Dataset Information

Automatic Hip Fracture Identification and Functional Subclassification with Deep Learning.

ABSTRACT:

Purpose

To investigate the feasibility of automatic identification and classification of hip fractures using deep learning, which may improve outcomes by reducing diagnostic errors and decreasing time to operation.

Materials and methods

Hip and pelvic radiographs from 1118 studies were reviewed, and 3026 hips were labeled via bounding boxes and classified as normal, displaced femoral neck fracture, nondisplaced femoral neck fracture, intertrochanteric fracture, previous open reduction and internal fixation, or previous arthroplasty. A deep learning-based object detection model was trained to automate the placement of the bounding boxes. A Densely Connected Convolutional Neural Network (or DenseNet) was trained on a subset of the bounding box images, and its performance was evaluated on a held-out test set and by comparison on a 100-image subset with two groups of human observers: fellowship-trained radiologists and orthopedists; senior residents in emergency medicine, radiology, and orthopedics.

Results

The binary accuracy for detecting a fracture of this model was 93.7% (95% confidence interval [CI]: 90.8%, 96.5%), with a sensitivity of 93.2% (95% CI: 88.9%, 97.1%) and a specificity of 94.2% (95% CI: 89.7%, 98.4%). Multiclass classification accuracy was 90.8% (95% CI: 87.5%, 94.2%). When compared with the accuracy of human observers, the accuracy of the model achieved an expert-level classification, at the very least, under all conditions. Additionally, when the model was used as an aid, human performance improved, with aided resident performance approximating unaided fellowship-trained expert performance in the multiclass classification.

Conclusion

A deep learning model identified and classified hip fractures with expert-level performance, at the very least, and when used as an aid, improved human performance, with aided resident performance approximating that of unaided fellowship-trained attending physicians.Supplemental material is available for this article.© RSNA, 2020.

SUBMITTER: Krogue JD

PROVIDER: S-EPMC8017394 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Hip fractures are a leading cause of death and disability among older adults. Hip fractures are also the most commonly missed diagnosis on pelvic radiographs, and delayed diagnosis leads to higher cost and worse outcomes. Computer-aided diagnosis (CAD) algorithms have shown promise for helping radiologists detect fractures, but the image features underpinning their predictions are notoriously difficult to understand. In this study, we trained deep-learning models on 17,587 radiographs to classify fracture, 5 patient traits, and 14 hospital process variables. All 20 variables could be individually predicted from a radiograph, with the best performances on scanner model (AUC?=?1.00), scanner brand (AUC?=?0.98), and whether the order was marked "priority" (AUC?=?0.79). Fracture was predicted moderately well from the image (AUC?=?0.78) and better when combining image features with patient data (AUC?=?0.86, DeLong paired AUC comparison, p?=?2e-9) or patient data plus hospital process features (AUC?=?0.91, p?=?1e-21). Fracture prediction on a test set that balanced fracture risk across patient variables was significantly lower than a random test set (AUC?=?0.67, DeLong unpaired AUC comparison, p?=?0.003); and on a test set with fracture risk balanced across patient and hospital process variables, the model performed randomly (AUC?=?0.52, 95% CI 0.46-0.58), indicating that these variables were the main source of the model's fracture predictions. A single model that directly combines image features, patient, and hospital process data outperforms a Naive Bayes ensemble of an image-only model prediction, patient, and hospital process data. If CAD algorithms are inexplicably leveraging patient and process variables in their predictions, it is unclear how radiologists should interpret their predictions in the context of other known patient data. Further research is needed to illuminate deep-learning decision processes so that computers and clinicians can effectively cooperate.

