Abstract:Importance: Scaphoid fractures are the most common carpal fracture, but as many as 20% are not visible (ie, occult) in the initial injury radiograph; untreated scaphoid fractures can lead to degenerative wrist arthritis and debilitating pain, detrimentally affecting productivity and quality of life. Occult scaphoid fractures are among the primary causes of scaphoid nonunions, secondary to delayed diagnosis.Objective: To develop and validate a deep convolutional neural network (DCNN) that can reliably detect both apparent and occult scaphoid fractures from radiographic images.Design, Setting, and Participants: This diagnostic study used a radiographic data set compiled for all patients presenting to Chang Gung Memorial Hospital (Taipei, Taiwan) and Michigan Medicine (Ann Arbor) with possible scaphoid fractures between January 2001 and December 2019. This group was randomly split into training, validation, and test data sets. The images were passed through a detection model to crop around the scaphoid and were then used to train a DCNN model based on the EfficientNetB3 architecture to classify apparent and occult scaphoid fractures. Data analysis was conducted from January to October 2020.Exposures: A DCNN trained to discriminate radiographs with normal and fractured scaphoids.Main Outcomes and Measures: Area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity. Fracture localization was assessed using gradient-weighted class activation mapping.Results: Of the 11 838 included radiographs (4917 [41.5%] with scaphoid fracture; 6921 [58.5%] without scaphoid fracture), 8356 (70.6%) were used for training, 1177 (9.9%) for validation, and 2305 (19.5%) for testing. In the testing test, the first DCNN achieved an overall sensitivity and specificity of 87.1% (95% CI, 84.8%-89.2%) and 92.1% (95% CI, 90.6%-93.5%), respectively, with an AUROC of 0.955 in distinguishing scaphoid fractures from scaphoids without fracture. Gradient-weighted class activation mapping closely corresponded to visible fracture sites. The second DCNN achieved an overall sensitivity of 79.0% (95% CI, 70.6%-71.6%) and specificity of 71.6% (95% CI, 69.0%-74.1%) with an AUROC of 0.810 when examining negative cases from the first model. Two-stage examination identified 20 of 22 cases (90.9%) of occult fracture.Conclusions and Relevance: In this study, DCNN models were trained to identify scaphoid fractures. This suggests that such models may be able to assist with radiographic detection of occult scaphoid fractures that are not visible to human observers and to reliably detect fractures of other small bones.

Scaphoid fractures: has magnetic resonance imaging become a cost-effective alternative to definitive diagnosis?

Diagnosis of occult scaphoid fractures: an exploratory study to assess whether clusters of clinical features are diagnostic in comparison to reduced sequence magnetic resonance imaging

The Feasibility of High-Resolution Peripheral Quantitative Computed Tomography (HR-pQCT) in Patients with Suspected Scaphoid Fractures

The current management of scaphoid fractures in the emergency department across an Australian metropolitan public health service: A retrospective cohort study

MRI for Scaphoid Nonunion: Utilization Rates, Factors Associated with Utilization, and Subsequent Vascularized Bone Graft Use

6-week radiographs unsuitable for diagnosis of suspected scaphoid fractures

Arthroscope Assisted Minimally Invasive Management of Scaphoid Fractures with Missed Diagnosis and Delayed Treatment

An Umbrella Review and Updated Meta-Analysis of Imaging Modalities in Occult Scaphoid and Hip and Femoral Fractures

Measurement of Scapholunate Joint Space Width on Real-Time MRI—A Feasibility Study

A comparative study of image quality and diagnostic confidence in diagnosis and follow-up of scaphoid fractures using photon-counting detector CT and energy-integrating detector CT

Reasons for scaphoid non-union: analysis of behavior of health care providers and patients

Development and Validation of a Deep Learning Model Using Convolutional Neural Networks to Identify Scaphoid Fractures in Radiographs

Clinical Validation of an Artificial Intelligence Model for Detecting Distal Radius, Ulnar Styloid, and Scaphoid Fractures on Conventional Wrist Radiographs

Editorial Commentary: Magnetic Resonance Imaging Is Not Inferior to the Gold Standard of Diagnostic Arthroscopy for Identification of Wrist Ligamentous Pathology

Comparison of Operative and Non-Operative Treatment of Acute Undisplaced or Minimally-Displaced Scaphoid Fractures: a Meta-Analysis of Randomized Controlled Trials.

Improving Radiographic Fracture Recognition Performance and Efficiency Using Artificial Intelligence

Magnetic Resonance Imaging Has Utility in Clinical Decision-Making for Children and Adolescents With Symptomatic Subfibular Ossicles

Deciphering acute shoulder trauma with normal initial X-ray: Contributions of ultrasonography and MRI

Diagnostic accuracy of wrist MRI in comparison to wrist arthroscopy regarding TFCC lesions in clinical practice

Assessment of Scapholunate Instability on 4D CT Scans in Patients with Inconclusive Conventional Images

High-resolution MRI (3T-MRI) in diagnosis of wrist pain: is diagnostic arthroscopy still necessary?