Influence of learned landmark correspondences on lung CT registration.

BACKGROUND: Disease or injury may cause a change in the biomechanical properties of the lungs, which can alter lung function. Image registration can be used to measure lung ventilation and quantify volume change, which can be a useful diagnostic aid. However, lung registration is a challenging problem because of the variation in deformation along the lungs, sliding motion of the lungs along the ribs, and change in density. PURPOSE: Landmark correspondences have been used to make deformable image registration robust to large displacements. METHODS: To tackle the challenging task of intra-patient lung computed tomography (CT) registration, we extend the landmark correspondence prediction model deep convolutional neural network-Match by introducing a soft mask loss term to encourage landmark correspondences in specific regions and avoid the use of a mask during inference. To produce realistic deformations to train the landmark correspondence model, we use data-driven synthetic transformations. We study the influence of these learned landmark correspondences on lung CT registration by integrating them into intensity-based registration as a distance-based penalty. RESULTS: Our results on the public thoracic CT dataset COPDgene show that using learned landmark correspondences as a soft constraint can reduce median registration error from approximately 5.46 to 4.08 mm compared to standard intensity-based registration, in the absence of lung masks. CONCLUSIONS: We show that using landmark correspondences results in minor improvements in local alignment, while significantly improving global alignment.

Reconstruction and completion of high-resolution 3D cardiac shapes using anisotropic CMRI segmentations and continuous implicit neural representations.

Since the onset of computer-aided diagnosis in medical imaging, voxel-based segmentation has emerged as the primary methodology for automatic analysis of left ventricle (LV) function and morphology in cardiac magnetic resonance images (CMRI). In standard clinical practice, simultaneous multi-slice 2D cine short-axis MR imaging is performed under multiple breath-holds resulting in highly anisotropic 3D images. Furthermore, sparse-view CMRI often lacks whole heart coverage caused by large slice thickness and often suffers from inter-slice misalignment induced by respiratory motion. Therefore, these volumes only provide limited information about the true 3D cardiac anatomy which may hamper highly accurate assessment of functional and anatomical abnormalities. To address this, we propose a method that learns a continuous implicit function representing 3D LV shapes by training an auto-decoder. For training, high-resolution segmentations from cardiac CT angiography are used. The ability of our approach to reconstruct and complete high-resolution shapes from manually or automatically obtained sparse-view cardiac shape information is evaluated by using paired high- and low-resolution CMRI LV segmentations. The results show that the reconstructed LV shapes have an unconstrained subvoxel resolution and appear smooth and plausible in through-plane direction. Furthermore, Bland-Altman analysis reveals that reconstructed high-resolution ventricle volumes are closer to the corresponding reference volumes than reference low-resolution volumes with bias of [limits of agreement] -3.51 [-18.87, 11.85] mL, and 12.96 [-10.01, 35.92] mL respectively. Finally, the results demonstrate that the proposed approach allows recovering missing shape information and can indirectly correct for limited motion-induced artifacts.

Impact of bone and cartilage segmentation from CT and MRI on both bone forearm osteotomy planning.

INTRODUCTION: The use of MRI scans for pre-operative surgical planning of forearm osteotomies provides additional information of joint cartilage and soft tissue structures and reduces radiation exposure in comparison with the use of CT scans. In this study, we investigated whether using 3D information obtained from MRI with and without cartilage information leads to a different outcome of pre-operative planning. METHODS: Bilateral CT and MRI scans of the forearms of 10 adolescent and young adult patients with a unilateral bone deformation were acquired in a prospective study. The bones were segmented from CT and MRI, and cartilage only from MRI. The deformed bones were virtually reconstructed, by registering the joint ends to the healthy contralateral side. An optimal osteotomy plane was determined that minimized the distance between the resulting fragments. This process was performed in threefold: using the CT and MRI bone segmentations, and the MRI cartilage segmentations. RESULTS: Comparison of bone segmentation from MRI and CT scan resulted in a 0.95 ± 0.02 Dice Similarity Coefficient and 0.42 ± 0.07 mm Mean Absolute Surface Distance. All realignment parameters showed excellent reliability across the different segmentations. However, the mean differences in translational realignment between CT and MRI bone segmentations (4.5 ± 2.1 mm) and between MRI bone and MRI bone and cartilage segmentations (2.8 ± 2.1 mm) were shown to be clinically and statistically significant. A significant positive correlation was found between the translational realignment and the relative amount of cartilage. CONCLUSION: This study indicates that although bone realignment remained largely similar when using MRI with and without cartilage information compared to using CT, the small differences in segmentation could induce statistically and clinically significant differences in the osteotomy planning. We also showed that endochondral cartilage might be a non-negligible factor when planning osteotomies for young patients.

