A Deep Learning Algorithm for Radiographic Measurements of the Hip in Adults-A Reliability and Agreement Study

Automated weight-bearing foot measurements using an artificial intelligence-based software

OBJECTIVE

To assess the accuracy of an artificial intelligence (AI) software (BoneMetrics, Gleamer) in performing automated measurements on weight-bearing forefoot and lateral foot radiographs.

METHODS

Consecutive forefoot and lateral foot radiographs were retrospectively collected from three imaging institutions. Two senior musculoskeletal radiologists independently annotated key points to measure the hallux valgus, first-second metatarsal, and first-fifth metatarsal angles on forefoot radiographs and the talus-first metatarsal, medial arch, and calcaneus inclination angles on lateral foot radiographs. The ground truth was defined as the mean of their measurements. Statistical analysis included mean absolute error (MAE), bias assessed with Bland-Altman analysis between the ground truth and AI prediction, and intraclass coefficient (ICC) between the manual ratings.

RESULTS

Eighty forefoot radiographs were included (53 ± 17 years, 50 women), and 26 were excluded. Ninety-seven lateral foot radiographs were included (51 ± 20 years, 46 women), and 21 were excluded. MAE for the hallux valgus, first-second metatarsal, and first-fifth metatarsal angles on forefoot radiographs were respectively 1.2° (95% CI [1; 1.4], bias =  - 0.04°, ICC = 0.98), 0.7° (95% CI [0.6; 0.9], bias =  - 0.19°, ICC = 0.91) and 0.9° (95% CI [0.7; 1.1], bias = 0.44°, ICC = 0.96). MAE for the talus-first, medial arch, and calcaneal inclination angles on the lateral foot radiographs were respectively 3.9° (95% CI [3.4; 4.5], bias = 0.61° ICC = 0.88), 1.5° (95% CI [1.2; 1.8], bias =  - 0.18°, ICC = 0.95) and 1° (95% CI [0.8; 1.2], bias = 0.74°, ICC = 0.99). Bias and MAE between the ground truth and the AI prediction were low across all measurements. ICC between the two manual ratings was excellent, except for the talus-first metatarsal angle.

CONCLUSION

AI demonstrated potential for accurate and automated measurements on weight-bearing forefoot and lateral foot radiographs.

A Nordic survey on artificial intelligence in the radiography profession - Is the profession ready for a culture change?

INTRODUCTION

The impact of artificial intelligence (AI) on the radiography profession remains uncertain. Although AI has been increasingly used in clinical radiography, the perspectives of the radiography professionals in Nordic countries have yet to be examined. The primary aim was to examine views of Nordic radiographers 'on AI, with focus on perspectives, engagement, and knowledge of AI.

METHODS

Radiographers from Denmark, Norway, Sweden, Iceland, Greenland, and the Faroe Island were invited through social media platforms to participate in an online survey from March to June 2023. The survey encompassed 29-items and included 4 sections a) demographics, b) barriers and enablers on AI, c) perspectives and experiences of AI and d) knowledge of AI in radiography. Edgars Schein's model of organizational culture was employed to analyse Nordic radiographers' perspectives on AI.

RESULTS

Overall, a total of 421 respondents participated in the survey. A majority were positive/somewhat positive towards AI in radiography e.g., 77.9 % (n = 342) thought that AI would have a positive effect on the profession, and 26% thought that AI would reduce the administrative workload. Most radiographers agreed or strongly agreed that clinicians may have access to AI generated reports (76.8 %, n = 297). Nevertheless, a total of 86 (20.1%) agree or somewhat agreed that AI a potential risk for radiography.

CONCLUSION

Nordic radiographers are generally positive towards AI, yet uncertainties regarding its implementation persist. The findings underscore the importance of understanding these challenges for the responsible integration of AI systems. Carefully weighing the expected influence of AI against key incentives will support a seamless integration of AI for the benefit not just of the patients, but also of the radiography profession.

IMPLICATIONS FOR PRACTICE

Understanding incentives factors and barriers can help address uncertainties during implementation of AI in clinical practice.

Nordic radiographers' and students' perspectives on artificial intelligence - A cross-sectional online survey

INTRODUCTION

The integration of artificial intelligence (AI) into the domain of radiography holds substantial potential in various aspects including workflow efficiency, image processing, patient positioning, and quality assurance. The successful implementation of AI within a Radiology department necessitates the participation of key stakeholders, particularly radiographers. The study aimed to provide a comprehensive investigation about Nordic radiographers' perspectives and attitudes towards AI in radiography.

METHODS

An online 29-item survey was distributed via social media platforms to Nordic students and radiographers working in Denmark, Norway, Sweden, Iceland, Greenland, and the Faroe Islands including items on demographics, specialization, educational background, place of work and perspectives and knowledge on AI. The items were a mix of closed-type and scaled questions, with the option for free-text responses when relevant.

