Search

BackgroundOne in eight men will be affected by prostate cancer (PCa) in their lives. While the current clinical standard prognostic marker for PCa is the Gleason score, it is subject to interreviewer variability. This study compares two machine learning methods for discriminating between high- and low-grade PCa on histology from 47 PCa patients.MethodsDigitized slides were annotated by a GU fellowship-trained pathologist. High-resolution tiles were extracted from annotated and unlabeled tissue. Glands were segmented and pathomic features were calculated and averaged across each patient. Patients were separated into a training set of 31 patients (Cohort A, n=9345 tiles) and a testing cohort of 16 patients (Cohort B, n=4375 tiles). Tiles from Cohort A were used to train a compact classification ensemble model and a ResNet model to discriminate tumor and were compared to pathologist annotations.ResultsThe ensemble and ResNet models had overall accuracies of 89% and 88%, respectively. The ResNet model was additionally able to differentiate Gleason patterns on data from Cohort B while the ensemble model was not.ConclusionsOur results suggest that quantitative pathomic features calculated from PCa histology can distinguish regions of cancer; how-ever, texture features captured by deep learning frameworks better differentiate unique Gleason patterns.

BackgroundThis study identified a clinically significant subset of glioma patients with tumor outside of contrast-enhancement present at autopsy, and subsequently developed a method for detecting non-enhancing tumor using radio-pathomic mapping. We tested the hypothesis that autopsy-based radio-pathomic tumor probability maps would be able to non-invasively identify areas of infiltrative tumor beyond traditional imaging signatures.MethodsA total of 159 tissue samples from 65 subjects were aligned to MRI acquired nearest to death for this study. Demographic and survival characteristics for patients with and without tumor beyond the contrast-enhancing margin were computed. An ensemble algorithm was used to predict pixelwise tumor presence from pathological annotations using segmented cellularity (Cell), extracellular fluid (ECF), and cytoplasm (Cyt) density as input (6 train/3 test subjects). A second level of ensemble algorithms were used to predict voxel-wise Cell, ECF, and Cyt on the full dataset (43 train/22 test subjects) using 5-by-5 voxel tiles from T1, T1+C, FLAIR, and ADC as input. The models were then combined to generate non-invasive whole brain maps of tumor probability.ResultsTumor outside of contrast was identified in 41.5 percent of patients, who showed worse survival outcomes (HR=3.90, p<0.001). Tumor probability maps reliably tracked non-enhancing tumor in the test set, external data collected pre-surgery, and longitudinal data to identify treatment-related changes and anticipate recurrence.ConclusionsThis study developed a multi-1 stage model for mapping gliomas using autopsy tissue samples as ground truth, which was able to identify regions of tumor beyond traditional imaging signatures.

Background: Anterior cruciate ligament (ACL) injury is a military occupational hazard that may be attributed to an individual’s knee biomechanics and joint anatomy. This study sought to determine if greater flexion when landing with load resulted in knee biomechanics thought to decrease ACL injury risk and whether knee biomechanics during landing relate to knee anatomic metrics. Hypothesis: Anatomic metrics regarding the slope and concavity of the tibial plateau will exhibit a significant relation to the increased anterior shear force on the knee and decreased knee flexion posture during landing with body-borne load. Study Design: Descriptive laboratory study. Methods: Twenty male military personnel completed a drop landing task with 3 load conditions: light (~6 kg), medium (15% body weight), and heavy (30% body weight). Participants were divided into groups based on knee flexion exhibited when landing with the heavy load (high- and low-Δflexion). Tibial slopes and depth were measured on weightbearing volumetric images of the knee obtained with a prototype cone beam computed tomography system. Knee biomechanics were submitted to a linear mixed model to evaluate the effect of landing group and load, with the anatomic metrics considered covariates. Results: Load increased peak proximal anterior tibial shear force ( P = .034), knee flexion angle ( P = .024), and moment ( P = .001) during landing. Only the high flexion group increased knee flexion ( P < .001) during weighted landings with medium and heavy loads. The low flexion group used greater knee abduction angle ( P = .030) and peak proximal anterior tibial shear force ( P = .034) when landing with load. Anatomic metrics did not differ between groups, but ratio of medial-to-lateral tibial slope and medial tibial depth predicted peak proximal anterior tibial shear force ( P = .009) and knee flexion ( P = .034) during landing, respectively. Conclusion: Increasing knee flexion is an attainable strategy to mitigate risk of ACL injury, but certain individuals may be predisposed to knee forces and biomechanics that load the ACL during weighted landings. Clinical Relevance: The ability to screen individuals for anatomic metrics that predict knee flexion may identify soldiers and athletes who require additional training to mitigate the risk of lower extremity injury.

