Artificial intelligence for radiological paediatric fracture assessment: a systematic review

Abstract

Background Majority of research and commercial efforts have focussed on use of artificial intelligence (AI) for fracture detection in adults, despite the greater long-term clinical and medicolegal implications of missed fractures in children. The objective of this study was to assess the available literature regarding diagnostic performance of AI tools for paediatric fracture assessment on imaging, and where available, how this compares with the performance of human readers. Materials and methods MEDLINE, Embase and Cochrane Library databases were queried for studies published between 1 January 2011 and 2021 using terms related to ‘fracture’, ‘artificial intelligence’, ‘imaging’ and ‘children’. Risk of bias was assessed using a modified QUADAS-2 tool. Descriptive statistics for diagnostic accuracies were collated. Results Nine eligible articles from 362 publications were included, with most (8/9) evaluating fracture detection on radiographs, with the elbow being the most common body part. Nearly all articles used data derived from a single institution, and used deep learning methodology with only a few (2/9) performing external validation. Accuracy rates generated by AI ranged from 88.8 to 97.9%. In two of the three articles where AI performance was compared to human readers, sensitivity rates for AI were marginally higher, but this was not statistically significant. Conclusions Wide heterogeneity in the literature with limited information on algorithm performance on external datasets makes it difficult to understand how such tools may generalise to a wider paediatric population. Further research using a multicentric dataset with real-world evaluation would help to better understand the impact of these tools.

Commercially available artificial intelligence tools for fracture detection: the evidence

“…There are understandably significant technical challenges in adapting AI solutions to be effective in paediatrics due to the variability in bone structure, predominantly between the ages of 0-16 years. 5 Development in this more specialized field is also slowed and restricted by legal issues relating to the collection of data for training the algorithms and more complex and difficult procedures for obtaining certification or approval from respective legal bodies. 64 …”

“…In adults, high sensitivity rates of 92% for plain radiographs and 89% for computerised tomography (CT) scans have been reported, with specificities of 91% for plain radiographs and 92% for CT scans 2–4 ; with accuracy rates of between 89% and 98% in children. 5 Despite such promising results, over half of these studies had a high risk of bias, 2 were only used in a research setting, were not ready for clinical deployment, nor externally validated. 6 This is particularly concerning given that 24% of AI solutions in one study yielded a substantial decline in performance when evaluated on external data (compared to their internal data set) and the majority (81% of algorithms) reported negative impact.…”

Artificial intelligence-based detection of paediatric appendicular skeletal fractures: performance and limitations for common fracture types and locations

“…Additionally, smaller patient populations hamper the creation of large databases, which are necessary to train the software sufficiently. Despite growing interest in AI-based software applications for automated paediatric fracture detection, research in this area remains scarce [2]. Most studies have focused on one specific body part and have described the development and evaluation of the software in a singlecentre setting [3][4][5].…”

Elbow trauma in children: development and evaluation of radiological artificial intelligence models

“…However, most of the artificial intelligence (A.I.) models that can detect fractures on X-ray are not validated in children, except recently [14] , [15] , [16] , [17] . Moreover, despite numerous publications, focus is mostly made on the in silico performance of algorithms, but sparsely on the performance resulting from the interaction between humans and A.I.…”