RESULTS

The survey received responses from all Nordic countries with 586 respondents, 26.8% males, 72.1% females, and 1.1% non-binary/self-defined or preferred not to say. The mean age was 37.2 with a standard deviation (SD) of ±12.1 years, and the mean number of years since qualification was 14.2 SD ± 10.3 years. A total of 43% (n = 254) of the respondents had not received any AI training in clinical practice. Whereas 13% (n = 76) had received AI during radiography undergrad training. A total of 77.9% (n = 412) expressed interest in pursuing AI education. The majority of respondents were aware of the potential use of AI (n = 485, 82.8%) and 39.1% (n = 204) had no reservations about AI.

CONCLUSION

Overall, this study found that Nordic radiographers have a positive attitude toward AI. Very limited training or education has been provided to the radiographers. Especially since 82.8% reports on plans to implement AI in clinical practice. In general, awareness of AI applications is high, but the educational level is low for Nordic radiographers.

IMPLICATION FOR PRACTICE

This study emphasises the favourable view of AI held by students and Nordic radiographers. However, there is a need for continuous professional development to facilitate the implementation and effective utilization of AI tools within the field of radiography.

Application of deep learning for automated diagnosis and classification of hip dysplasia on plain radiographs

BACKGROUND

Hip dysplasia is a condition where the acetabulum is too shallow to support the femoral head and is commonly considered a risk factor for hip osteoarthritis. The objective of this study was to develop a deep learning model to diagnose hip dysplasia from plain radiographs and classify dysplastic hips based on their severity.

METHODS

We collected pelvic radiographs of 571 patients from two single-center cohorts and one multicenter cohort. The radiographs were split in half to create hip radiographs (n = 1022). One orthopaedic surgeon and one resident assessed the radiographs for hip dysplasia on either side. We used the center edge (CE) angle as the primary diagnostic criteria. Hips with a CE angle < 20°, 20° to 25°, and > 25° were labeled as dysplastic, borderline, and normal, respectively. The dysplastic hips were also classified with both Crowe and Hartofilakidis classification of dysplasia. The dataset was divided into train, validation, and test subsets using 80:10:10 split-ratio that were used to train two deep learning models to classify images into normal, borderline and (1) Crowe grade 1-4 or (2) Hartofilakidis grade 1-3. A pre-trained on Imagenet VGG16 convolutional neural network (CNN) was utilized by performing layer-wise fine-turning.

RESULTS

Both models struggled with distinguishing between normal and borderline hips. However, achieved high accuracy (Model 1: 92.2% and Model 2: 83.3%) in distinguishing between normal/borderline vs. dysplastic hips. The overall accuracy of Model 1 was 68% and for Model 2 73.5%. Most misclassifications for the Crowe and Hartofilakidis classifications were +/- 1 class from the correct class.

CONCLUSIONS

This pilot study shows promising results that a deep learning model distinguish between normal and dysplastic hips with high accuracy. Future research and external validation are warranted regarding the ability of deep learning models to perform complex tasks such as identifying and classifying disorders using plain radiographs.

LEVEL OF EVIDENCE

Diagnostic level IV.

Plain radiographic indices are reliable indicators for quantitative bone mineral density in male and female patients before total hip arthroplasty

Osteoporosis is underdiagnosed in patients undergoing total hip arthroplasty (THA). Bone mineral density measurement by dual-energy X-ray absorptiometry (DXA) is the gold standard, but indices on plain hip radiographs also seemed to be reliable screening tools in female or Asian ethnicities in previous studies. Given the lack of knowledge about male patients and Caucasian ethnicities, this study was conducted to evaluate plane hip radiographic indices as a screening tool for osteopenia and osteoporosis in Caucasian female and also male patients before undergoing THA. A retrospective analysis of 216 elderly patients with pre-existing DXA before hip arthroplasty was performed and four indices were calculated on plain hip radiographs: Canal-Flare-Index (CFI), Canal-Calcar-Ratio (CCR), Canal-Bone-Ratio (CBR) 7 and 10 cm below the lesser trochanter. They were correlated with femoral neck DXA T-scores by Pearson's correlation and intraclass correlation coefficient, and a ROC analysis was performed. A total of 216 patients (49.5% male) were included. CBR-7 and -10 were highly correlated (p < 0.001) with femoral neck T-score in males (Pearson's correlation CBR-7 r = - 0.60, CBR-10 r = - 0.55) and females (r = - 0.74, r = - 0.77). CBR-7 and -10 also showed good diagnostic accuracy for osteoporosis in the ROC analysis in males (CBR-7: AUC = 0.75, threshold = 0.51; CBR-10: 0.63; 0.50) and females (CBR-7: AUC = 0.87, threshold = 0.55; CBR-10: 0.90; 0.54). Indices such as the Canal Bone Ratio (CBR) 7 or 10 cm below the lesser trochanter on plain hip radiographs are a good screening tool for osteopenia and osteoporosis on plain hip radiographs and can be used to initiate further diagnostics like the gold standard DXA. They differ between male and female patients.