Tumor heterogeneity in glioblastoma complicates delineation of active tumor using standard MR imaging. Pseudo-progression following treatment with chemotherapy and radiation (chemoRT) further complicates how tumors appear. T1-weighted subtraction maps (T1S) have been used to better identify subtly enhancing regions containing infiltrative tumor. This study examines the differences in tumor appearance between patients treated with chemoRT compared to a cohort opting out at autopsy, to understand how chemoRT changes contrast enhancement on MRI. Ten patients diagnosed with glioblastoma were recruited for whole brain donation for this study. Three patients received no treatment (chemoRT-), and seven received a combination of chemoRT and additional treatments (chemoRT+), including but not limited to bevacizumab (Bev) and tumor treating fields (TTF). Large tissue samples were taken at autopsy from whole brain samples sliced axially to align with the last clinical MRI using patient-specific 3D-printed slicing jigs. All tissue samples were hematoxylin and eosin (HE) stained and digitized at 40X resolution (27 total samples). The whole slide images (WSI) were annotated to outline regions containing necrosis without pseudopalisading cells, tumor with pseudopalisading necrosis, and infiltrative tumor. T1S were created for each patient by subtracting intensity normalized T1-weighted images from T1 post-contrast images (T1C). The annotated WSIs were aligned and resampled into MRI space using a custom software. A mixed effect model was used to compare the T1S intensity between chemoRT +/- cohorts within each pathological annotation, incorporating a random effect for subject. Mean T1S intensity was greater in untreated subjects when compared to chemoRT-subjects within each pathological annotation, including necrosis without pseudopalisading cells, tumor with pseudopalisading necrosis, and infiltrative tumor (p< 0.001). We show that chemoRT reduces the contrast enhancement in all aspects of pathologically validated tumor compartments, including infiltrative tumor. Future research is needed to examine if other patterns are evident in additional MR sequences.

BACKGROUND Diffusion-weighted imaging (DWI) is commonly used to detect prostate cancer, and a major clinical challenge is differentiating aggressive from indolent disease. PURPOSE To compare 14 site-specific parametric fitting implementations applied to the same dataset of whole-mount pathologically validated DWI to test the hypothesis that cancer differentiation varies with different fitting algorithms. STUDY TYPE Prospective. POPULATION Thirty-three patients prospectively imaged prior to prostatectomy. FIELD STRENGTH/SEQUENCE 3 T, field-of-view optimized and constrained undistorted single-shot DWI sequence. ASSESSMENT Datasets, including a noise-free digital reference object (DRO), were distributed to the 14 teams, where locally implemented DWI parameter maps were calculated, including mono-exponential apparent diffusion coefficient (MEADC), kurtosis (K), diffusion kurtosis (DK), bi-exponential diffusion (BID), pseudo-diffusion (BID*), and perfusion fraction (F). The resulting parametric maps were centrally analyzed, where differentiation of benign from cancerous tissue was compared between DWI parameters and the fitting algorithms with a receiver operating characteristic area under the curve (ROC AUC). STATISTICAL TEST Levene's test, P

PURPOSE Glioblastoma is one of the most common and deadly adult brain tumors. Current standard treatment is surgical resection followed by radiation and concomitant chemotherapy (chemoRT). Glioblastoma progression is monitored using MRI, primarily relying on post-contrast T1-weighted imaging (T1C). Unfortunately, tumor invasion is known to extend beyond traditional contrast enhancement. T1-subtraction (T1S) maps have been introduced as a better tumor volume estimate. In this study we compare T1S map derived tumor annotations to a radiologist for identifying histologically confirmed tumor in patients with differing treatment histories at autopsy. METHODS Ten patients with autopsy confirmed glioblastoma and MRI within 1 month of death were recruited for this study. Seven patients received chemoRT combined (chemoRT+) and three patients received no treatment beyond surgery (chemoRT-). Patient’s brains were sliced axially in the same orientation as their final MRI using a patient-specific slicing jig. Large tissue samples were taken, processed, embedded in paraffin, stained for hematoxylin and eosin, and digitized at 40x resolution. Digital images were annotated for infiltrative tumor, pseudopalisading necrosis, and necrosis without palisading cells. T1S and radiologist annotations were created for each patient using their final MRI (mean 18 days prior to death). The annotated histology images were aligned and resampled into MRI space using custom software and the overlap of pathologically confirmed tumor and MRI derived annotations was compared. RESULTS T1S maps alone were significantly better at identifying areas of histologically confirmed tumor in chemoRT+ patients compared to chemoRT- patients (p=0.043). T1S derived annotations overlapped with 52% of histologically confirmed tumor in the chemoRT- patients and 78% in the chemoRT+. The radiologist drawn tumor masks were more accurate in chemoRT+ patients, identifying 61% confirmed tumor (trending, p=0.097, chemoRT+=61%). CONCLUSION These results demonstrate the difficulty of identifying pathologically confirmed tumor outside contrast enhancement in glioblastoma patients, even in the untreated state.