Deep Learning-Based Segmentation of Locally Advanced Breast Cancer on MRI in Relation to Residual Cancer Burden: A Multi-Institutional Cohort Study.

BACKGROUND: While several methods have been proposed for automated assessment of breast-cancer response to neoadjuvant chemotherapy on breast MRI, limited information is available about their performance across multiple institutions. PURPOSE: To assess the value and robustness of deep learning-derived volumes of locally advanced breast cancer (LABC) on MRI to infer the presence of residual disease after neoadjuvant chemotherapy. STUDY TYPE: Retrospective. SUBJECTS: Training cohort: 102 consecutive female patients with LABC scheduled for neoadjuvant chemotherapy (NAC) from a single institution (age: 25-73 years). Independent testing cohort: 55 consecutive female patients with LABC from four institutions (age: 25-72 years). FIELD STRENGTH/SEQUENCE: Training cohort: single vendor 1.5 T or 3.0 T. Testing cohort: multivendor 3.0 T. Gradient echo dynamic contrast-enhanced sequences. ASSESSMENT: A convolutional neural network (nnU-Net) was trained to segment LABC. Based on resulting tumor volumes, an extremely randomized tree model was trained to assess residual cancer burden (RCB)-0/I vs. RCB-II/III. An independent model was developed using functional tumor volume (FTV). Models were tested on an independent testing cohort and response assessment performance and robustness across multiple institutions were assessed. STATISTICAL TESTS: The receiver operating characteristic (ROC) was used to calculate the area under the ROC curve (AUC). DeLong's method was used to compare AUCs. Correlations were calculated using Pearson's method. P values <0.05 were considered significant. RESULTS: Automated segmentation resulted in a median (interquartile range [IQR]) Dice score of 0.87 (0.62-0.93), with similar volumetric measurements (R = 0.95, P < 0.05). Automated volumetric measurements were significantly correlated with FTV (R = 0.80). Tumor volume-derived from deep learning of DCE-MRI was associated with RCB, yielding an AUC of 0.76 to discriminate between RCB-0/I and RCB-II/III, performing similar to the FTV-based model (AUC = 0.77, P = 0.66). Performance was comparable across institutions (IQR AUC: 0.71-0.84). DATA CONCLUSION: Deep learning-based segmentation estimates changes in tumor load on DCE-MRI that are associated with RCB after NAC and is robust against variations between institutions. EVIDENCE LEVEL: 2. TECHNICAL EFFICACY: Stage 4.

Automatic Assessment of Lower-Limb Alignment from Computed Tomography.

BACKGROUND: Preoperative planning of lower-limb realignment surgical procedures necessitates the quantification of alignment parameters by using landmarks placed on medical scans. Conventionally, alignment measurements are performed on 2-dimensional (2D) standing radiographs. To enable fast and accurate 3-dimensional (3D) planning of orthopaedic surgery, automatic calculation of the lower-limb alignment from 3D bone models is required. The goal of this study was to develop, validate, and apply a method that automatically quantifies the parameters defining lower-limb alignment from computed tomographic (CT) scans. METHODS: CT scans of the lower extremities of 50 subjects were both manually and automatically segmented. Thirty-two manual landmarks were positioned twice on the bone segmentations to assess intraobserver reliability in a subset of 20 subjects. The landmarks were also positioned automatically using a shape-fitting algorithm. The landmarks were then used to calculate 25 angles describing the lower-limb alignment for all 50 subjects. RESULTS: The mean absolute difference (and standard deviation) between repeat measurements using the manual method was 2.01 ± 1.64 mm for the landmark positions and 1.05° ± 1.48° for the landmark angles, whereas the mean absolute difference between the manual and fully automatic methods was 2.17 ± 1.37 mm for the landmark positions and 1.10° ± 1.16° for the landmark angles. The manual method required approximately 60 minutes of manual interaction, compared with 12 minutes of computation time for the fully automatic method. The intraclass correlation coefficient showed good to excellent reliability between the manual and automatic assessments for 23 of 25 angles, and the same was true for the intraobserver reliability in the manual method. The mean for the 50 subjects was within the expected range for 18 of the 25 automatically calculated angles. CONCLUSIONS: We developed a method that automatically calculated a comprehensive range of 25 measurements that defined lower-limb alignment in considerably less time, and with differences relative to the manual method that were comparable to the differences between repeated manual assessments. This method could thus be used as an efficient alternative to manual assessment of alignment. LEVEL OF EVIDENCE: Diagnostic Level III . See Instructions for Authors for a complete description of levels of evidence.

