Hip xray normal vs fracture4/1/2024 ![]() ![]() In contrast, the company Imagen Technologies’ OsteoDetect DL system reported improving humans from an unaided AUC 0.84 to AUC 0.89 (improvement of 0.05, 95% CI 0.02–0.08), according to a letter from the FDA ( ). 14 These academic DL reports compared isolated image model performance against humans, but none tested whether algorithms could aid human diagnosis. developed the only previously reported hip fracture detector using DL their model achieved an area under the receiver-operating curve (AUC) of 0.994. 10 Convolutional Neural Networks (CNNs), the DL models best suited for image recognition, have recently been used to detect fracture in the appendicular skeleton, including wrists, 11 shoulders, 12 and hands and feet. performed a clinical trial to study how algorithms can augment radiologists and found radiologists were significantly better at detecting vertebral fractures when aided by an ML model that had a standalone sensitivity of 81%. Most studies detect fracture in algorithm-only systems. Past studies used machine learning (ML) to identify combinations of hand-engineered features associated with fracture, and more recent studies used deep learning (DL) to discover hierarchical pixel patterns from many images with a known diagnosis. Statistical learning models can both detect fractures and help radiologists detect fractures. 3 Fractures are the most commonly missed diagnosis on radiographs of the spine and extremities, and the majority of these errors are perceptual (i.e., a radiologist not noticing some abnormality as opposed to misinterpreting a recognized anomaly). 4, 5 If a patient with high clinical suspicion of fracture has a negative or indeterminant radiograph, then it is usually appropriate to follow-up with a pelvic MRI. 3 However, not all fractures are detectable on radiographs. 2 When a middle-aged or elderly patient presents with acute hip pain and fracture is suspected, clinical guidelines recommend first ordering a hip radiograph. 1 The chance of death in the 3 months following a hip fracture increases by fivefold for women and eightfold for men, relative to age- and sex-matched controls. Similar content being viewed by othersĪn estimated, 1.3 million hip fractures occur annually and are associated with 740,000 deaths, and 1.75 million disability-adjusted life years. Further research is needed to illuminate deep-learning decision processes so that computers and clinicians can effectively cooperate. 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. 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. 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. 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). 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). In this study, we trained deep-learning models on 17,587 radiographs to classify fracture, 5 patient traits, and 14 hospital process variables. 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. Hip fractures are also the most commonly missed diagnosis on pelvic radiographs, and delayed diagnosis leads to higher cost and worse outcomes. Hip fractures are a leading cause of death and disability among older adults. ![]()
0 Comments
Leave a Reply.AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |