Accurate prediction of disease-risk factors from volumetric medical scans by a deep vision model pre-trained with 2D scans.

Autor:	Avram O; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA. orenavram@gmail.com.; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA. orenavram@gmail.com.; Department of Anesthesiology and Perioperative Medicine, University of California, Los Angeles, Los Angeles, CA, USA. orenavram@gmail.com., Durmus B; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Rakocz N; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Corradetti G; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA.; Department of Ophthalmology, University of California, Los Angeles, Los Angeles, CA, USA., An U; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Nittala MG; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA.; Department of Ophthalmology, University of California, Los Angeles, Los Angeles, CA, USA., Terway P; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Rudas A; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA., Chen ZJ; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Wakatsuki Y; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Hirabayashi K; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Velaga S; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Tiosano L; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA.; Department of Ophthalmology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel., Corvi F; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Verma A; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA.; Department of Ophthalmology and Visual Sciences, University of Louisville, Louisville, KY, USA., Karamat A; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Lindenberg S; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Oncel D; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Almidani L; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Hull V; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Fasih-Ahmad S; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Esmaeilkhanian H; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA., Cannesson M; Department of Anesthesiology and Perioperative Medicine, University of California, Los Angeles, Los Angeles, CA, USA., Wykoff CC; Retina Consultants of Texas, Retina Consultants of America, Houston, TX, USA.; Blanton Eye Institute, Houston Methodist Hospital, Houston, TX, USA., Rahmani E; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA., Arnold CW; Department of Radiology, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Pathology, University of California, Los Angeles, Los Angeles, CA, USA., Zhou B; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA., Zaitlen N; Department of Neurology, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Human Genetics, University of California, Los Angeles, Los Angeles, CA, USA., Gronau I; School of Computer Science, Reichman University, Herzliya, Israel., Sankararaman S; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Human Genetics, University of California, Los Angeles, Los Angeles, CA, USA., Chiang JN; Department of Computational Medicine, University of California, Los Angeles, Los Angeles, CA, USA.; Department of Neurosurgery, University of California, Los Angeles, Los Angeles, CA, USA., Sadda SR; Doheny Eye Institute, University of California, Los Angeles, Pasadena, CA, USA. ssadda@doheny.org.; Department of Ophthalmology, University of California, Los Angeles, Los Angeles, CA, USA. ssadda@doheny.org., Halperin E; Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA. ehalperin@cs.ucla.edu.
Jazyk:	angličtina
Zdroj:	Nature biomedical engineering [Nat Biomed Eng] 2024 Oct 01. Date of Electronic Publication: 2024 Oct 01.
DOI:	10.1038/s41551-024-01257-9
Abstrakt:	The application of machine learning to tasks involving volumetric biomedical imaging is constrained by the limited availability of annotated datasets of three-dimensional (3D) scans for model training. Here we report a deep-learning model pre-trained on 2D scans (for which annotated data are relatively abundant) that accurately predicts disease-risk factors from 3D medical-scan modalities. The model, which we named SLIViT (for 'slice integration by vision transformer'), preprocesses a given volumetric scan into 2D images, extracts their feature map and integrates it into a single prediction. We evaluated the model in eight different learning tasks, including classification and regression for six datasets involving four volumetric imaging modalities (computed tomography, magnetic resonance imaging, optical coherence tomography and ultrasound). SLIViT consistently outperformed domain-specific state-of-the-art models and was typically as accurate as clinical specialists who had spent considerable time manually annotating the analysed scans. Automating diagnosis tasks involving volumetric scans may save valuable clinician hours, reduce data acquisition costs and duration, and help expedite medical research and clinical applications. (© 2024. The Author(s), under exclusive licence to Springer Nature Limited.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu Full text from SpringerLink