Although physical and ultrasound (US)-based screening for congenital deformities of the hip (developmental dysplasia of the hip, or DDH) is routinely performed in most countries, one of the most commonly performed manoeuvres done under ultrasound observation - dynamic assessment - has been shown to be relatively unreliable and is associated with significant misdiagnosis rates, on the order of 29%.

Our overall research objective is to develop a quantitative method of assessing hip instability, which we hope will standardise diagnosis across different raters and health-centres, and may perhaps improve reliability of diagnosis. To quantify dynamic assessment, we propose to use the variability in femoral head coverage (FHC) measurements within multiple US scans collected during a dynamic assessment. In every US scan, we use our recently-developed automatic FHC measuring tool which leverages phase symmetry features to approximate vertical cortex of ilium and a random forest classifier to identify approximate location of the femoral head. Having estimated FHC in each scan, we estimate the change in FHC across all the US scans during a dynamic assessment and compare this change with variability of FHC found in previous studies.

Our findings - in a dynamic assessment on an infant done by an orthopaedic surgeon, the femoral centre moved by up to 19% of its diameter during distraction, from 55% FHC to 74% FHC. This variability is similar to the variability of FHC in static US scans reported in previous studies, so the variability in FHC readings we found are not indicative of any subluxation or dislocation of the infant's femoral head. Our clinician's qualitative assessment concluded the hip to be normal and not indicative of instability. This suggests that our technique likely has sufficient resolution and repeatability to quantify differences in laxity between stable and unstable hips, although this presumption will have to be confirmed in a subsequent study with additional subjects. The long-term significance of this approach to evaluating dynamic assessments may lie in increasing early diagnostic sensitivity in order to prevent dysplasia remaining undetected prior to manifesting itself in early adulthood joint disease.

Ultrasound (US) is the standard imaging modality used to screen for developmental dysplasia of the hip in infants. Currently, radiologists or orthopaedic surgeons review scan images and judge them to be adequate or inadequate for interpretation. If considered adequate, diagnostic dysplasia metrics are determined; however, there is no standardised method for this process. There is significant inter-observer variability in this manual process which can affect misdiagnosis rates. To eliminate this subjectivity, we developed an automatic method to identify adequate US images and extract dysplasia metrics. The purpose of this study was to validate the efficacy of this automatic method by comparing results with observer-determined dysplasia metrics on a set of US images.

A total of 693 US images from scans of 35 infants were analysed. Trained clinicians at a single institution labeled each image as adequate or inadequate, and subsequently measured alpha and beta angles on adequate images to diagnose dysplasia. We trained our image classifier on random sets of 415 images and used it to assess the remaining 278 images. Alpha and beta angles were automatically estimated on all adequate images. We compared the manual and automatic methods for discrepancies in adequacy determination, metric variability and incidences of missed early diagnosis or over-treatment.

There was excellent agreement between the automatic and manual methods in image adequacy classification (Kappa coefficient = 0.912). On each adequate US image, alpha and beta angle measurements were compared, producing mixed levels of agreement between methods. Mean discrepancies of 1.78°±4.72° and 8.91°±6.437° were seen for alpha and beta angles, respectively. Standard deviations of the angle measures across multiple images from a single patient scan were significantly reduced by the automatic method for both alpha (p<0.05) and beta (p<0.01) angles. Additionally, the automatic method classified three hips (two patients) as Graf type II and two hips (two patients) as type III, while the manual method classified them as type I and II, respectively. Both cases flagged as type III patients by the automatic system subsequently failed Pavlik harness treatment and were booked for surgery.

The automatic method produced excellent agreement with radiologists in scan adequacy classification and significantly reduced measurement variability. Good agreement between methods was found in Graf classification. In instances of disagreement, subsequent clinical findings seemed to support the classification of the automatic method. This proposed method presents an alternative automatic, near-real-time analysis for US images that may potentially significantly improve dysplasia metric reliability and reduce missed early diagnoses without increasing over-treatment.

Ultrasound (US) imaging is recommended for early detection of Developmental Dysplasia of the Hip (DDH) to guide decisions about possible surgical treatment. However, a number of studies have raised concerns over the efficacy of US in early diagnosis. The main limitation of US-based diagnosis is sub-standard reliability of the primary dysplasia metric measurements: namely, the alpha and beta angles. In this study, we have proposed a novel and automatic method to extract dysplasia metrics from 2D US, which we hope will improve the overall reliability of US-based DDH measurements by removing error due to subjective measurements. We hypothesise that improvements in reliability of dysplasia metric measurements will reduce the chances of missed early-diagnosis, and therefore reduce the need for later complex surgical treatments.

We evaluated performance of the algorithm on 4 infants diagnosed with US scans for DDH. The typical runtime of our algorithm is less than 1 second for an US image. We found a 6° bias between manual and automatic measurements, with automatic measurements tending to be lower in value; the standard deviation in the discrepancy values was also relatively high at 7°. This suggests that there is considerable variability in the angle estimation process, which is typically done manually, which supports our contention that further work needs to be done to establish an accurate and repeatable measurement technique. Further, we found agreements in the Graf-classification types in six out of seven sessions. For the one patient where there was a discrepancy in classification, later US sessions suggest the manual technique possibly missed the opportunity for early detection, in contrast to the automatic method which classified the patient as having evidence of dysplasia. Thus, such an automatic method may improve the reliability of current US-based DDH diagnosis techniques. The primary limitation of this study is that we have done strictly an intra-image discrepancy analysis and have not compared the results with what could be considered a ‘gold standard’ reference. In future work, we plan to assess these indices on 3D images of the hip and assess the accuracy of proposed 2D and 3D-based automatic index calculation techniques against a 3D reference model.

