Seeing what ophthalmologists can't | DELL Technologies
CUSTOMER: Voxeleron LLC, a company specialised in the development of Artifical Vision and Machine Learning software.
TECHNOLOGY: AI, CNN, Dell Precision and GPU NVIDIA workstations.
Addressing needs and bringing benefits
At present, incurable Age-related Macular Degeneration (AMD) affects nearly 200 million people in the world. It is the leading cause of blindness in people over 60. The disease has two stages: an early dry stage and a late wet stage of advanced AMD when vision loss and blindness can rapidly occur. Nowadays ophthalmologists use optical coherence tomography to create 3D images of the retina, so as to diagnose and monitor the disease through regular examinations to detect any changes.
The objective is to help ophthalmologists predict the likelihood of a patient progressing from the dry stage to the wet stage. To achieve the goal, 3D datasets need to be processed in a deep learning convolutional neural network model. By applying artificial intelligence models and using Dell Precision 7920 Tower workstations with NVIDIA Quadro GV100 GPU, it was possible to save up to three months in the execution of the models.
Voxeleron is broadening the horizons of ophthalmologic diagnosis with image analysis based on AI models, trained with Dell Precision workstations with GPUs NVIDIA. Voxeleron aims to improve ophthalmologists’ capacity to predict the likelihood of a patient with AMD progressing from the dry stage to the wet stage. 3D retinal images may contain useful clues that can be spotted by artificial intelligence (AI) in the form of a deep learning convolutional neural network (CNN) model.
Furthermore, doctors can potentially use their 3D retinal imaging tools to also diagnose neurological disorders. Although the retina is at the back of the eye, it is really at the front of the brain and is a window to the central nervous system. The way in which AI can be applied is changing, particularly in tasks that are difficult for humans to do, such as finding complex patterns in large datasets.