Median Technologies announces results of deep learning study for enhanced prostate segmentation – helping to advance precision medicine

Sophia Antipolis, France

Press Release – For Immediate Release – 5:45pm CET

Median Technologies, the industry-leading Imaging Phenomics Company® (Paris:ALMDT) announced today results of a deep learning study for enhanced prostate segmentation. With more accurate segmentation, new therapeutic decision-making tools can help advance precision medicine. Study results will be presented at the SPIE Medical Imaging conference in Houston, TX, USA on February 14.

The study, “Assessing the relevance of multi-planar MRI acquisitions for prostate segmentation using deep learning techniques,” relates to prostate cancer detection and presents a deep learning-based method for prostate segmentation using magnetic resonance images acquired in two imaging views. The results were compared to manual tracings performed by medical image analysts and were evaluated by the number of cancerous lesions within the segmentation. The study shows the proposed method performs better than methods relying on a single imaging view. This highlights the possibilities for design of fully automatic, accurate, fast and reproducible prostate segmentation computer systems.

Deep learning, a field of artificial intelligence, enables computer systems to improve analysis by using experience and data. It has very promising applications in healthcare, particularly in the medical imaging field to develop computer-aided systems for the management of diseases such as cancer. A key step in developing computer-aided systems for cancer detection is organ segmentation.

The study will be presented during the poster session, “Machine Learning and Artificial Intelligence,” to be held on Wednesday 14 February, from 5:30p to 7:00p CST.

Please visit SPIE Medical Imaging Conference, for more information on the conference.

Download 20180213_SPIE2018_EN.pdf