Long-term Survival in Breast Cancer Patients Is Associated with Contralateral Parenchymal Enhancement at MRI: Outcomes of the SELECT Study.

Background Several single-center studies found that high contralateral parenchymal enhancement (CPE) at breast MRI was associated with improved long-term survival in patients with estrogen receptor (ER)-positive and human epidermal growth factor receptor 2 (HER2)-negative breast cancer. Due to varying sample sizes, population characteristics, and follow-up times, consensus of the association is currently lacking. Purpose To confirm whether CPE is associated with long-term survival in a large multicenter retrospective cohort, and to investigate if CPE is associated with endocrine therapy effectiveness. Materials and Methods This multicenter observational cohort included women with unilateral ER-positive HER2-negative breast cancer (tumor size ≤50 mm and ≤three positive lymph nodes) who underwent MRI from January 2005 to December 2010. Overall survival (OS), recurrence-free survival (RFS), and distant RFS (DRFS) were assessed. Kaplan-Meier analysis was performed to investigate differences in absolute risk after 10 years, stratified according to CPE tertile. Multivariable Cox proportional hazards regression analysis was performed to investigate whether CPE was associated with prognosis and endocrine therapy effectiveness. Results Overall, 1432 women (median age, 54 years [IQR, 47-63 years]) were included from 10 centers. Differences in absolute OS after 10 years were stratified according to CPE tertile as follows: 88.5% (95% CI: 88.1, 89.1) in tertile 1, 85.8% (95% CI: 85.2, 86.3) in tertile 2, and 85.9% (95% CI: 85.4, 86.4) in tertile 3. CPE was independently associated with OS, with a hazard ratio (HR) of 1.17 (95% CI: 1.0, 1.36; P = .047), but was not associated with RFS (HR, 1.11; P = .16) or DRFS (HR, 1.11; P = .19). The effect of endocrine therapy on survival could not be accurately assessed; therefore, the association between endocrine therapy efficacy and CPE could not reliably be estimated. Conclusion High contralateral parenchymal enhancement was associated with a marginally decreased overall survival in patients with estrogen receptor-positive and human epidermal growth factor receptor 2-negative breast cancer, but was not associated with recurrence-free survival (RFS) or distant RFS. Published under a CC BY 4.0 license. Supplemental material is available for this article. See also the editorial by Honda and Iima in this issue.

Thermosensitive Biodegradable Hydrogels for Local and Controlled Cerebral Delivery of Proteins: MRI-Based Monitoring of In Vitro and In Vivo Protein Release.

Hydrogels have been suggested as novel drug delivery systems for sustained release of therapeutic proteins in various neurological disorders. The main advantage these systems offer is the controlled, prolonged exposure to a therapeutically effective dose of the released drug after a single intracerebral injection. Characterization of controlled release of therapeutics from a hydrogel is generally performed in vitro, as current methods do not allow for in vivo measurements of spatiotemporal distribution and release kinetics of a loaded protein. Importantly, the in vivo environment introduces many additional variables and factors that cannot be effectively simulated under in vitro conditions. To address this, in the present contribution, we developed a noninvasive in vivo magnetic resonance imaging (MRI) method to monitor local protein release from two injected hydrogels of the same chemical composition but different initial water contents. We designed a biodegradable hydrogel formulation composed of low and high concentration thermosensitive polymer and thiolated hyaluronic acid, which is liquid at room temperature and forms a gel due to a combination of physical and chemical cross-linking upon injection at 37 °C. The in vivo protein release kinetics from these gels were assessed by MRI analysis utilizing a model protein labeled with an MR contrast agent, i.e. gadolinium-labeled albumin (74 kDa). As proof of principle, the release kinetics of the hydrogels were first measured with MRI in vitro. Subsequently, the protein loaded hydrogels were administered in male Wistar rat brains and the release in vivo was monitored for 21 days. In vitro, the thermosensitive hydrogels with an initial water content of 81 and 66% released 64 ± 3% and 43 ± 3% of the protein loading, respectively, during the first 6 days at 37 °C. These differences were even more profound in vivo, where the thermosensitive hydrogels released 83 ± 16% and 57 ± 15% of the protein load, respectively, 1 week postinjection. Measurement of volume changes of the gels over time showed that the thermosensitive gel with the higher polymer concentration increased more than 4-fold in size in vivo after 3 weeks, which was substantially different from the in vitro behavior where a volume change of 35% was observed. Our study demonstrates the potential of MRI to noninvasively monitor in vivo intracerebral protein release from a locally administered in situ forming hydrogel, which could aid in the development and optimization of such drug delivery systems for brain disorders.

Generative models for reproducible coronary calcium scoring.

Purpose: Coronary artery calcium (CAC) score, i.e., the amount of CAC quantified in CT, is a strong and independent predictor of coronary heart disease (CHD) events. However, CAC scoring suffers from limited interscan reproducibility, which is mainly due to the clinical definition requiring application of a fixed intensity level threshold for segmentation of calcifications. This limitation is especially pronounced in non-electrocardiogram-synchronized computed tomography (CT) where lesions are more impacted by cardiac motion and partial volume effects. Therefore, we propose a CAC quantification method that does not require a threshold for segmentation of CAC. Approach: Our method utilizes a generative adversarial network (GAN) where a CT with CAC is decomposed into an image without CAC and an image showing only CAC. The method, using a cycle-consistent GAN, was trained using 626 low-dose chest CTs and 514 radiotherapy treatment planning (RTP) CTs. Interscan reproducibility was compared to clinical calcium scoring in RTP CTs of 1662 patients, each having two scans. Results: A lower relative interscan difference in CAC mass was achieved by the proposed method: 47% compared to 89% manual clinical calcium scoring. The intraclass correlation coefficient of Agatston scores was 0.96 for the proposed method compared to 0.91 for automatic clinical calcium scoring. Conclusions: The increased interscan reproducibility achieved by our method may lead to increased reliability of CHD risk categorization and improved accuracy of CHD event prediction.

Explainable artificial intelligence (XAI) in deep learning-based medical image analysis.

With an increase in deep learning-based methods, the call for explainability of such methods grows, especially in high-stakes decision making areas such as medical image analysis. This survey presents an overview of explainable artificial intelligence (XAI) used in deep learning-based medical image analysis. A framework of XAI criteria is introduced to classify deep learning-based medical image analysis methods. Papers on XAI techniques in medical image analysis are then surveyed and categorized according to the framework and according to anatomical location. The paper concludes with an outlook of future opportunities for XAI in medical image analysis.

Efficient cascaded V-net optimization for lower extremity CT segmentation validated using bone morphology assessment.

Semantic segmentation of bone from lower extremity computerized tomography (CT) scans can improve and accelerate the visualization, diagnosis, and surgical planning in orthopaedics. However, the large field of view of these scans makes automatic segmentation using deep learning based methods challenging, slow and graphical processing unit (GPU) memory intensive. We investigated methods to more efficiently represent anatomical context for accurate and fast segmentation and compared these with state-of-the-art methodology. Six lower extremity bones from patients of two different datasets were manually segmented from CT scans, and used to train and optimize a cascaded deep learning approach. We varied the number of resolution levels, receptive fields, patch sizes, and number of V-net blocks. The best performing network used a multi-stage, cascaded V-net approach with 1283 -643 -323 voxel patches as input. The average Dice coefficient over all bones was 0.98 ± 0.01, the mean surface distance was 0.26 ± 0.12 mm and the 95th percentile Hausdorff distance 0.65 ± 0.28 mm. This was a significant improvement over the results of the state-of-the-art nnU-net, with only approximately 1/12th of training time, 1/3th of inference time and 1/4th of GPU memory required. Comparison of the morphometric measurements performed on automatic and manual segmentations showed good correlation (Intraclass Correlation Coefficient [ICC] >0.8) for the alpha angle and excellent correlation (ICC >0.95) for the hip-knee-ankle angle, femoral inclination, femoral version, acetabular version, Lateral Centre-Edge angle, acetabular coverage. The segmentations were generally of sufficient quality for the tested clinical applications and were performed accurately and quickly compared to state-of-the-art methodology from the literature.

Synthetic CT for the planning of MR-HIFU treatment of bone metastases in pelvic and femoral bones: a feasibility study.