Over half of all cancer patients receive radiation therapy (RT). However, radiation exposure to the heart can cause cardiotoxicity. Nevertheless, there is a paucity of data on RT-induced cardiac damage, with limited understanding of safe regional RT doses, early detection, prevention and management. A common initial feature of cardiotoxicity is asymptomatic dysfunction, which if left untreated may progress to heart failure. The current paradigm for cardiotoxicity detection and management relies primarily upon assessment of ejection fraction (EF). However, cardiac injury can occur without a clear change in EF.To identify magnetic resonance imaging (MRI) markers of early RT-induced cardiac dysfunction.We investigated the effect of RT on global and regional cardiac function and myocardial T1/T2 values at two timepoints post-RT using cardiac MRI in a rat model of localized cardiac RT. Rats who received image-guided whole-heart radiation of 24Gy were compared to sham-treated rats.The rats maintained normal global cardiac function post-RT. However, a deterioration in strain was particularly notable at 10-weeks post RT, and changes in circumferential strain were larger than changes in radial or longitudinal strain. Compared to sham, circumferential strain changes occurred at the basal, mid-ventricular and apical levels (p0.05 for all at both 8-weeks and 10-weeks post-RT), most of the radial strain changes occurred at the mid-ventricular (p=0.044 at 8-weeks post-RT) and basal (p=0.018 at 10-weeks post-RT) levels, and most of the longitudinal strain changes occurred at the apical (p=0.002 at 8-weeks post-RT) and basal (p=0.035 at 10-weeks post-RT) levels. Regionally, lateral myocardial segments showed the greatest worsening in strain measurements, and histologic changes supported these findings. Despite worsened myocardial strain post-RT, myocardial tissue displacement measures were maintained, or even increased. T1/T2 measurements showed small non-significant changes post-RT compared to values in non-irradiated rats.Our findings suggest MRI regional myocardial strain is a sensitive imaging biomarker for detecting RT-induced subclinical cardiac dysfunction prior to compromise of global cardiac function.

Current MRI signatures of brain cancer often fail to identify regions of hypercellularity beyond the contrast enhancing region. Therefore, this study used autopsy tissue samples aligned to clinical MRIs in order to quantify the relationship between intensity values and cellularity, as well as to develop a radio-pathomic model to predict cellularity using MRI data. This study used 93 samples collected at autopsy from 44 brain cancer patients. Tissue samples were processed, stained for hematoxylin and eosin (HE) and digitized for nuclei segmentation and cell density calculation. Pre- and post-gadolinium contrast T1-weighted images (T1, T1C), T2 fluid-attenuated inversion recovery (FLAIR) images, and apparent diffusion coefficient (ADC) images calculated from diffusion imaging were collected from each patients’ final acquisition prior to death. In-house software was used to align tissue samples to the FLAIR image via manually defined control points. Mixed effect models were used to assess the relationship between single image intensity and cellularity for each image. An ensemble learner was trained to predict cellularity using 5 by 5 voxel tiles from each image, employing a 2/3-1/3 train-test split for validation. Single image analyses found subtle associations between image intensity and cellularity, with a less pronounced relationship within GBM patients. The radio-pathomic model was able to accurately predict cellularity in the test set (RMSE = 1015 cells/mm2) and identified regions of hypercellularity beyond the contrast enhancing region. We concluded that a radio-pathomic model for cellularity is able to identify regions of hypercellular tumor beyond traditional imaging signatures.

Tumor treating fields (TTFields) are thought to disrupt the cell division process in glioma. This study uses autopsy tissue samples from patients recruited for brain donation to determine whether TTFields treatment duration and usage affects cellularity distributions beyond the gadolinium contrast enhancing region on T1-weighted (T1w) magnetic resonance imaging (MRI). At autopsy, 43 tissue samples from 18 patients who had undergone TTFields treatment (inclusion criteria: >50% compliance and 25-day duration) were collected from brain slices sectioned in-line with slices from the patients’ final MRI prior to death. Nuclei were segmented on the digitized hematoxylin and eosin (HE) stained tissue samples, which were then used to compute cell count across the slides. Tissue was then aligned to the patient’s final MRI scan prior to death using a custom in-house software, where manually defined control points were used to compute a nonlinear transform to warp the tissue to match the MRI. Histogram features including mean, median, 90th percentile, variance, and skewness, were computed from the cellularity distributions within regions outside the traditional tumor margin defined by T1w contrast enhancement for each subject. General linear models were fit to assess the effects of TTFields usage (percent of day) and duration (in days) on each histogram feature, controlling for age, overall survival, and time between TTFields treatment and death, as well as time between MRI acquisition and death. Longer TTFields treatment duration and overall survival was associated with significantly decreased skewness in the non-enhancing cellularity distributions (p< 0.05) while associations with other metrics remained statistically insignificant. These preliminary results in a small patient cohort suggest that TTFields duration may reduce the presence of high cellularity tails in the non-contrast enhancing distributions. Additional research in larger patient populations is warranted to better understand these findings given the number of confounding factors.

Purpose: Multiplex staining allows molecular co-localization analysis of same cell or tissue sections and maximizes the amount of data acquired from individual samples. We developed a non-linear registration framework for automated alignment of whole mount multiplexed histology images of prostate cancer. To achieve a precise automatic fit of bleached and restained high resolution tissue samples, a patch-based approach is proposed to improve annotation speed and analysis. Methods: The three-step co-registration process begins with a coarse low-resolution registration of the IHC stained image to the fixed H&E-stained image. The initial registration is then refined separately for each high-resolution patch using a smaller search window. Finally, registered patches are stitched back together using speeded up robust features (SURF). We apply the method to five multiplex whole mount prostate histology slides. To determine its effectiveness, we compare the automatic registration to the initial coarse registration and a manual control point based-method varying the number of control points. Results: For the control point-based approach, 25, 50, and 100 manually placed set landmarks resulted in a decrease of - 76.20%, -75.9% and -75.48% of the root mean squared error (RMSE), respectively. Compared to the initial registration an improvement of -76.29% of RMSE can be seen, illustrating the potential benefits of a patch-based automatic approach. Conclusion: The proposed method achieved excellent registration of the IHC to the input image. The automated method for registration of multiplexed histology images achieves high accuracy in shorter time and with greater reproducibility than conventional registration approaches and semi-automatic control points without the need for time consuming and subjective manual control-point setting. The accuracy of the fit especially improves in complex areas close to tissue tears and folds.

PURPOSE Tumor treatment fields (TTFields) are approved by the FDA for newly diagnosed as well as recurrent glioblastoma (GBM). TTFields have been shown to extend survival by 4.9 months in newly diagnosed GBM, and a landmark overall survival rate of 13% at 5 years. However, the specific effects remain widely unknown, which has prevented widespread clinical use of this treatment. METHODS This case study examines two glioblastoma patients, IDH-1 wildtype, MGMT unmethylated, that received TTFields (Optune) in addition to maintenance temozolomide (TMZ) following radiation (RT). Both cases were followed using standard MR imaging. Second resections were performed due to radiographic progression of contrast enhancement. RESULTS Although imaging was concerning for tumor progression, pathology showed only treatment effect, ultimately leading to the diagnosis of pseudoprogression. Both patients fell outside the normal expected timeline for chemo-radiation induced pseudoprogression. Based on the pathology, both patients resumed treatment with TMZ/TTFields. One patient expired at 25 months and one is still alive. CONCLUSIONS Pathologic confirmation was essential for guiding further treatment and allowed patients to continue treatment that was effective despite contrary indications on imaging. These findings suggest that pathological confirmation should be strongly considered in patients receiving TMZ/TTFields who develop radiographic progression, potentially with a less invasive biopsy procedure. Future studies should look to characterize the underlying mechanism of interaction between TTFields/TMZ and quantify the prevalence, associated risk factors and whether there is a genotype more susceptible. Both patients reported here had O(6)-methylguanine-DNA methyltransferase (MGMT) unmethylated GBM, and while about 60% of glioblastomas are diagnosed likewise, it is possible that MGMT methylation status plays a role in TTFields response. Better characterization of this phenomenon will improve treatment guidance, potentially reducing unnecessary surgeries and the discontinuance of successful therapies.

