Metadata-independent classification of MRI sequences using convolutional neural networks: Successful application to prostate MRI.

Autor: Baumgärtner GL; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: georg.baumgaertner@charite.de., Hamm CA; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany; Berlin Institute of Health (BIH), Anna-Louisa-Karsch-Straße 2, 10178 Berlin, Germany. Electronic address: charlie.hamm@charite.de., Schulze-Weddige S; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: sophia.schulze-weddige@charite.de., Ruppel R; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: richard.ruppel@charite.de., Beetz NL; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: nick-lasse.beetz@charite.de., Rudolph M; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: madhuri.rudolph@charite.de., Dräger F; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: franziska.draeger@charite.de., Froböse KP; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: konrad.froboese@charite.de., Posch H; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: helena.posch@charite.de., Lenk J; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany. Electronic address: julian.lenk@charite.de., Biessmann F; Faculty VI - Informatics and Media, Berliner Hochschule für Technik (BHT), Einstein Center Digital Future, 13353 Berlin, Germany. Electronic address: felix.biessmann@bht-berlin.de., Penzkofer T; Department of Radiology, Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Campus Virchow Klinikum, Augustenburgerplatz 1, 13353 Berlin, Germany; Berlin Institute of Health (BIH), Anna-Louisa-Karsch-Straße 2, 10178 Berlin, Germany. Electronic address: tobias.penzkofer@charite.de.
Jazyk: angličtina
Zdroj: European journal of radiology [Eur J Radiol] 2023 Sep; Vol. 166, pp. 110964. Date of Electronic Publication: 2023 Jul 08.
DOI: 10.1016/j.ejrad.2023.110964
Abstrakt: Purpose: The ever-increasing volume of medical imaging data and interest in Big Data research brings challenges to data organization, categorization, and retrieval. Although the radiological value chain is almost entirely digital, data structuring has been widely performed pragmatically, but with insufficient naming and metadata standards for the stringent needs of image analysis. To enable automated data management independent of naming and metadata, this study focused on developing a convolutional neural network (CNN) that classifies medical images based solely on voxel data.
Method: A 3D CNN (3D-ResNet18) was trained using a dataset of 31,602 prostate MRI volumes with 10 different sequence types of 1243 patients. A five-fold cross-validation approach with patient-based splits was chosen for training and testing. Training was repeated with a gradual reduction in training data assessing classification accuracies to determine the minimum training data required for sufficient performance. The trained model and developed method were tested on three external datasets.
Results: The model achieved an overall accuracy of 99.88 % ± 0.13 % in classifying typical prostate MRI sequence types. When being trained with approximately 10 % of the original cohort (112 patients), the CNN still achieved an accuracy of 97.43 % ± 2.10 %. In external testing the model achieved sensitivities of > 90 % for 10/15 tested sequence types.
Conclusions: The herein developed CNN enabled automatic and reliable sequence identification in prostate MRI. Ultimately, such CNN models for voxel-based sequence identification could substantially enhance the management of medical imaging data, improve workflow efficiency and data quality, and allow for robust clinical AI workflows.
Competing Interests: Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: G.L.B. Stock in Siemens. C.A.H. Seed funding from University Medicine Greifswald; payment for lectures from Bracco Imaging Deutschland. F.B. Research project Sondierungsprojekt zu KI in der Pflege (SOKIP), funded by German Federal Ministry of Research and Education; AI Lighthouse Project Green Consumption Assistant with ecosia and TU Berlin, funded by German Federal Ministry for the Environment; Research grant on Prozessentwicklung und -begleitung zum KI-Einsatz in der Pflege (ProKIP), funded by the German Federal Ministry of Research; Research grant on KI in der Pflege-Sturz / Delir / Medikation (KIP-SDM), funded by the German Federal Ministry of Research; Research grant on Reduction of the Impact of untreated Waste Water on the Environment in case of torrential Rain (RIWWER), funded by the German Federal Ministry for Economic Affairs and Climate Action; Research grant on Citizen-based Monitoring for Peace & Security in the Era of Synthetic Media and Deepfakes, funded by the German Foundation for Peace Research. T.P. Berlin Institute of Health (Clinician Scientist Grant, Platform Grant), Ministry of Education and Research (BMBF, 01KX2021, 01KX2121, 68GX21001A), German Research Foundation (DFG, SFB 1340/2), Horizon 2020 (952172); AGO, Aprea AB, ARCAGY-GINECO, Astellas Pharma Global Inc. (APGD), Astra Zeneca, Clovis Oncology, Dohme Corp, Holaira, Incyte Corporation, Karyopharm, Lion Biotechnologies, MedImmune, Merck Sharp, Millennium Pharmaceuticals, Morphotec Inc., NovoCure Ltd., PharmaMar S.A. and PharmaMar USA, Roche, Siemens Healthineers, and TESARO; royalties from Elsevier; patent pending.
(Copyright © 2023 Elsevier B.V. All rights reserved.)
Databáze: MEDLINE
Externí odkaz: