Health information management Technology Offers

Servei de Gestió de la Innovació posted this:
Licensing Manager at Universitat Politècnica de Catalunya - UPC

We have developed a new algorithm for ordering anterior chamber OCT images in such a way that it is possible to classify them, in a fully unsupervised manner, in meaningful groups according to relevant features. We have tested the algorithm with a large set of images classified by two expert ophthalmologists, and with a larger set of annotated images. We have verified that the separation in the different classes defined by the ophthalmologists (closed, narrow, open, and wide open_ figure 6) is similar when using the manually extracted features, or when using the features that are returned by the unsupervised algorithm (View figures). Therefore, the abstract features generated by the algorithm provide novel tools for assessing OCT images of the anterior chamber. They can be used for direct classification of the images and, furthermore, they can be linked to established quantities used for characterizing diseased eyes (like chamber depth, iris-corneal angle) resulting in an automatic detection system. As the algorithm is fully unsupervised, it can be easily automated and set up in OCT imaging systems to aid technicians and doctors in an early diagnosis. The two main advantages of the algorithm demonstrated here over previous works are that it doesn’t need any ground truth or gold standard for training, and it does not rely on specific landmarks; thus, it can analyze images in which relevant landmarks are not visible or not easy to locate.
Image processing method for glaucoma detection and computer program