This study used large format autopsy tissue samples to compare radio-pathomic maps of brain cancer to a current tumor segmentation algorithm. We hypothesized that an MRI-based machine learning model trained with actual histology rather than radiologist annotations cellularity would 1) improve delineation between tumor and treatment effect, and 2) detect abnormal pathology beyond the contrast-enhancing tumor region. Seventeen patients with pathologically confirmed glioma were included in this study. At autopsy, 43 tissue samples were collected from 17 subjects from whole brain slices sectioned to align with the last axial MRI prior to death. Cellularity was calculated using a color deconvolution on 40X digitized H&E stained slides from the tissue samples. In-house custom software was used to align tissue samples and cellularity information to the FLAIR image using manually defined control points. The DeepMedic algorithm was trained to segment tumors using the BraTs 2017 dataset, and then applied to our patients in order to create automated tumor probability maps. An MRI-based ensemble algorithm using a 5x5 voxel searchlight (input: T1, T1C, FLAIR, ADC) was used to predict cellularity at each voxel, using tissue samples from 14 subjects as ground truth. Both models were applied to 3 withheld test subjects in order to compare tumor probability and cellularity predictions to the pathological ground truth. The mutual information between tumor probability and actual cellularity was 1X10-15, relatively low compared to the rad-path predicted cellularity (=0.16), despite the tumor prediction model accurately highlighting regions of contrast enhancement. Additionally, the radio-pathomic ensemble model correctly identify regions of hypercellularity beyond the tumor segmentation model as well as regions of within the segmented tumor area. This study demonstrates the utility of training machine learning models with pathological ground truth rather than radiologist annotations for predicting localized tumor information, particularly in the post-treatment stage.

Purpose: Our study predictively maps epithelium density in magnetic resonance imaging (MRI) space while varying the ground truth labels provided by five pathologists to quantify the downstream effects of interobserver variability. Approach: Clinical imaging and postsurgical tissue from 48 recruited prospective patients were used in our study. Tissue was sliced to match the MRI orientation and whole-mount slides were stained and digitized. Data from 28 patients (n = 33 slides) were sent to five pathologists to be annotated. Slides from the remaining 20 patients (n = 123 slides) were annotated by one of the five pathologists. Interpathologist variability was measured using Krippendorff’s alpha. Pathologist-specific radiopathomic mapping models were trained using a partial least-squares regression using MRI values to predict epithelium density, a known marker for disease severity. An analysis of variance characterized intermodel means difference in epithelium density. A consensus model was created and evaluated using a receiver operator characteristic classifying high grade versus low grade and benign, and was statistically compared to apparent diffusion coefficient (ADC). Results: Interobserver variability ranged from low to acceptable agreement (0.31 to 0.69). There was a statistically significant difference in mean predicted epithelium density values (p

To evaluate the influence of weight-bearing (WB) load in standard axial ankle syndesmotic measurements using cone beam CT (CBCT) examination of asymptomatic uninjured ankles. In this IRB approved, prospective study, patients with previous unilateral ankle fractures were recruited. We simultaneously scanned the injured ankles and asymptomatic contralateral ankles of 27 patients in both WB and NWB modes. For this study, only asymptomatic contralateral ankles with normal plain radiographs were included. Twelve standardized syndesmosis measurements at two axial planes (10 mm above the tibial plafond and 5 mm below the talar dome) were obtained by two expert readers using a custom CBCT viewer with the capability for geometric measurements between user-identified anatomical landmarks. Inter-reader reliability between two readers was obtained using the intra-class correlation coefficient (ICC). We compared the WB and NWB measurements using paired t test. Significant agreement was observed between two readers for both WB and NWB measurements (p

Infiltrative glioma beyond contrast enhancement on MRI is often difficult to identify with conventional imaging. In this study, we use large-format autopsy samples aligned to multi-parametric MRI to test the hypothesis that radio-pathomic machine learning models are able to accurately identify areas of infiltrative tumor beyond the contrast enhancing region. At autopsy, 140 tissue samples from 62 brain cancer patients were collected from brain slices sectioned to align with the patients’ last clinical MRI prior to death. Cell, extra-cellular fluid (ECF), and cytoplasm densities were computed from digitized, hematoxylin and eosin-stained samples, and a subset of 20 slides from 9 patients were annotated for tumor presence by a pathologist-trained technician. In-house custom software was used to align the tissue samples to the patients’ last clinical imaging, which included pre- and post-contrast T1, FLAIR, and ADC images. Bagging random forest models were then trained to predict cellularity, ECF, and cytoplasm density using 5-by-5 voxel tiles from each MRI as input. A 2/3-1/3 train-test split was used to validate model generalizability. A naïve Bayes classifier was trained to predict tumor class using cellularity, ECF, and cytoplasm segmentations within the annotation data set, again using a 2/3-1/3 train-test split to validate performance. The random forest models each accurately predicted cellularity, ECF, and cytoplasm density within the test data set, with root-mean-squared error values for each falling within one standard deviation of the ground truth. The histology-based tumor prediction model accurately predicted tumor, with a test set ROC AUC of 0.86. When using whole brain cellularity, ECF, and cytoplasm predictions from the random forest models as inputs for the naïve Bayes classifier, tumor probability maps identified regions of infiltrative tumor beyond contrast enhancement. Our results suggest that radio-pathomic maps of tumor probability accurately identify regions of infiltrative tumor beyond currently accepted MRI signatures.

Purpose: A prototype high-resolution extremity cone-beam CT (CBCT) system based on a CMOS detector was developed to support quantitative in vivo assessment of bone microarchitecture. We compare the performance of CMOS CBCT to an amorphous silicon (a-Si:H) FPD extremity CBCT in imaging of trabecular bone. Methods: The prototype CMOS-based CBCT involves a DALSA Xineos3030 detector (99 μm pixels) with 400 μm-thick CsI scintillator and a compact 0.3 FS rotating anode x-ray source. We compare the performance of CMOS CBCT to an a- Si:H FPD scanner built on a similar gantry, but using a Varian PaxScan2530 detector with 0.137 mm pixels and a 0.5 FS stationary anode x-ray source. Experimental studies include measurements of Modulation Transfer Function (MTF) for the detectors and in 3D image reconstructions. Image quality in clinical scenarios is evaluated in scans of a cadaver ankle. Metrics of trabecular microarchitecture (BV/TV, Bone Volume/Total Volume, TbSp, Trabecular Spacing, and TbTh, trabecular thickness) are obtained in a human ulna using CMOS CBCT and a-Si:H FPD CBCT and compared to gold standard μCT. Results: The CMOS detector achieves ~40% increase in the f20 value (frequency at which MTF reduces to 0.20) compared to the a-Si:H FPD. In the reconstruction domain, the FWHM of a 127 μm tungsten wire is also improved by ~40%. Reconstructions of a cadaveric ankle reveal enhanced modulation of trabecular structures with the CMOS detector and soft-tissue visibility that is similar to that of the a-Si:H FPD system. Correlations of the metrics of bone microarchitecture with gold-standard μCT are improved with CMOS CBCT: from 0.93 to 0.98 for BV/TV, from 0.49 to 0.74 for TbTh, and from 0.9 to 0.96 for TbSp. Conclusion: Adoption of a CMOS detector in extremity CBCT improved spatial resolution and enhanced performance in metrics of bone microarchitecture compared to a conventional a-Si:H FPD. The results support development of clinical applications of CMOS CBCT in quantitative imaging of bone health.

Purpose: In-vivo evaluation of bone microarchitecture remains challenging because of limited resolution of conventional orthopaedic imaging modalities. We investigate the performance of flat-panel detector extremity Cone-Beam CT (CBCT) in quantitative analysis of trabecular bone. To enable accurate morphometry of fine trabecular bone architecture, advanced CBCT pre-processing and segmentation algorithms are developed. Methods: The study involved 35 transilliac bone biopsy samples imaged on extremity CBCT (voxel size 75 μm, imaging dose ~13 mGy) and gold standard μCT (voxel size 7.67 μm). CBCT image segmentation was performed using (i) global Otsu’s thresholding, (ii) Bernsen’s local thresholding, (iii) Bernsen’s local thresholding with additional histogram-based global pre-thresholding, and (iv) the same as (iii) but combined with contrast enhancement using a Laplacian Pyramid. Correlations between extremity CBCT with the different segmentation algorithms and gold standard μCT were investigated for measurements of Bone Volume over Total Volume (BV/TV), Trabecular Thickness (Tb.Th), Trabecular Spacing (Tb.Sp), and Trabecular Number (Tb.N). Results: The combination of local thresholding with global pre-thresholding and Laplacian contrast enhancement outperformed other CBCT segmentation methods. Using this optimal segmentation scheme, strong correlation between extremity CBCT and μCT was achieved, with Pearson coefficients of 0.93 for BV/TV, 0.89 for Tb.Th, 0.91 for Tb.Sp, and 0.88 for Tb.N (all results statistically significant). Compared to a simple global CBCT segmentation using Otsu’s algorithm, the advanced segmentation method achieved ~20% improvement in the correlation coefficient for Tb.Th and ~50% improvement for Tb.Sp. Conclusions: Extremity CBCT combined with advanced image pre-processing and segmentation achieves high correlation with gold standard μCT in measurements of trabecular microstructure. This motivates ongoing development of clinical applications of extremity CBCT in in-vivo evaluation of bone health e.g. in early osteoarthritis and osteoporosis.

