At the National Institutes of Health (NIH), scientists are making significant strides in the application of artificial intelligence (AI) to enhance the quality and speed of eye imaging. This is a major step forward in the diagnosis and treatment of eye diseases such as age-related macular degeneration (AMD) and other retinal conditions.
Dr. Johnny Tam, who heads the Clinical and Translational Imaging Section at NIH’s National Eye Institute, is at the forefront of this technological innovation. He and his team are enhancing an existing imaging technology known as adaptive optics (AO) using AI. The AO-based imaging equipment, similar to ultrasound, is non-invasive, quick, painless, and a standard feature in many eye clinics.
However, the application of AO in imaging Retinal Pigment Epithelium (RPE) cells presents a unique challenge, a phenomenon referred to as speckle. Speckle can obscure parts of the image, similar to how clouds can interfere with aerial photography. To counter this issue, researchers continually image cells over a prolonged period, allowing different parts of the cells to become visible as the speckle shifts. The resulting images are then meticulously assembled to present a speckle-free image of the RPE cells.
To streamline this process, Dr. Tam’s team developed an AI-based method known as Parallel Discriminator Generative Adverbial Network (P-GAN). This deep learning algorithm was trained using around 6,000 manually analyzed AO-OCT images of human RPE cells, each paired with its speckled original, allowing it to identify and restore obscured cellular features.
The results were impressive. The newly trained P-GAN was able to successfully de-speckle RPE images, revealing cellular details. It could achieve in a single image capture what previously required the acquisition and averaging of 120 images. The P-GAN also improved the image contrast by about 3.5 times and reduced the imaging acquisition and processing time by approximately 100-fold.
Dr. Tam compares the addition of AO to OCT-based imaging to moving from a balcony seat to a front row seat to image the retina. It allows for the revelation of 3D retinal structures at cellular-scale resolution, making it easier to identify early signs of disease.
The focus of Dr. Tam’s recent work is the retinal pigment epithelium (RPE), a layer of tissue behind the retina that supports metabolically active retinal neurons, including the photoreceptors. The RPE is a subject of interest because many retina diseases occur when it starts to degrade.
By integrating AI with AO-OCT, Dr. Tam is confident that one of the major hurdles in routine clinical imaging using AO-OCT, especially for diseases that affect the RPE, has been surmounted. He sees the potential of AI to revolutionize how images are captured and believes that their P-GAN AI will make AO imaging more accessible for routine clinical applications and for studies aimed at understanding the structure, function, and pathophysiology of blinding retinal diseases. This represents an exciting paradigm shift in the field of AI application in imaging.
Comments are closed for this post.