Project description:BackgroundMyopia is the leading cause of visual impairment and affects millions of children worldwide. Timely and annual manual optometric screenings of the entire at-risk population improve outcomes, but screening is challenging due to the lack of availability and training of assessors and the economic burden imposed by the screenings. Recently, deep learning and computer vision have shown powerful potential for disease screening. However, these techniques have not been applied to large-scale myopia screening using ocular appearance images.MethodsWe trained a deep learning system (DLS) for myopia detection using 2,350 ocular appearance images (processed by 7,050 pictures) from children aged 6 to 18. Myopia is defined as a spherical equivalent refraction (SER) [the algebraic sum in diopters (D), sphere + 1/2 cylinder] ≤-0.5 diopters. Saliency maps and gradient class activation maps (grad-CAM) were used to highlight the regions recognized by VGG-Face. In a prospective clinical trial, 100 ocular appearance images were used to assess the performance of the DLS.ResultsThe area under the curve (AUC), sensitivity, and specificity of the DLS were 0.9270 (95% CI, 0.8580-0.9610), 81.13% (95% CI, 76.86-5.39%), and 86.42% (95% CI, 82.30-90.54%), respectively. Based on the saliency maps and grad-CAMs, the DLS mainly focused on eyes, especially the temporal sclera, rather than the background or other parts of the face. In the prospective clinical trial, the DLS achieved better diagnostic performance than the ophthalmologists in terms of sensitivity [DLS: 84.00% (95% CI, 73.50-94.50%) versus ophthalmologists: 64.00% (95% CI, 48.00-72.00%)] and specificity [DLS: 74.00% (95% CI, 61.40-86.60%) versus ophthalmologists: 53.33% (95% CI, 30.00-66.00%)]. We also computed AUC subgroups stratified by sex and age. DLS achieved comparable AUCs for children of different sexes and ages.ConclusionsThis study for the first time applied deep learning to myopia screening using ocular images and achieved high screening accuracy, enabling the remote monitoring of the refractive status in children with myopia. The application of our DLS will directly benefit public health and relieve the substantial burden imposed by myopia-associated visual impairment or blindness.

Project description:BackgroundMedication errors account for a large proportion of all medical errors. In most homes, patients take a variety of medications for a long period. However, medication errors frequently occur because patients often throw away the containers of their medications.ObjectiveWe proposed a deep learning-based system for reducing medication errors by accurately identifying prescription pills. Given the pill images, our system located the pills in the respective pill databases in South Korea and the United States.MethodsWe organized the system into a pill recognition step and pill retrieval step, and we applied deep learning models to train not only images of the pill but also imprinted characters. In the pill recognition step, there are 3 modules that recognize the 3 features of pills and their imprints separately and correct the recognized imprint to fit the actual data. We adopted image classification and text detection models for the feature and imprint recognition modules, respectively. In the imprint correction module, we introduced a language model for the first time in the pill identification system and proposed a novel coordinate encoding technique for effective correction in the language model. We identified pills using similarity scores of pill characteristics with those in the database.ResultsWe collected the open pill database from South Korea and the United States in May 2022. We used a total of 24,404 pill images in our experiments. The experimental results show that the predicted top-1 candidates achieve accuracy levels of 85.6% (South Korea) and 74.5% (United States) for the types of pills not trained on 2 different databases (South Korea and the United States). Furthermore, the predicted top-1 candidate accuracy of our system was 78% with consumer-granted images, which was achieved by training only 1 image per pill. The results demonstrate that our system could identify and retrieve new pills without additional model updates. Finally, we confirmed through an ablation study that the language model that we emphasized significantly improves the pill identification ability of the system.ConclusionsOur study proposes the possibility of reducing medical errors by showing that the introduction of artificial intelligence can identify numerous pills with high precision in real time. Our study suggests that the proposed system can reduce patients' misuse of medications and help medical staff focus on higher-level tasks by simplifying time-consuming lower-level tasks such as pill identification.

Project description:Osteoporotic hip fracture is a major public health issue. Estimation of the outcome and maximization of functional recovery after fracture is very important in the treatment of older patients. The purposes of this study were to clarify the functional outcomes after the treatment of hip fracture and to identify the factors that influence functional recovery. In the present study, 228 patients admitted to an acute-care hospital from January 2016 to June 2018 were evaluated. The patients were categorized into a trochanteric fracture group (n = 128) and a neck fracture group (n = 100). We retrospectively reviewed their ambulation ability 6 months after fracture using the Functional Ambulation Category (FAC) score. The other survey items were the presurgical duration, length of hospital stay, time until beginning to walk using parallel bars, complications affecting treatment, and mortality rate. The 6-month follow-up rate was 54.4% (n = 124). The results showed that the patients with trochanteric fracture were significantly older than those with neck fracture (86 vs. 82 years, respectively; p = 0.03). In total, 85.0% of patients with trochanteric fracture and 92.2% of patients with neck fracture were independent ambulators before injury (FAC score of 4 or 5). The FAC score 6 months after fracture was positively correlated with the FAC score before fracture and at discharge (all p<0.001) and negatively correlated with patient age (p<0.001) and presurgical duration for patients with neck fracture (p = 0.04). There was no statistically significant correlation with the length of hospital stay or the time until beginning to walk using parallel bars. In conclusion, patients with trochanteric fractures were older than those with neck fractures. In both fracture types, walking recovery 6 months after hip fracture was related to the FAC score before injury and at discharge from an acute-care hospital but not to the time until beginning to walk using parallel bars.