Date of case study: 12/2019
Industry of business: Academic organization
LightningChart solution: XY
Efficient applications for visualization and data analysis of oral and dental multispectral images
The University of Eastern Finland is one of the largest universities in Finland. Their research is ranked among the best in the world in several fields. As a multidisciplinary university, UEF trains experts for the various sectors of society. UEF seeks solutions to global challenges through multidisciplinary research and education. In collaboration with their partners, they generate new knowledge, expertise and innovations.
UEF and Arction collaborated in DIGIDENT project, which focused on developing spectral imaging technologies, databases and machine learning for dental applications. The project utilizes spectrally optimized optical imaging and digital spectral image database (Big Data) for more effective detection and visualization of oral and dental diseases. Methods based on artificial intelligence create an opportunity for early detection of illness using secure optical imaging. The aim is to make prototypes for spectral optimized optical imaging of the mouth and teeth, which in the future could be integrated into the dental chairs.
Arction’s LightningChart enabled The University of Eastern Finland to create fast and efficient applications for data visualization and data analysis of oral and dental multispectral images. It was also used to design and create a user-friendly spectral image annotation tool, which was used by dentists for manual annotation of oral and dental spectral images.
“These LightningChart-based applications helped us to visualize our data, to analyze them, and to gather the data necessary for developing machine-learning methods for the detection, classification and segmentation of the signs of oral and dental diseases. The co-operation with the staff of Arction. Ltd. was very fluent.”
– Markku Hauta-Kasari, PhD, Professor, Head of the School of Computing
Below in pictures 1-3 you can see few examples of Multispectral Image Lab in use.
Multispectral Image Lab also contains annotation tools for manual oral and dental image segmentation. The tool contains 40 different classes, varying from technical markings, like specular reflection, shadow, or out-of-focus area, to hard and soft tissue like enamel and gingiva, and diseases like gingivitis, calculus, or caries. Annotation tool allows annotating using polygons and lines. The annotations are saved in a csv-like text file. Each annotation class can be hidden from the view by toggling the checkbox off next to the annotation class selection button. Hiding individual annotation markings is not possible.
The software contains a principal component analysis -mode. This mode computes the PCA eigenvectors of the loaded spectral image cube and projects it onto these new basis vectors. The eigenvectors can be studied using the ”Eigenvector tools”-panel on the right side. In this mode the slider above the image refers to the nth eigenvector used to transform the spectral image cube.
The program also has a histogram adjustment tooling. The “Histogram tools”-section on the right side shows the histogram of the current wavelength band image, simple brightness control, adaptive histogram equalization functionality, and of course reset changes -buttons. Each of the latter three operations can be applied on the selected band image or on the whole image cube.