13 February 2018 in Journal of Medical Imaging
The Journal of Medical Imaging covers fundamental and translational research, as well as applications, focused on medical imaging, which continue to yield physical and biomedical advancements in the early detection, diagnostics, and therapy of disease as well as in the understanding of normal.
18 February 2020 in Journal of Medical Imaging
<!-- *** Custom HTML *** --> Purpose: Data-intensive modeling could provide insight on the broad variability in outcomes in spine surgery. Previous studies were limited to analysis of demographic and clinical characteristics. We report an analytic framework called “SpineCloud” that incorporates quantitative features extracted from perioperative images to predict spine surgery outcome. Approach: A retrospective study was conducted in which patient demographics, imaging, and outcome data were collected. Image features were automatically computed from perioperative CT. Postoperative 3- and 12-month functional and pain outcomes were analyzed in terms of improvement relative to the preoperative state. A boosted decision tree classifier was trained to predict outcome using demographic and image features as predictor variables. Predictions were computed based on SpineCloud and conventional demographic models, and features associated with poor outcome were identified from weighting terms evident in the boosted tree. Results: Neither approach was predictive of 3- or 12-month outcomes based on preoperative data alone in the current, preliminary study. However, SpineCloud predictions incorporating image features obtained during and immediately following surgery (i.e., intraoperative and immediate postoperative images) exhibited significant improvement in area under the receiver operating characteristic (AUC): AUC = 0.72 (CI95 = 0.59 to 0.83) at 3 months and AUC = 0.69 (CI95 = 0.55 to 0.82) at 12 months. Conclusions: Predictive modeling of lumbar spine surgery outcomes was improved by incorporation of image-based features compared to analysis based on conventional demographic data. The SpineCloud framework could improve understanding of factors underlying outcome variability and warrants further investigation and validation in a larger patient cohort.
9 February 2018 in Journal of Medical Imaging
Dynamic reference frames (DRFs) are a common component of modern surgical tracking systems; however, the limited number of commercially available DRFs poses a constraint in developing systems, especially for research and education. This work presents the design and validation of a large, open-source library of DRFs compatible with passive, single-face tracking systems, such as Polaris stereoscopic infrared trackers (NDI, Waterloo, Ontario). An algorithm was developed to create new DRF designs consistent with intra- and intertool design constraints and convert to computer-aided design (CAD) files suitable for three-dimensional printing. A library of 10 such groups, each with 6 to 10 DRFs, was produced and tracking performance was validated in comparison to a standard commercially available reference, including pivot calibration, fiducial registration error (FRE), and target registration error (TRE). Pivot tests showed calibration error , indistinguishable from the reference. FRE was , and TRE in a CT head phantom was , both equivalent to the reference. The library of DRFs offers a useful resource for surgical navigation research and could be extended to other tracking systems and alternative design constraints.
10 May 2017 in IEEE Transactions on Medical Imaging
In magnetic resonance (MR)-targeted, 3-D transrectal ultrasound (TRUS)-guided biopsy, prostate motion during the procedure increases the needle targeting error and limits the ability to accurately sample MR-suspicious tumor volumes. The robustness of the 2-D-3-D registration methods for prostate motion compensation is impacted by local optima in the search space. In this paper, we analyzed the prostate motion characteristics and investigated methods to incorporate such knowledge into the registration optimization framework to improve robustness against local optima. Rigid motion of the prostate was analyzed adopting a mixture-of-Gaussian (MoG) model using 3-D TRUS images acquired at bilateral sextant probe positions with a mechanically assisted biopsy system. The learned motion characteristics were incorporated into Powell's direction set method by devising multiple initial search positions and initial search directions. Experiments were performed on data sets acquired during clinical biopsy procedures, and registration error was evaluated using target registration error (TRE) and converged image similarity metric values after optimization. After incorporating the learned initialization positions and directions in Powell's method, 2-D-3-D registration to compensate for motion during prostate biopsy was performed with rms ± std TRE of 2.33 ± 1.09 mm with ~3 s mean execution time per registration. This was an improvement over 3.12 ± 1.70 mm observed in Powell's standard approach. For the data acquired under clinical protocols, the converged image similarity metric value improved in ≥8% of the registrations whereas it degraded only ≤1% of the registrations. The reported improvements in optimization indicate useful advancements in robustness to ensure smooth clinical integration of a registration solution for motion compensation that facilitates accurate sampling of the smallest clinically significant tumors.