e14035 Background: With an average overall survival of 12-18 months, glioblastoma has a particularly grim diagnosis. Standard treatment of glioblastoma, following detection on MRI, is surgical resection followed by radiation therapy and chemotherapy and is monitored through MR imaging. Glioblastoma has a unique heterogenous nature that complicates visualization of subtly enhancing tumor. This study used autopsy tissue samples taken from glioblastoma patients with varying treatment, to examine the effects of treatment on cell density within regions of contrast enhancement, using T1-weighted subtraction maps (T1S) from the last MR images to death. Methods: Eight patients diagnosed with glioblastoma at autopsy were recruited for this study. Two patients had no treatment and six received a combination of chemo-radiation and other treatments, including but not limited to bevacizumab (Bev) and TTFields therapy. At autopsy, whole brain samples were sliced axially aligned to the patient’s final MRI to death. Time between last MRI and death ranged from 4-27 days (mean 16 days). Overall survival (OS) ranged from 4-538 days (mean 307 days). Large tissue samples were taken from regions of suspected tumor or treatment effect, for a total of 18 tissue samples. Tissue samples were processed, H&E stained, and digitized at 40X resolution (Huron Slide Scanner). Cell density (cells/mm2) was calculated using digital histology. T1S were created for each patient by subtracting intensity normalized T1 weighted images from T1 post contrast images (T1C). Digital histology was aligned and resampled into MRI space using manual control point registration. Mixed effect models were used to compare differences in cell density across contrast enhancement (T1SE vs. NE) as well as across treatment groups (treatment vs. no treatment). Results: Cellularity was compared across regions of T1S enhancement (T1SE) and non-enhancement (NE) within manually selected regions of interest. Cell density compared between regions of T1SE and NE was not different (p=0.219). Total cell density was increased in patients who had received treatment compared to no treatment in both regions of T1SE and NE (p=0.014). Conclusions: Overall, cell density was increased in patients who had received treatment after diagnosis of glioblastoma. Additional research is needed to examine the extent of treatment’s effect on cellularity of glioblastoma. This work begins to characterize the use of T1S in evaluating glioblastoma tumor burden in patients with varying treatment histories.[Table: see text]

In orthopedic surgeries, it is important to avoid intra-articular implant placements, which increase revision rates and the risk of arthritis. In order to support the intraoperative assessment and correction of surgical implants, we present an automatic detection approach using cone-beam computed tomography (CBCT). Multiple active shape models (ASM) with specified articular surface regions are used to isolate the joint spaces. Fast and easy-to-implement methods are integrated in the ASM segmentation to optimize the robustness and accuracy for intraoperative application. A cylinder detection method is applied to determine metal implants. Intersections between articular surfaces and cylinders are detected and used to find intra-articular collisions. Segmentations of two calcaneal articular surfaces were evaluated on 50 patient images and have shown average surface distance errors of 0.59 and 0.46 mm, respectively. The proposed model-independent segmentation at the specified articular surface regions allowed to significantly decrease the error by 22 and 25 % on average. The method was able to compensate suboptimal initializations for translations of up to 16 mm and rotations of up to 21 $$^{\circ }$$ . In a human cadaver test, articular perforations could be localized with an accuracy of 0.80 mm on average. A concept for automatic intraoperative detection of intra-articular implants in CBCT images was presented. The results show a reliable segmentation of articular surfaces in retrospective patient data and an accurate localization of misplaced implants in artificially created human cadaver test cases.