OBJECTIVES: Visualization of the bone distribution is an important prerequisite for MRI-guided high-intensity focused ultrasound (MRI-HIFU) treatment planning of bone metastases. In this context, we evaluated MRI-based synthetic CT (sCT) imaging for the visualization of cortical bone. METHODS: MR and CT images of nine patients with pelvic and femoral metastases were retrospectively analyzed in this study. The metastatic lesions were osteolytic, osteoblastic or mixed. sCT were generated from pre-treatment or treatment MR images using a UNet-like neural network. sCT was qualitatively and quantitatively compared to CT in the bone (pelvis or femur) containing the metastasis and in a region of interest placed on the metastasis itself, through mean absolute difference (MAD), mean difference (MD), Dice similarity coefficient (DSC), and root mean square surface distance (RMSD). RESULTS: The dataset consisted of 3 osteolytic, 4 osteoblastic and 2 mixed metastases. For most patients, the general morphology of the bone was well represented in the sCT images and osteolytic, osteoblastic and mixed lesions could be discriminated. Despite an average timespan between MR and CT acquisitions of 61 days, in bone, the average (± standard deviation) MAD was 116 ± 26 HU, MD - 14 ± 66 HU, DSC 0.85 ± 0.05, and RMSD 2.05 ± 0.48 mm and, in the lesion, MAD was 132 ± 62 HU, MD - 31 ± 106 HU, DSC 0.75 ± 0.2, and RMSD 2.73 ± 2.28 mm. CONCLUSIONS: Synthetic CT images adequately depicted the cancellous and cortical bone distribution in the different lesion types, which shows its potential for MRI-HIFU treatment planning. KEY POINTS: • Synthetic computed tomography was able to depict bone distribution in metastatic lesions. • Synthetic computed tomography images intrinsically aligned with treatment MR images may have the potential to facilitate MR-HIFU treatment planning of bone metastases, by combining visualization of soft tissues and cancellous and cortical bone.

Deep learning-based whole-heart segmentation in 4D contrast-enhanced cardiac CT.

Automatic cardiac chamber and left ventricular (LV) myocardium segmentation over the cardiac cycle significantly extends the utilization of contrast-enhanced cardiac CT, potentially enabling in-depth assessment of cardiac function. Therefore, we evaluate an automatic method for cardiac chamber and LV myocardium segmentation in 4D cardiac CT. In this study, 4D contrast-enhanced cardiac CT scans of 1509 patients selected for transcatheter aortic valve implantation with 21,605 3D images, were divided into development (N = 12) and test set (N = 1497). 3D convolutional neural networks were trained with end-systolic (ES) and end-diastolic (ED) images. Dice similarity coefficient (DSC) and average symmetric surface distance (ASSD) were computed for 3D segmentations at ES and ED in the development set via cross-validation, and for 2D segmentations in four cardiac phases for 81 test set patients. Segmentation quality in the full test set of 1497 patients was assessed visually on a three-point scale per structure based on estimated overlap with the ground truth. Automatic segmentation resulted in a mean DSC of 0.89 ± 0.10 and ASSD of 1.43 ± 1.45 mm in 12 patients in 3D, and a DSC of 0.89 ± 0.08 and ASSD of 1.86 ± 1.20 mm in 81 patients in 2D. The qualitative evaluation in the whole test set of 1497 patients showed that automatic segmentations were assigned grade 1 (clinically useful) in 98.5%, 92.2%, 83.1%, 96.3%, and 91.6% of cases for LV cavity and myocardium, right ventricle, left atrium, and right atrium. Our automatic method using convolutional neural networks performed clinically useful segmentation across the cardiac cycle in a large set of 4D cardiac CT images, potentially enabling in-depth assessment of cardiac function.

AI-Based Radiation Dose Quantification for Estimation of Heart Disease Risk in Breast Cancer Survivors After Radiation Therapy.

PURPOSE: To investigate whether the dose planned for cardiac structures is associated with the risk of heart disease (HD) in patients with breast cancer treated with radiation therapy, and whether this association is modified by the presence of coronary artery calcification (CAC). METHODS AND MATERIALS: Radiation therapy planning computed tomographic (CT) scans and corresponding dose distribution maps of 5561 patients were collected, 5300 patients remained after the exclusion of ineligible patients and duplicates; 1899 patients received their CT scan before 2011, allowing long follow-up. CAC was detected automatically. Using an artificial intelligence-based method, the cardiac structures (heart, cardiac chambers, large arteries, 3 main coronary arteries) were segmented. The planned radiation dose to each structure separately and to the whole heart were determined. Patients were assigned to a low-, medium-, or high-dose group based on the dose to the respective heart structure. Information on HD hospitalization and mortality was obtained for each patient. The association of planned radiation dose to cardiac structures with risk of HD was investigated in patients with and without CAC using Cox proportional hazard analysis in the long follow-up population. Tests for interaction were performed. RESULTS: After a median follow-up of 96.0 months (interquartile range, 84.2-110.4 months) in the long follow-up group, 135 patients were hospitalized for HD or died of HD. If the dose to a structure increased 1 Gy, the relative HD risk increased by 3% to 11%. The absolute increase in HD risk was substantially higher in patients with CAC (event-ratelow-dose = 14-15 vs event-ratehigh-dose = 15-34 per 1000 person-years) than in patients without CAC (event-ratelow-dose = 6-8 vs event-ratehigh-dose = 5-17 per 1000 person-years). No interaction between CAC and radiation dose was found. CONCLUSIONS: Radiation exposure of cardiac structures is associated with increased risk of HD. Automatic segmentation of cardiac structures enables spatially localized dose estimation, which can aid in the prevention of radiation therapy-induced cardiac damage. This could be especially valuable in patients with breast cancer and CAC.

AI-Based Quantification of Planned Radiation Therapy Dose to Cardiac Structures and Coronary Arteries in Patients With Breast Cancer.

PURPOSE: The purpose of this work is to develop and evaluate an automatic deep learning method for segmentation of cardiac chambers and large arteries, and localization of the 3 main coronary arteries in radiation therapy planning on computed tomography (CT). In addition, a second purpose is to determine the planned radiation therapy dose to cardiac structures for breast cancer therapy. METHODS AND MATERIALS: Eighteen contrast-enhanced cardiac scans acquired with a dual-layer-detector CT scanner were included for method development. Manual reference annotations of cardiac chambers, large arteries, and coronary artery locations were made in the contrast scans and transferred to virtual noncontrast images, mimicking noncontrast-enhanced CT. In addition, 31 noncontrast-enhanced radiation therapy treatment planning CTs with corresponding dose-distribution maps of breast cancer cases were included for evaluation. For reference, cardiac chambers and large vessels were manually annotated in two 2-dimensional (2D) slices per scan (26 scans, totaling 52 slices) and in 3-dimensional (3D) scan volumes in 5 scans. Coronary artery locations were annotated on 3D imaging. The method uses an ensemble of convolutional neural networks with 2 output branches that perform 2 distinct tasks: (1) segmentation of the cardiac chambers and large arteries and (2) localization of coronary arteries. Training was performed using reference annotations and virtual noncontrast cardiac scans. Automatic segmentation of the cardiac chambers and large vessels and the coronary artery locations was evaluated in radiation therapy planning CT with Dice score (DSC) and average symmetrical surface distance (ASSD). The correlation between dosimetric parameters derived from the automatic and reference segmentations was evaluated with R2. RESULTS: For cardiac chambers and large arteries, median DSC was 0.76 to 0.88, and the median ASSD was 0.17 to 0.27 cm in 2D slice evaluation. 3D evaluation found a DSC of 0.87 to 0.93 and an ASSD of 0.07 to 0.10 cm. Median DSC of the coronary artery locations ranged from 0.80 to 0.91. R2 values of dosimetric parameters were 0.77 to 1.00 for the cardiac chambers and large vessels, and 0.76 to 0.95 for the coronary arteries. CONCLUSIONS: The developed and evaluated method can automatically obtain accurate estimates of planned radiation dose and dosimetric parameters for the cardiac chambers, large arteries, and coronary arteries.

MRI-based synthetic CT shows equivalence to conventional CT for the morphological assessment of the hip joint.

This study evaluated the accuracy of synthetic computed tomography (sCT), as compared to CT, for the 3D assessment of the hip morphology. Thirty male patients with asymptomatic hips, referred for magnetic resonance (MR) imaging and CT, were included in this retrospective study. sCT images were generated from three-dimensional radiofrequency-spoiled T1-weighted multi-echo gradient-echo MR images using a commercially available deep learning-enabled software and were compared with CT images through mean error and surface distance computation and by means of eight clinical morphometric parameters relevant for hip care. Parameters included center-edge angle (CEA), sharp angle, acetabular index, extrusion index, femoral head center-to-midline distance, acetabular version (AV), and anterior and posterior acetabular sector angles. They were measured by two senior orthopedic surgeons and a radiologist in-training on CT and sCT images. The reliability and agreement of CT- and sCT-based measurements were assessed using intraclass correlation coefficients (ICCs) for absolute agreement, Bland-Altman plots, and two one-sided tests for equivalence. The surface distance between CT- and sCT-based bone models were on average submillimeter. CT- and sCT-based measurements showed moderate to excellent interobserver and intraobserver correlation (0.56 < ICC < 0.99). In particular, the inter/intraobserver agreements were good for AV (ICC > 0.75). For CEA, the intraobserver agreement was good (ICC > 0.75) and the interobserver agreement was moderate (ICC > 0.69). Limits of agreements were similar between intraobserver CT and intermodal measurements. All measurements were found statistically equivalent, with average intermodal differences within the intraobserver limits of agreement. In conclusion, sCT and CT were equivalent for the assessment of the hip joint bone morphology.

