Detection, classification, and characterization of proximal humerus fractures on plain radiographs.

Autor:	Spek RWA; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia.; Department of Orthopaedic Surgery, University Medical Center Groningen, and University of Groningen, Groningen, Netherlands.; Department of Orthopaedic Surgery, OLVG, Amsterdam, Netherlands., Smith WJ; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Sverdlov M; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Broos S; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Zhao Y; Australian Institute for Machine Learning, Adelaide, Australia., Liao Z; Australian Institute for Machine Learning, Adelaide, Australia., Verjans JW; Australian Institute for Machine Learning, Adelaide, Australia.; Department of Cardiology, Royal Adelaide Hospital, Adelaide, Australia.; Adelaide Medical School, University of Adelaide, Adelaide, Australia., Prijs J; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia.; Department of Orthopaedic Surgery, University Medical Center Groningen, and University of Groningen, Groningen, Netherlands., To MS; South Australia Medical Imaging, Flinders Medical Centre, Adelaide, Australia., Åberg H; Department of Orthopaedic Surgery, Institution of Surgical Sciences, Uppsala University, Uppsala, Sweden., Chiri W; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., IJpma FFA; Department of Orthopaedic Surgery, University Medical Center Groningen, and University of Groningen, Groningen, Netherlands., Jadav B; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., White J; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Bain GI; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Jutte PC; Department of Orthopaedic Surgery, University Medical Center Groningen, and University of Groningen, Groningen, Netherlands., van den Bekerom MPJ; Shoulder and Elbow Expertise Center, Department of Orthopaedic Surgery, OLVG, Amsterdam, Netherlands.; Department of Human Movement Sciences, Faculty of Behavioral and Movement Sciences, Vrije Universiteit, Amsterdam, Netherlands., Jaarsma RL; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia., Doornberg JN; Department of Orthopaedic Surgery, Flinders Medical Centre, and Flinders University, Adelaide, Australia.; Department of Orthopaedic Surgery, University Medical Center Groningen, and University of Groningen, Groningen, Netherlands., Ashkani S, Assink N, Colaris JW, der Gaast NV, Jayakumar P, Kim LJ, de Klerk HH, Kuipers J, Mallee WH, Meesters AML, Mennes SRJ, Oldhof MGE, Pijpker PAJ, Yiu Lau C, Wijffels MME, Wolf AD
Jazyk:	angličtina
Zdroj:	The bone & joint journal [Bone Joint J] 2024 Nov 01; Vol. 106-B (11), pp. 1348-1360. Date of Electronic Publication: 2024 Nov 01.
DOI:	10.1302/0301-620X.106B11.BJJ-2024-0264.R1
Abstrakt:	Aims: The purpose of this study was to develop a convolutional neural network (CNN) for fracture detection, classification, and identification of greater tuberosity displacement ≥ 1 cm, neck-shaft angle (NSA) ≤ 100°, shaft translation, and articular fracture involvement, on plain radiographs. Methods: The CNN was trained and tested on radiographs sourced from 11 hospitals in Australia and externally validated on radiographs from the Netherlands. Each radiograph was paired with corresponding CT scans to serve as the reference standard based on dual independent evaluation by trained researchers and attending orthopaedic surgeons. Presence of a fracture, classification (non- to minimally displaced; two-part, multipart, and glenohumeral dislocation), and four characteristics were determined on 2D and 3D CT scans and subsequently allocated to each series of radiographs. Fracture characteristics included greater tuberosity displacement ≥ 1 cm, NSA ≤ 100°, shaft translation (0% to < 75%, 75% to 95%, > 95%), and the extent of articular involvement (0% to < 15%, 15% to 35%, or > 35%). Results: For detection and classification, the algorithm was trained on 1,709 radiographs (n = 803), tested on 567 radiographs (n = 244), and subsequently externally validated on 535 radiographs (n = 227). For characterization, healthy shoulders and glenohumeral dislocation were excluded. The overall accuracy for fracture detection was 94% (area under the receiver operating characteristic curve (AUC) = 0.98) and for classification 78% (AUC 0.68 to 0.93). Accuracy to detect greater tuberosity fracture displacement ≥ 1 cm was 35.0% (AUC 0.57). The CNN did not recognize NSAs ≤ 100° (AUC 0.42), nor fractures with ≥ 75% shaft translation (AUC 0.51 to 0.53), or with ≥ 15% articular involvement (AUC 0.48 to 0.49). For all objectives, the model's performance on the external dataset showed similar accuracy levels. Conclusion: CNNs proficiently rule out proximal humerus fractures on plain radiographs. Despite rigorous training methodology based on CT imaging with multi-rater consensus to serve as the reference standard, artificial intelligence-driven classification is insufficient for clinical implementation. The CNN exhibited poor diagnostic ability to detect greater tuberosity displacement ≥ 1 cm and failed to identify NSAs ≤ 100°, shaft translations, or articular fractures. Competing Interests: R. W. A. Spek received payments of an amount between USD 10,000 and USD 100,000 from the Flinders Foundation (Adelaide, Australia) for the purpose of this study. During the study period, R. W. A. Spek received payments of an amount between USD 10,000 and USD 100,000 from Prins Bernhard Cultuurfonds (Amsterdam, the Netherlands), Stichting Zabawas (The Hague, The Netherlands), and with an amount of less than USD 10,000 from Michael van Vloten Foundation (Rotterdam, The Netherlands) and Anna Fonds NOREF (Mijdrecht, the Netherlands), all of which were unrelated to this specific study. B. Jadav provided paid consultations for Johnson & Johnson, unrelated to the current study. G. I. Bain received a Flinders Foundation grant for this study, paid to Flinders University, as well as royalties or licenses from Fusetec, stock or stock options in Fusetec, and speaker payments or honoraria from Depuy Synthes and Medartis, none of which are related to this study. G. I. Bain also holds fiduciary roles in the Australian Hand Surgery Society, the Shoulder and Elbow Society of Australia, the Asia Pacific Wrist Association, the International Federation for the Societies for Surgery of the Hand, and the Journal of Wrist Surgery. J. L. Jaarsma is an unpaid executive of the Australian Orthopaedic Association. B. Jadav received a one-off consultation payment and payment for teaching courses from Johnson & Johnson, unrelated to this study. M. P. J. van den Bekerom receives fellowship support from Smith & Nephew, which contributed to this study. (© 2024 The British Editorial Society of Bone & Joint Surgery.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu