A analysis staff from Nationwide Cheng Kung College (NCKU), led by Chih-Chung Hsu, an assistant professor on the Institute of Knowledge Science, NCKU, has demonstrated that machine studying can help within the detection of age-related macular degeneration from fundus photographs.

The analysis was a part of the ADAM problem (Automated Detection problem on Age-related Macular degeneration), which was a satellite tv for pc occasion of the Worldwide Symposium on Biomedical Imaging (ISBI) 2020 convention. The ADAM problem drew participation from 11 groups from all around the world. The main points of the problem and its outcomes have been not too long ago printed within the journal IEEE Transactions on Medical Imaging.

Age-related macular degeneration (AMD) is a number one explanation for imaginative and prescient loss. If left untreated, AMD can result in everlasting imaginative and prescient loss and irreversible injury to the retina. Early detection is crucial for well timed treatement. Fundus images is a robust software that can be utilized to detect the illness, however it’s time-consuming and requires the experience of a medical skilled. That is the place the ADAM problem is available in.

The problem was damaged down into 4 duties addressing the essential points of detecting and characterizing age-related macular degeneration — AMD classification, optic disc detection and segmentation, fovea localization, and lesion detection and segmentation. A dataset of 1200 fundus pictures was made obtainable for the problem. The staff from NCKU used 4 totally different machine studying architectures to handle every job of the ADAM problem.

For the primary job, the staff used an EfficientNet structure for the binary classification of AMD. Then, for the optic disc detection and segmentation job, they used the EfficientNet structure for classification mixed with U-Internet for segmentation with a weight cross-entropy loss perform. For the fovea localization job, the staff used a mix of two U-Internet architectures, a Masks-RCNN and ResNet. And for the final job of lesion detection and segmentation, they used a DeepLab-v3 with Resnet.

Out of the 4 architectures used, the one used for fovea localization significantly stood out. The staff used a multiple-blocks regression technique to divide the fundus picture into a number of blocks, after which every block was one-hot-encoded.

“Early detection of AMD is essential for its treatment. We wanted to achieve a highly accurate symptom localization for fundus image, helping the doctors to have external references for their diagnosis,” Hsu stated concerning the machine studying structure.

The mannequin proposed by the NCKU staff was comparable in efficiency to different state-of-the-art fashions whereas being extraordinarily light-weight. Moreover, the ensemble technique employed improved the efficiency of the mannequin and the usage of prior scientific information helped to attain higher outcomes.

“Given the fundus image, it is possible to diagnose AMD in a real-time sense, which might reduce the effort of doctors. Further, the proposed model can be integrated into hardware to be one of the standard pieces of equipment in any clinic. That would provide us with an AI-aided diagnosis, which could be very helpful for junior doctors or even interns,” Hsu stated.

The staff has made obtainable the video beneath to explain the examine.


What's Your Reaction?

hate hate
confused confused
fail fail
fun fun
geeky geeky
love love
lol lol
omg omg
win win
The Obsessed Guy
Hi, I'm The Obsessed Guy and I am passionate about artificial intelligence. I have spent years studying and working in the field, and I am fascinated by the potential of machine learning, deep learning, and natural language processing. I love exploring how these technologies are being used to solve real-world problems and am always eager to learn more. In my spare time, you can find me tinkering with neural networks and reading about the latest AI research.


Your email address will not be published. Required fields are marked *