CT to MR registration of complex deformations in the knee joint through dual quaternion interpolation of rigid transforms.

Purpose.To develop a method that enables computed tomography (CT) to magnetic resonance (MR) image registration of complex deformations typically encountered in rotating joints such as the knee joint.Methods.We propose a workflow, denoted quaternion interpolated registration (QIR), consisting of three steps, which makes use of prior knowledge of tissue properties to initialise deformable registration. In the first step, the rigid skeletal components were individually registered. Next, the deformation of soft tissue was estimated using a dual quaternion-based interpolation method. In the final step, the registration was fine-tuned with a rigidity-constrained deformable registration step. The method was applied to paired, unregistered CT and MR images of the knee of 92 patients. It was compared to registration using B-Splines (BS) and B-Splines with a rigidity penalty (BSRP). Registration accuracy was evaluated using mutual information, and by calculating Dice similarity coefficient (DSC), mean absolute surface distance (MASD) and 95th percentile Hausdorff distance (HD95) on bone, and DSC on water and fat dominated tissue. To evaluate the rigidity of bone in the registration, the Jacobian determinant (JD) was calculated.Results.QIR achieved improved results with 0.93, 0.76 mm and 1.88 mm on the DSC, MASD and HD95 metrics on bone, compared to 0.87, 1.40 mm and 4.99 mm for method and 0.87, 1.40 mm and 3.56 mm for the BSRP method. The average DSC of water and fat was 0.77 and 0.86 for the QIR, 0.75 and 0.84 for BS and 0.74 and 0.84 for BSRP. Comparison of the median JD and median interquartile (IQR) ranges of the JD indicated that the QIR (1.00 median, 0.03 IQR) resulted in higher rigidity in the rigid skeletal tissues compared to the BS (0.98 median, 0.19 IQR) and BSRP (1.00 median, 0.05 IQR) methods.Conclusion.This study showed that QIR could improve the outcome of complex registration problems, encountered in joints involving rigid and non-rigid bodies such as occur in the knee, as compared to a conventional registration approach.

Fiber orientation distribution from diffusion MRI: Effects of inaccurate response function calibration.

BACKGROUND AND PURPOSE: Diffusion MRI of the brain enables to quantify white matter fiber orientations noninvasively. Several approaches have been proposed to estimate such characteristics from diffusion MRI data with spherical deconvolution being one of the most widely used methods. Spherical deconvolution requires to define--or derive from the data--a response function, which is used to compute the fiber orientation distribution (FOD). Different characteristics of the response function are expected to affect the FOD computation and the subsequent fiber tracking. METHODS: In this work, we explored the effects of inaccuracies in the shape factors of the response function on the FOD characteristics. RESULTS: With simulations, we show that the apparent fiber density could be doubled in the presence of underestimated shape factors in the response functions, whereas the overestimation of the shape factor will cause more spurious peaks in the FOD, especially when the signal-to-noise ratio is below 15. Moreover, crossing fiber populations with a separation angle smaller than 60° were more sensitive to inaccuracies in the response function than fiber populations with more orthogonal separation angles. Results with in vivo data demonstrate angular deviations in the FODs and spurious peaks as a result of modified shape factors of the response function, while the reconstruction of the main parts of fiber bundles is well preserved. CONCLUSIONS: This work sheds light on how specific aspects of the response function shape can affect the estimated FODs, and highlights the importance of a proper calibration/definition of the response function.

Interleaved water and fat MR thermometry for monitoring high intensity focused ultrasound ablation of bone lesions.

PURPOSE: To demonstrate that interleaved MR thermometry can monitor temperature in water and fat with adequate temporal resolution. This is relevant for high intensity focused uUltrasounds (HIFU) treatment of bone lesions, which are often found near aqueous tissues, as muscle, or embedded in adipose tissues, as subcutaneous fat and bone marrow. METHODS: Proton resonance frequency shift (PRFS)-based thermometry scans and T1 -based 2D variable flip angle (2D-VFA) thermometry scans were acquired alternatingly over time. Temperature in water was monitored using PRFS thermometry, and in fat by 2D-VFA thermometry with slice profile effect correction. The feasibility of interleaved water/fat temperature monitoring was studied ex vivo in porcine bone during MR-HIFU sonication. Precision and stability of measurements in vivo were evaluated in a healthy volunteer under non-heating conditions. RESULTS: The method allowed observing temperature change over time in muscle and fat, including bone marrow, during MR-HIFU sonication, with a temporal resolution of 6.1 s. In vivo, the apparent temperature change was stable on the time scale of the experiment: In 7 min the systematic drift was <0.042°C/min in muscle (PRFS after drift correction) and <0.096°C/min in bone marrow (2D-VFA). The SD of the temperature change averaged over time was 0.98°C (PRFS) and 2.7°C (2D-VFA). CONCLUSIONS: Interleaved MR thermometry allows temperature measurements in water and fat with a temporal resolution high enough for monitoring HIFU ablation. Specifically, combined fat and water thermometry provides uninterrupted information on temperature changes in tissue close to the bone cortex.

