Clinically applicable deep learning for diagnosis and referral in retinal disease
Autor: | Kareem Ayoub, Trevor Back, Rosalind Raine, Alan Karthikesalingam, Pearse A. Keane, Xavier Glorot, Hugh Montgomery, Julien Cornebise, Demis Hassabis, Daniel Visentin, Brendan O'Donoghue, Adnan Tufail, Catherine A Egan, Jeffrey De Fauw, Olaf Ronneberger, Peng T. Khaw, Harry Askham, Reena Chopra, Faith Mackinder, Joseph R. Ledsam, Dominic King, Nenad Tomasev, Julian Hughes, Simon Bouton, Clemens Meyer, George van den Driessche, Cian Hughes, Stanislav Nikolov, Mustafa Suleyman, Balaji Lakshminarayanan, Dawn A Sim, Bernardino Romera-Paredes, Sam Blackwell, Geraint Rees |
---|---|
Rok vydání: | 2018 |
Předmět: |
Male
0301 basic medicine medicine.medical_specialty Referral Computer science Clinical Decision-Making ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Disease Retina General Biochemistry Genetics and Molecular Biology 03 medical and health sciences Deep Learning 0302 clinical medicine Clinical pathway Retinal Diseases Optical coherence tomography medicine Medical imaging Humans Medical physics Set (psychology) Referral and Consultation Aged medicine.diagnostic_test business.industry Deep learning General Medicine Middle Aged Multiple pathologies 030104 developmental biology 030221 ophthalmology & optometry Female Artificial intelligence business Tomography Optical Coherence |
Zdroj: | Nature Medicine. 24:1342-1350 |
ISSN: | 1546-170X 1078-8956 |
Popis: | The volume and complexity of diagnostic imaging is increasing at a pace faster than the availability of human expertise to interpret it. Artificial intelligence has shown great promise in classifying two-dimensional photographs of some common diseases and typically relies on databases of millions of annotated images. Until now, the challenge of reaching the performance of expert clinicians in a real-world clinical pathway with three-dimensional diagnostic scans has remained unsolved. Here, we apply a novel deep learning architecture to a clinically heterogeneous set of three-dimensional optical coherence tomography scans from patients referred to a major eye hospital. We demonstrate performance in making a referral recommendation that reaches or exceeds that of experts on a range of sight-threatening retinal diseases after training on only 14,884 scans. Moreover, we demonstrate that the tissue segmentations produced by our architecture act as a device-independent representation; referral accuracy is maintained when using tissue segmentations from a different type of device. Our work removes previous barriers to wider clinical use without prohibitive training data requirements across multiple pathologies in a real-world setting. |
Databáze: | OpenAIRE |
Externí odkaz: |