What is this talk about?

NOUS, is the on-site deep-learning system that is being trained to detect the difference between healthy and abnormal cilia in cases suspected of PCD using electron microscopy (EM) images. Different teaching approaches are being tested that include (i) Detection of cilia that are usable for diagnosis (ii) Classification of normal and abnormal cilia, (iii) second classification of abnormal cilia in classes of abnormalities. Over 20,000 EM images of cilia have been input to NOUS and it is showing consistently improved accuracy as the training datasets enlarge. When testing it against diagnostic specialists (n=5) using blinded image datasets, an agreement of >75% in the classification of images was found. This is similar to the agreement measured between individual diagnostic specialists. Furthermore, it is currently being trialled alongside the current diagnostic protocol in order to validate the system for clinical use.

About Dr Andreia Pinto:

She started as a biomedical scientist in histocellular pathology but early refined her area of interest in EM. Dr Pinto worked for more than ten years in Lisbon, solo managing an electron microscopy department, taking in fundamental to translational research. Since 2014, she has been working on PCD, in 2019 she joined the Royal Brompton as a thoracic research associate. Dr Pinto is currently teaching a deep machine learning platform to recognize patterns in EM images of cilia with the aim to roll it out as a tool (PCD-AID) in diagnosis. She has also been involved in investigating new insights of Sars-CoV2 infection of the respiratory airway.

Registration:

Click here.

The meeting link will be sent out 24 hours before the event!