Rapid 2D variable flip angle method for accurate and precise T1 measurements over a wide range of T1 values.

PURPOSE: To perform dynamic T1 mapping using a 2D variable flip angle (VFA) method, a correction for the slice profile effect is needed. In this work we investigated the impact of flip angle selection and excitation RF pulse profile on the performance of slice profile correction when applied to T1 mapping over a range of T1 values. METHODS: A correction of the slice profile effect is proposed, based on Bloch simulation of steady-state signals. With this correction, Monte Carlo simulations were performed to assess the accuracy and precision of 2D VFA T1 mapping in the presence of noise, for RF pulses with time-bandwidth products of 2, 3 and 10 and with flip angle pairs in the range [1°-90°]. To evaluate its performance over a wide range of T1 , maximum errors were calculated for six T1 values between 50 ms and 1250 ms. The method was demonstrated using in vitro and in vivo experiments. RESULTS: Without corrections, 2D VFA severely underestimates T1 . Slice profile errors were effectively reduced with the correction based on simulations, both in vitro and in vivo. The precision and accuracy of the method depend on the nominal T1 values, the FA pair, and the RF pulse shape. FA pairs leading to <5% errors in T1 can be identified for the common RF shapes, for T1 values between 50 ms and 1250 ms. CONCLUSIONS: 2D VFA T1 mapping with Bloch-simulation-based correction can deliver T1 estimates that are accurate and precise to within 5% over a wide T1 range.

Toward Computer-Assisted Triaging of Magnetic Resonance Imaging-Guided Biopsy in Preoperative Breast Cancer Patients.

OBJECTIVES: Incidental MR-detected breast lesions (ie, additional lesions to the index cancer) pose challenges in the preoperative workup of patients with early breast cancer. We pursue computer-assisted triaging of magnetic resonance imaging (MRI)-guided breast biopsy of additional lesions at high specificity. MATERIALS AND METHODS: We investigated 316 consecutive female patients (aged 26 to 76 years; mean, 54 years) with early breast cancer who received preoperative multiparametric breast MRI between 2013 and 2016. In total, 82 (26%) of 316 patients had additional breast lesions on MRI. These 82 patients had 101 additional lesions in total, 51 were benign and 50 were malignant. We collected 4 clinical features and 46 MRI radiomic features from T1-weighted dynamic contrast-enhanced imaging, high-temporal-resolution dynamic contrast-enhanced imaging, T2-weighted imaging, and diffusion-weighted imaging. A multiparametric computer-aided diagnosis (CAD) model using 10-fold cross-validated ridge regression was constructed. The sensitivities were calculated at operating points corresponding to 98%, 95%, and 90% specificity. The model calibration performance was evaluated by calibration plot analysis and goodness-of-fit tests. The model was tested in an independent testing cohort of 187 consecutive patients from 2017 and 2018 (aged 35 to 76 years; mean, 59 years). In this testing cohort, 45 (24%) of 187 patients had 55 additional breast lesions in total, 23 were benign and 32 were malignant. RESULTS: The multiparametric CAD model correctly identified 48% of the malignant additional lesions with a specificity of 98%. At specificity 95% and 90%, the sensitivity was 62% and 72%, respectively. Calibration plot analysis and goodness-of-fit tests indicated that the model was well fitted.In the independent testing cohort, the specificity was 96% and the sensitivity 44% at the 98% specificity operating point of the training set. At operating points 95% and 90%, the specificity was 83% at 69% sensitivity and the specificity was 78% at 81% sensitivity, respectively. CONCLUSIONS: The multiparametric CAD model showed potential to identify malignant disease extension with near-perfect specificity in approximately half the population of preoperative patients originally indicated for a breast biopsy. In the other half, patients would still proceed to MRI-guided biopsy to confirm absence of malignant disease. These findings demonstrate the potential to triage MRI-guided breast biopsy.