Medical & Scientific Research
Customised Vision Applications for Medical & Scientific Research
Our image processing makes it possible to detect objects and substances that are not visible to the naked eye, such as enzymes and bacteria. At Qtechnology, we have worked with scientists and chemists to implement a variety of solutions for science and research projects.
Combining our cameras with a range of optics, spectrographs, microscopes, and other elements, facilitate application development for scientific observations or dedicated inspection tasks.
Typical Applications – Medical & Scientific Research
Inhibition Zones Around Antibiotic Containers
- Light Setting (Multiple Exposures and HDR to Capture Both OCR Capable Characters in Bright Regions Detection of Inhibition Zones in Dark Areas)
- Image Processing
Based on some test imaging, we were requested to develop a common algorithm to determine the inhibition zones around antibiotic containers. Each of the inhibition zones has a specific numbering relating to the type of antibiotic, which is used to find the antibiotics with greatest effect.
However, disturbing elements such as the labels on the Petri dishes and lighting made it difficult to perform well, with common image processing. Inhibition zones needed to be detected from the inner contour, as they can not grow outside the Petri dish, it sometimes gave overlapping zones. The medium in the Petri dish can have different colors and inhibition zones can be almost transparent. At the same time, the number on top of each antibiotic container must be read accurately.
Nevertheless, this requires two or more images with different exposures, to get the best possible detection. In improving the imaging technology and lighting, it is possible to get the code running using Python, enabling us to develop a solution that determines the inhibition zones around antibiotic containers with high accuracy.
Image Flow Cytometry
Combining cell sorting with Imaging Flow Cytometry (IFC) to fully realise its tremendous potential, requires real-time image construction and analysis. However, all IFC systems demonstrated to date perform image analysis offline, and the ability to produce, measure, analyse cell images, and to sort cells in a real-time manner, will be the next major milestone for IFC.
Consequently, we developed a solution that extracts cell characteristics in real-time, including the use of Field-Programmable Gate Arrays (FPGA) or Graphics Processing Unit (GPU) to implement various image processing and machine learning algorithms. In other words, we developed an evaluation platform for online monitoring of bacteria and other contaminants in enzyme production, counting bacteria in a given volume eg. 1 mL.
One of the two cameras used in the process is equipped with a sensor for bright-field imaging and the other camera is equipped with two sensors and blue lasers for fluorescence excitation. The bright-field and fluorescence imaging are both controlled from the cameras and the cameras are synchronised to each other with minimum latency between image capture.
The enzymes are pumped out from the main process pipe to a sample tube, that is passed through the system, where particle sizes and types are determined and counted from the combination of bright-field and fluorescence images.
- Laser Integration
- Light Design (Combination of Laser and Bright Field Light)
- Pump Control (Control of External Sample Pump to Fed System)
- Optical Design (Integration of Light and Splitting Light to Different Camera Sensors)
- Multi-Camera Sensor and Light Synchronisation
Recovering Hidden Medieval Texts
- Hyperspectral Imaging
As parchment was an expensive material in the 16th century, the paper was often reused, especially during the 16th century when Catholic books were discarded in favor of new Protestant. For centuries, it was common for bookbinders to reuse books as bracing in – and cover on – contemporary books as well as scraping the text clean because it was parchment and then write new text on top.
Withal, it has been a challenge for the librarians to decode these hidden texts. Hence, in 2017, our hyperspectral imaging (HSI) cameras were applied in a project with Newtec A/S, attempting to decode a selection of old books containing unreadable texts in their bindings. In the project, a selection of monographs from Herlufsholm Special collection was moved down a conveyor belt underneath HSI cameras and special lighting.
The result was that unreadable passages in the monographs were made considerably more legible, which enabled further analysis of texts in the scholarly community using data mining. The project concluded that the combination of HSI and Data Mining is a powerful tool when it comes to the reading and rapid identification of hidden medieval manuscript fragments.