Background

Previously, we demonstrated the effectiveness of phase symmetry (PS) features for segmentation and localisation of bone fractures in 3D ultrasound for the purpose of orthopedic fracture reduction surgery. We recently proposed a novel real-time image-processing method of bone surface extraction from local phase features of clinical 3D B-mode ultrasound data. We are presenting a computational study and outline planned future developments for integration into a computer aided orthopedic surgery framework.

Previously, we demonstrated the effectiveness of phase symmetry (PS) features for segmentation and localisation of bone fractures in 3D ultrasound for the purpose of orthopaedic fracture reduction surgery. We recently proposed a novel real-time image-processing method of bone surface extraction from local phase features of clinical 3D B-mode ultrasound data. We are presenting a computational study and outline of planned future developments for integration into a computer aided orthopaedic surgery framework.

Our image-processing pipeline was implemented on three platforms: (1) using an existing PS extraction C++ algorithm on a dual processor machine with two Xeon x5472 CPUs @ 3GHz with 8GB of RAM, (2) using our proposed method implemented in MATLAB running on the same machine as in (1), and (3) CUDA implementation of our method implemented on a professional GPU (Nvidia Tesla c2050).

We ran these three implementations 20 times each on 128×128×128 scans of the iliac crest in live subjects and repeated the processing for 15 combinations of filter parameters. On average, the C++ implementation took 1.93s per volume, the MATLAB implementation 1.28s, and the GPU implementation 0.08s. Overall, our GPU implementation is between 15 and 25 times faster than the state-of-the-art method.

Implementing our algorithm on a professional grade GPU produced dramatic computational improvements, enabling full 3D datasets to be processed in an average time of under 100ms, which, if proven in a clinical system, would allow for near real time computation. We are currently implementing our algorithm on an open research sonography platform (Ultrasonix Medical Corporation). High-powered graphic cards can easily be integrated into the open architecture of this system, thus enabling GPU computation on diagnostic medical and research ultrasound devices.

We intend to use this platform within a surgical environment for accurate and automatic detection of fractures and as an integral part of our developing computer aided surgery pipeline, in which we use PS features to register intra-operative ultrasound to pre-operative computed tomography images.

Due to its ease of use, portability, low cost, real-time response and absence of ionising radiation, ultrasound (US) imaging could potentially be an important tool for non-invasive diagnostic imaging in orthopaedics. Unfortunately, nonlinear characteristics of ultrasound, low signal-to-noise ratio and speckle make it difficult to accurately and reliably determine the location and shape of the bone surface. Recently, local phase-based image processing methods, named phase symmetry (PS), have been shown to perform very well at locating bone surfaces in ultrasound images, with reported accuracies of better than 0.4mm. The local phase features are extracted by filtering the B-mode US image in the frequency domain with a Log-Gabor filter. Although successful results were achieved, accurate localization is highly affected by the choice of filter parameters. Recently, our group proposed a method of automatically selecting the scale, bandwidth and orientation parameters of Log-Gabor filters. Previously, we showed our first clinical results using local phase information to identify distal radius fractures from B-mode US images using automatically selected filter parameters.

The objective of the current study was to determine if the proposed automatic parameter selection method could produce accurate pelvic bone surface shapes in a live clinical setting.

CT scans were obtained as part of normal clinical care from ten patients admitted to Vancouver General Hospital for pelvic fractures. A ‘gold standard’ bone surface was computed from the CT scan. After obtaining informed consent, we performed an additional US scan using a commercially-available real-time scanner (Voluson 730, GE Healthcare, Waukesha, WI) with a 3D US transducer. The PS bone surfaces were extracted from the US scans using the empirical Log-Gabor filter parameters and optimised Log-Gabor filter parameters. The bone surfaces on CT were extracted using a standard thresholding approach that minimises the intra-class variance. The US images were then registered to the CT images using a feature-based rigid registration algorithm with manual landmarking. The quality of the resulting surface matching was evaluated by computing the root mean square distance between the two surface representations.

The average fiducial registration error was 0.31mm (SD 0.25mm). The average surface fitting error (SFE) was 0.72mm (SD 1.24 mm) for PS surfaces extracted using empirical filter parameters and 0.41mm (SD 0.44 mm) using the optimized filter parameters.

In this study, we have demonstrated that our automatic filter parameter selection process can be applied successfully to a bone surface extraction task on 3D US images acquired under clinically realistic conditions. The accuracy of the resulting bone surface is excellent, with an average discrepancy relative to a CT standard of well under a millimeter. This level of accuracy is likely to be sufficiently good for a number of important surgical tasks, including CT to US registration.

Purpose

Radiographs are the most common imaging modality used to guide orthopaedic interventions. Ultrasound (US) imaging offers potential advantages for intraoperative imaging by its portability and ability to produce real-time 2D or 3D images without radiation to either the patient or surgical team. Our objective in this study was to determine in a live emergency room setting, if a newly-developed image processing method for 3D US would allow us to accurately extract (reproduce) the surfaces of fractured bones.