Deep Learning Reconstruction in Magnetic Resonance Imaging


Deep Learning Reconstruction (DLR) used in combination with Magnetic Resonance Imaging (MRI) has the potential to help diagnose diseases earlier, faster and better. It improves image quality by eliminating noise, and increasing signal to noise ratio, helping to obtain ultra-high-resolution images. Bordeaux University Hospital, in France, is working with Canon Medical’s DLR solution in numerous applications in research and clinical practice. Through this, it has been able to make observations on anatomy that have previously not been possible on 3T systems. The team has been using the system routinely for about a year now and is working to validate the tool scientifically.

Speeding workflows

In their daily routine, one of the biggest challenges the Hospital faces is the growing number of requests they receive and the difficult task of examining all the patients. They anticipate that DLR will help speed up workflows to enable them to meet this demand. DLR functionality is easily integrated into the image reconstruction chain. Radiologists only have to plug in the option to improve the image quality. Switching to this new routine is effortless and brings real benefits. High resolution images can be achieved with DLR without losing time or signal. There is also a reduction of image acquisition time, as the signal quality matters less, because noise can be eliminated following the scan. DLR allows correction afterwards and signal or spatial resolution no longer requires improvement.

Spectacular resolution

In particular, DLR has an important clinical impact in imaging anatomical regions that require a very high resolution. For example, the spatial resolution achieved with DLR had never been seen before in neuro-logic imaging by Prof. Dousset, Head of the Neuro Radiology Department. In this he refers to the imaging of brain areas, such as the claustrum, which is almost invisible on standard MRI images, even with very high resolution or high field devices.

Whereas a colleague, Prof. Tourdias, has worked with 7T to visualize extremely fine structures of the hippocampus area of the brain, while at Stanford University in the US. With DLR, he can now do this task with a 3T and achieve similar results as with 7T.

Research synergies

The collaboration between the hospital and Canon Medical creates opportunities to find solutions for patients, often in areas that had never been explored before. Working with Canon Medical enables the physicians to work with the most advanced technology on the company’s latest MRI scanner for their ongoing clinical research. And there is an interest in transferring the technology back to Canon Medical, in the hope that research results help to advance industry swiftly.

Please click HERE if you would like to read the complete article.