Analyzing the reduction quality of the distal radioulnar joint after closed K-wire transfixation in a cadaver model: is supination or neutral position superior?

INTRODUCTION: Distal radioulnar joint (DRUJ) instabilities are challenging and their optimal treatment is controversial. In special cases or when reconstruction of the stabilizing triangular fibrocartilage complex (TFCC) fails, K-wire transfixation can be performed. However, no consensus has been reached regarding the rotational position of the forearm in which this should be done. Therefore, it was investigated whether anatomical reduction would best be achieved by transfixation in neutral position or supination of the forearm. MATERIALS AND METHODS: Twelve cadaveric upper limbs were examined before dissection of the DRUJ stabilizing ligaments and after closed transfixation in both positions by C-arm cone-beam CT. Whether this was first done in neutral position or in supination was randomized. The change in the radioulnar ratio (RR) in percentage points (%points) was analyzed using Student's t-test. RR was used since it is a common and sensitive method to evaluate DRUJ reduction, expressing the ulnar head's position in the sigmoid notch as a length ratio. RESULTS: The analysis showed an increased change in RR in neutral position with 5.4 ± 9.7%points compared to fixation in supination with 0.2 ± 16.1%points, yet this was not statistically significant (p = 0.404). CONCLUSIONS: Neither position leads to a superior reduction in general. However, the result was slightly closer to the anatomical position in supination. Thus, transfixation of the DRUJ should be performed in the position in which reduction could best be achieved and based on these data, that tends to be in supination. Further studies are necessary to validate these findings and to identify influential factors.

Automatic plane adjustment of orthopedic intraoperative flat panel detector CT-volumes.

Purpose: To assess the result in orthopedic trauma surgery, usually three-dimensional volume data of the treated region is acquired. With mobile C-arm systems, these acquisitions can be performed intraoperatively, reducing the number of required revision surgeries. However, the acquired volumes are typically not aligned to the anatomical regions. Thus, the multiplanar reconstructed (MPR) planes need to be adjusted manually during the review of the volume. To speed up and ease the workflow, an automatic parameterization of these planes is needed. Approach: We present a detailed study of multitask learning (MTL) regression networks to estimate the parameters of the MPR planes. First, various mathematical descriptions for rotation, including Euler angle, quaternion, and matrix representation, are revised. Then, two different MTL network architectures based on the PoseNet are compared with a single task learning network. Results: Using a matrix description rather than the Euler angle description, the accuracy of the regressed normals improves from 7.7 deg to 7.3 deg in the mean value for single anatomies. The multihead approach improves the regression of the plane position from 7.4 to 6.1 mm, whereas the orientation does not benefit from this approach. Thus, the achieved accuracy meets the reported interrater variance in similarly complex body regions of up to 6.3 deg for the normals and up to 9.3 mm for the plane position. Conclusions: The use of a multihead approach with shared features leads to more accurate plane regression compared with the use of individual networks for each task. It also improves the angle estimation for the ankle region. The reported results are in the same range as manual plane adjustments. The use of a combined network with shared parameters requires less memory, which is a great benefit for the implementation of an application for the surgical environment.

Multi-Stage Platform for (Semi-)Automatic Planning in Reconstructive Orthopedic Surgery.

Intricate lesions of the musculoskeletal system require reconstructive orthopedic surgery to restore the correct biomechanics. Careful pre-operative planning of the surgical steps on 2D image data is an essential tool to increase the precision and safety of these operations. However, the plan's effectiveness in the intra-operative workflow is challenged by unpredictable patient and device positioning and complex registration protocols. Here, we develop and analyze a multi-stage algorithm that combines deep learning-based anatomical feature detection and geometric post-processing to enable accurate pre- and intra-operative surgery planning on 2D X-ray images. The algorithm allows granular control over each element of the planning geometry, enabling real-time adjustments directly in the operating room (OR). In the method evaluation of three ligament reconstruction tasks effect on the knee joint, we found high spatial precision in drilling point localization (ε<2.9mm) and low angulation errors for k-wire instrumentation (ε<0.75∘) on 38 diagnostic radiographs. Comparable precision was demonstrated in 15 complex intra-operative trauma cases suffering from strong implant overlap and multi-anatomy exposure. Furthermore, we found that the diverse feature detection tasks can be efficiently solved with a multi-task network topology, improving precision over the single-task case. Our platform will help overcome the limitations of current clinical practice and foster surgical plan generation and adjustment directly in the OR, ultimately motivating the development of novel 2D planning guidelines.