron microscopy (FIB-SEM), is increasingly widespread in microstructural analysis of natural materials. However, our ability to collect high-resolution tomographic datasets, each comprising thousands of two-dimensional (2D) images with millions of pixels, far outstrips our ability to analyse them. Pixel-level segmentation of each 2D image is the first step in any analysis pipeline, but creates a considerable human bottleneck in the workflow that can now be overcome using machine learning. Although advanced pre-trained models such as the Segment Anything Model (SAM) have emerged, conventional segmentation workflows for 3D tomographic data remain limited in comparison. To tackle this, we propose a machine learning workflow that combines SAM with a few-shot learning framework, automating segmentation and minimising user bias. Using SAM, we generate precise annotations from a limited subset of 2D images through basic input prompts, such as points and boxes. These annotations serve as the training data for the few-shot learning model. We benchmark this workflow using a complex 3D FIB-SEM tomographic dataset of the C2 ungrouped carbonaceous chondrite WIS91600. With only 0.6 % of the training data, our method achieves an intersection over union (IoU) score of 80.62 % compared to the ground truth, significantly outperforming widely used methods that achieve a maximum IoU score of 67.07 %. The strong performance on the challenging meteorite dataset highlights its potential for broader application across materials and imaging modalities. |
| format | Article |
| id | doaj-art-7efd6e9c128f4c68af5696f49eeb8a88 |
| institution | OA Journals |
| issn | 2949-8228 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Next Materials |
| spelling | doaj-art-7efd6e9c128f4c68af5696f49eeb8a882025-08-20T02:17:40ZengElsevierNext Materials2949-82282025-07-01810066310.1016/j.nxmate.2025.100663Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/ChemPo-Yen Tung0Richard J. Harrison1Deparrment of Earth Sciences, University of Cambridge, Cambridge, UK; Deparrment of Materials Science and Metallurgy, University of Cambridge, Cambridge, UK; Corresponding author at: Deparrment of Earth Sciences, University of Cambridge, Cambridge, UKDeparrment of Earth Sciences, University of Cambridge, Cambridge, UKThe application of three-dimensional (3D) imaging techniques, such as X-ray tomography and focussed ion beam scanning electron microscopy (FIB-SEM), is increasingly widespread in microstructural analysis of natural materials. However, our ability to collect high-resolution tomographic datasets, each comprising thousands of two-dimensional (2D) images with millions of pixels, far outstrips our ability to analyse them. Pixel-level segmentation of each 2D image is the first step in any analysis pipeline, but creates a considerable human bottleneck in the workflow that can now be overcome using machine learning. Although advanced pre-trained models such as the Segment Anything Model (SAM) have emerged, conventional segmentation workflows for 3D tomographic data remain limited in comparison. To tackle this, we propose a machine learning workflow that combines SAM with a few-shot learning framework, automating segmentation and minimising user bias. Using SAM, we generate precise annotations from a limited subset of 2D images through basic input prompts, such as points and boxes. These annotations serve as the training data for the few-shot learning model. We benchmark this workflow using a complex 3D FIB-SEM tomographic dataset of the C2 ungrouped carbonaceous chondrite WIS91600. With only 0.6 % of the training data, our method achieves an intersection over union (IoU) score of 80.62 % compared to the ground truth, significantly outperforming widely used methods that achieve a maximum IoU score of 67.07 %. The strong performance on the challenging meteorite dataset highlights its potential for broader application across materials and imaging modalities.http://www.sciencedirect.com/science/article/pii/S2949822825001819Few-shot learningSegment Anything Model (SAM)3D tomographic datasetsMicrostructure segmentationMachine learning in materials scienceHigh-resolution imaging analysis |
| spellingShingle | Po-Yen Tung Richard J. Harrison Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem Next Materials Few-shot learning Segment Anything Model (SAM) 3D tomographic datasets Microstructure segmentation Machine learning in materials science High-resolution imaging analysis |
| title | Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem |
| title_full | Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem |
| title_fullStr | Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem |
| title_full_unstemmed | Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem |
| title_short | Efficient microstructure segmentation in three-dimensional imaging: Combining few-shot learning with the segment anything modelEarth/Chem |
| title_sort | efficient microstructure segmentation in three dimensional imaging combining few shot learning with the segment anything modelearth chem |
| topic | Few-shot learning Segment Anything Model (SAM) 3D tomographic datasets Microstructure segmentation Machine learning in materials science High-resolution imaging analysis |
| url | http://www.sciencedirect.com/science/article/pii/S2949822825001819 |
| work_keys_str_mv | AT poyentung efficientmicrostructuresegmentationinthreedimensionalimagingcombiningfewshotlearningwiththesegmentanythingmodelearthchem AT richardjharrison efficientmicrostructuresegmentationinthreedimensionalimagingcombiningfewshotlearningwiththesegmentanythingmodelearthchem |