Sequential U-Net Architecture for Automatic Femoral Artery Segmentation in Ultrasound Images

This work compares three different approaches to automatically segment the femoral artery from 2D ultrasound images. Two of the architectures follow a sequential structure, where each ultrasound image is considered a slice of the whole vessel volume, and its previous segmentation result will be part of the input, thus leading to a spatial prior. The Dice score on test data show a better performance on the baseline U-Net (0.819) compared to the sequential U-Net approaches (0.633, 0.725) for the femoral artery segmentation. This could be attributed to the misalignment of the slices being used in those networks. A possible improvement could be assumed in the implementation of a spatially calibrated and tracked ultrasound probe. Overall, these results indicate promising approaches for an automatic segmentation of the femoral artery using 2D ultrasound data.

Three-Dimensional Registration of Freehand-Tracked Ultrasound to CT Images of the Talocrural Joint

A rigid surface–volume registration scheme is presented in this study to register computed tomography (CT) and free-hand tracked ultrasound (US) images of the talocrural joint. Prior to registration, bone surfaces expected to be visible in US are extracted from the CT volume and bone contours in 2D US data are enhanced based on monogenic signal representation of 2D US images. A 3D monogenic signal data is reconstructed from the 2D data using the position of the US probe recorded with an optical tracking system. When registering the surface extracted from the CT scan to the monogenic signal feature volume, six transformation parameters are estimated so as to optimize the sum of monogenic signal features over the transformed surface. The robustness of the registration algorithm was tested on a dataset collected from 12 cadaveric ankles. The proposed method was used in a clinical case study to investigate the potential of US imaging for pre-operative planning of arthroscopic access to talar (osteo)chondral defects (OCDs). The results suggest that registrations with a registration error of 2 mm and less is achievable, and US has the potential to be used in assessment of an OCD’ arthroscopic accessibility, given the fact that 51% of the talar surface could be visualized.

Portable Tracked Ultrasound System to Monitor Spinal Curvature

ntroduction: Adolescent idiopathic scoliosis (AIS) is the most frequent spinal deformation. It is assessed by frequent x-ray imaging, exposing patients to frequent radiation, increasing the risk of cancer. Tracked ultrasound imaging produce a three-dimensional visual of the spine without risk. We proposed using an optically tracked handheld USB Interson ultrasound probe, which is much less expensive than the current standard ultrasounds. Methods: A practical setup was developed for scoliosis monitoring. Reference markers were on the patient’s shoulder and the wall to account for patient motion, and simulate the plane of the x-ray detector to allow future comparisons. The optical tracker tracked the markers and ultrasound probe. This was compared to the electromagnetically tracked non-portable Ultrasonix ultrasound. Scanning captured the thoracic and lumbar regions. A three-dimensional image was composited by stacking a series of two-dimensional ultrasound slices in their tracked physical positions. To compare the two-ultrasound setups and ensure the ease of identification, a novice attempted to manually identify transverse processes. These were compared to see if all transverse processes scanned could be marked from both ultrasound setups. Two clinical experts then confirmed the markings were anatomically correct. Results and Conclusions: In all scans 100% of the transverse processes scanned (n = 51) were identified in Interson and Ultrasonix images. Curvatures measured using the Ultrasonix method have been previously validated to curvature measurements from x-ray images. Thus the 100% correspondence of the Interson and Ultrasonix setups indicates this inexpensive method is a promising tool to reduce radiation exposure during scoliosis monitoring.

Skull Localization using Ultrasound for Navigated Neurosurgery

Image-guided navigation for neurosurgery requires accurate localization of the skull. Localization can be problematic when the patient is in a facedown position. The posterior skull lacks unique identifiable landmarks, which complicates standard localization methods using a tracked pointer. In addition to the lack of anatomical landmarks, trying to access facial surfaces is error-prone when working under the table and problems arise with line-of-sight of the optical tracker. We proposed the use of ultrasound to perform localization and investigated the accuracy of this process. A simulation study was performed to test the feasibility of ultrasound for localization on a plastic skull. An initial localization, using an optically tracked pointer, was performed to partially align pre-operative images and the skull model. Skull surface points were localized by optically tracked ultrasound and used in a surface registration algorithm. Accuracy and reproducibility was then investigated. Evaluation of the proposed localization method found that the average distance of points off the skull surface was 0.6 ± 0.1mm, which meets the same standards set by current commercially available systems for face-up positions. Using tracked ultrasound for registration is feasible for patients in facedown position. We provided a non-invasive method of registration that could be accomplished using one optical tracking camera, and maintains a constant line-of-sight. This project was performed in cooperation with Dr. Gernot Kronreif and the Austrian Center for Medical Innovation and Technology. Dr. Kronreif and his staff are preparing for a clinical test of this localization process.

Improved Temporal Calibration of Tracked Ultrasound: An Open-Source Solution

In tracked ultrasound systems, temporal misalignment between image and tracker data results in incorrect image pose. We present a fully automatic temporal calibration. We image a flat plate in water with a tracked probe undergoing periodic uniaxial freehand translation. Using robust line detection scheme, we compute temporal misalignment as difference between probe and corresponding image position. From 240 sequences, standard deviation was under 5 ms for standard imaging parameters. Source code is available in Public Library for Ultrasound Research, PLUS ( www.plustoolkit.org ).