29 October 2018 in Physics in Medicine & Biology
Real-time fusion of magnetic resonance (MR) and ultrasound (US) images could facilitate safe and accurate needle placement in spinal interventions. We develop an entirely image-based registration method (independent of or complementary to surgical trackers) that includes an efficient US probe pose initialization algorithm. The registration enables the simultaneous display of 2D ultrasound image slices relative to 3D pre-procedure MR images for navigation. A dictionary-based 3D–2D pose initialization algorithm was developed in which likely probe positions are predefined in a dictionary with feature encoding by Haar wavelet filters. Feature vectors representing the 2D US image are computed by scaling and translating multiple Haar basis filters to capture scale, location, and relative intensity patterns of distinct anatomical features. Following pose initialization, fast 3D–2D registration was performed by optimizing normalized cross-correlation between intra- and pre-procedure images using Powell’s method. Experiments were performed using a lumbar puncture phantom and a fresh cadaver specimen presenting realistic image quality in spinal US imaging. Accuracy was quantified by comparing registration transforms to ground truth motion imparted by a computer-controlled motion system and calculating target registration error (TRE) in anatomical landmarks. Initialization using a 315-length feature vector yielded median translation accuracy of 2.7 mm (3.4 mm interquartile range, IQR) in the phantom and 2.1 mm (2.5 mm IQR) in the cadaver. By comparison, storing the entire image set in the dictionary and optimizing correlation yielded a comparable median accuracy of 2.1 mm (2.8 mm IQR) in the phantom and 2.9 mm (3.5 mm IQR) in the cadaver. However, the dictionary-based method reduced memory requirements by 47× compared to storing the entire image set. The overall 3D error after registration measured using 3D landmarks was 3.2 mm (1.8 mm IQR) mm in the phantom and 3.0 mm (2.3 mm IQR) mm in the cadaver. The system was implemented in a 3D Slicer interface to facilitate translation to clinical studies. Haar feature based initialization provided accuracy and robustness at a level that was sufficient for real-time registration using an entirely image-based method for ultrasound navigation. Such an approach could improve the accuracy and safety of spinal interventions in broad utilization, since it is entirely software-based and can operate free from the cost and workflow requirements of surgical trackers.
17 December 2019 in Journal of Medical Imaging
Convolutional neural networks (CNNs) offer a promising means to achieve fast deformable image registration with accuracy comparable to conventional, physics-based methods. A persistent question with CNN methods, however, is whether they will be able to generalize to data outside of the training set. We investigated this question of mismatch between train and test data with respect to first- and second-order image statistics (e.g., spatial resolution, image noise, and power spectrum). A UNet-based architecture was built and trained on simulated CT images for various conditions of image noise (dose), spatial resolution, and deformation magnitude. Target registration error was measured as a function of the difference in statistical properties between the test and training data. Generally, registration error is minimized when the training data exactly match the statistics of the test data; however, networks trained with data exhibiting a diversity in statistical characteristics generalized well across the range of statistical conditions considered. Furthermore, networks trained on simulated image content with first- and second-order statistics selected to match that of real anatomical data were shown to provide reasonable registration performance on real anatomical content, offering potential new means for data augmentation. Characterizing the behavior of a CNN in the presence of statistical mismatch is an important step in understanding how these networks behave when deployed on new, unobserved data. Such characterization can inform decisions on whether retraining is necessary and can guide the data collection and/or augmentation process for training.