Patient positioning during the radiographic procedure affects the radiological signs of acetabular retroversion - A systematic review

Young adults presenting with non-traumatic hip pain may suffer from acetabular retroversion (AR). The previous studies have suggested that patient positioning during the radiographic procedure, that is, pelvic tilt and/or rotation may alter the appearance of the acetabulum. The purpose of this systematic review was to explore and collate existing literature on the correlation between pelvic positioning in weight-bearing anterior-posterior radiographs and the radiographic signs of AR, namely, the ischial spine sign (ISS) the cross-over sign (COS) and posterior wall sign (PWS). The preferred reporting items for systematic reviews and meta-analysis guidelines were followed. MEDLINE, EMBASE, PubMed, The Cochrane Library, and CINAHL were searched. The search string included the following keywords: Pelvic, tilt, rotation, positioning, inclination, incidence, AR, ISS, COS, PWS, and acetabular version. Two authors independently screened the studies identified in the search, extracted data, and critically assessed included studies for quality using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. In total, 2289 publications were screened. Fifteen articles were found eligible for full-text screening, and four articles met the inclusion criteria. Although the studies varied methodologically, all reported that pelvic positioning impacted radiographic signs of AR investigated. One study suggested that more than 9° of pelvic inclination would result in positive COS. No other benchmark values on the degree of pelvic tilt and rotation that would compromise the diagnosis of AR, that is, the detection of ISS, COS, and PWS were reported. At present, literature reporting on the correlation between patient positioning and AR is sparse. Four studies met the inclusion criteria, and they all reported a link between pelvic positioning and the radiographic appearance of AR.

A retrospective evaluation of supine pelvic radiography image quality using centring points and anatomical axial rotation, including reliability of measurements (ARLEX-P STUDY)

INTRODUCTION

Pelvic radiographs are commonly used for the investigation of a variety of conditions. Comparison between examinations requires a consistent radiographic technique but variations in image quality and radiographic centring points are frequently reported in the literature. The aim of this study was to establish the amount of variation in the radiographic centring point (RCP) and pelvic axial rotation (PAR), with a secondary aim of reporting the reliability of such measures.

METHODS

Using a previously acquired imaging archive, 633 adult pelvis/hip radiographs were identified on a Picture Archiving and Communication System (PACS). Radiographs with bilateral prostheses, evidence of acute pelvic trauma, projections acquired on a stretcher/trolley and those demonstrating large discontinuity between the detector and X-ray field centre were excluded. To determine centring point variation (+ values denote superior variations) and axial rotation multiple measurements were obtained from each radiograph. A video was used to train five observers and each of these reviewed ten random cases to determine inter- and intra-rater reliability. One of the five observers then performed the measurements on all remaining radiographs.

RESULTS

Following exclusions 380 radiographs were evaluated. The median (IQR) RCP deviation from the inter-acetabular line was +22 (+2 to +43) mm where both iliac crests were present and -29 (-45 to -12) mm where they were not. Eleven (3%) cases demonstrate RCP variation from the midline of greater than 25 mm (no bias towards the left or right side). The median (IQR) PAR was 0.0 (-1.5 to 1.4) degrees with greater variance in PAR for male participants (p = 0.004). Almost 60% of inter-rater ICC measurements were categorised as excellent, good or moderate.

CONCLUSION

Variations in RCP and PAR exist when evaluating a sample of routinely acquired pelvis radiographs. Some initial factors, such as sex and sub-examination type (full pelvis [XPEL] or low centred pelvis [XHIPB]) have been identified as having a statistical affect on variability. Further research and methods to standardise radiographic techniques is required and must be multidimensional in nature.

IMPLICATIONS FOR PRACTICE

Selection of radiographic technique, including RCP, appears to influence components of the pelvis radiograph. Given the increasing clinical requirements for pelvic radiography further standardisation alongside individual optimisation is warranted.

Deep Learning for Fully Automated Radiographic Measurements of the Pelvis and Hip

The morphometry of the hip and pelvis can be evaluated in native radiographs. Artificial-intelligence-assisted analyses provide objective, accurate, and reproducible results. This study investigates the performance of an artificial intelligence (AI)-based software using deep learning algorithms to measure radiological parameters that identify femoroacetabular impingement and hip dysplasia. Sixty-two radiographs (124 hips) were manually evaluated by three observers and fully automated analyses were performed by an AI-driven software (HIPPO™, ImageBiopsy Lab, Vienna, Austria). We compared the performance of the three human readers with the HIPPO™ using a Bayesian mixed model. For this purpose, we used the absolute deviation from the median ratings of all readers and HIPPO™. Our results indicate a high probability that the AI-driven software ranks better than at least one manual reader for the majority of outcome measures. Hence, fully automated analyses could provide reproducible results and facilitate identifying radiographic signs of hip disorders.