We present a novel reconstruction algorithm based on a general cone-beam CT forward model, which is capable of incorporating the blur and noise correlations that are exhibited in flat-panel CBCT measurement data. Specifically, the proposed model may include scintillator blur, focal-spot blur, and noise correlations due to light spread in the scintillator. The proposed algorithm (GPL-BC) uses a Gaussian Penalized-Likelihood objective function, which incorporates models of blur and correlated noise. In a simulation study, GPL-BC was able to achieve lower bias as compared with deblurring followed by FDK as well as a model-based reconstruction method without integration of measurement blur. In the same study, GPL-BC was able to achieve better line-pair reconstructions (in terms of segmented-image accuracy) as compared with deblurring followed by FDK, a model-based method without blur, and a model-based method with blur but not noise correlations. A prototype extremities quantitative cone-beam CT test-bench was used to image a physical sample of human trabecular bone. These data were used to compare reconstructions using the proposed method and model-based methods without blur and/or correlation to a registered $\mu $ CT image of the same bone sample. The GPL-BC reconstructions resulted in more accurate trabecular bone segmentation. Multiple trabecular bone metrics, including trabecular thickness (Tb.Th.) were computed for each reconstruction approach as well as the $\mu $ CT volume. The GPL-BC reconstruction provided the most accurate Tb.Th. measurement, 0.255 mm, as compared with the $\mu $ CT derived value of 0.193 mm, followed by the GPL-B reconstruction, the GPL-I reconstruction, and then the FDK reconstruction (0.271 mm, 0.309 mm, and 0.335 mm, respectively).

With the help of an intra-operative mobile C-arm CT, medical interventions can be verified and corrected, avoiding the need for a post-operative CT and a second intervention. An exact adjustment of standard plane positions is necessary for the best possible assessment of the anatomical regions of interest but the mobility of the C-arm causes the need for a time-consuming manual adjustment. In this article, we present an automatic plane adjustment at the example of calcaneal fractures. We developed two feature detection methods (2D and pseudo-3D) based on SURF key points and also transferred the SURF approach to 3D. Combined with an atlas-based registration, our algorithm adjusts the standard planes of the calcaneal C-arm images automatically. The robustness of the algorithms is evaluated using a clinical data set. Additionally, we tested the algorithm’s performance for two registration approaches, two resolutions of C-arm images and two methods for metal artifact reduction. For the feature extraction, the novel 3D-SURF approach performs best. As expected, a higher resolution ( $$512^3$$ voxel) leads also to more robust feature points and is therefore slightly better than the $$256^3$$ voxel images (standard setting of device). Our comparison of two different artifact reduction methods and the complete removal of metal in the images shows that our approach is highly robust against artifacts and the number and position of metal implants. By introducing our fast algorithmic processing pipeline, we developed the first steps for a fully automatic assistance system for the assessment of C-arm CT images.

Purpose Quantitative assessment of trabecular bone microarchitecture in extremity cone-beam CT (CBCT) would benefit from the high spatial resolution, low electronic noise, and fast scan time provided by complementary metal-oxide semiconductor (CMOS) x-ray detectors. We investigate the performance of CMOS sensors in extremity CBCT, in particular with respect to potential advantages of thin (

Purpose: Anatomical metrics of the tibiofemoral joint support assessment of joint stability and surgical planning. We propose an automated, atlas-based algorithm to streamline the measurements in 3D images of the joint and reduce userdependence of the metrics arising from manual identification of the anatomical landmarks. Methods: The method is initialized with coarse registrations of a set of atlas images to the fixed input image. The initial registrations are then refined separately for the tibia and femur and the best matching atlas is selected. Finally, the anatomical landmarks of the best matching atlas are transformed onto the input image by deforming a surface model of the atlas to fit the shape of the tibial plateau in the input image (a mesh-to-volume registration). We apply the method to weight-bearing volumetric images of the knee obtained from 23 subjects using an extremity cone-beam CT system. Results of the automated algorithm were compared to an expert radiologist for measurements of Static Alignment (SA), Medial Tibial Slope (MTS) and Lateral Tibial Slope (LTS). Results: Intra-reader variability as high as ~10% for LTS and 7% for MTS (ratio of standard deviation to the mean in repeated measurements) was found for expert radiologist, illustrating the potential benefits of an automated approach in improving the precision of the metrics. The proposed method achieved excellent registration of the atlas mesh to the input volumes. The resulting automated measurements yielded high correlations with expert radiologist, as indicated by correlation coefficients of 0.72 for MTS, 0.8 for LTS, and 0.89 for SA. Conclusions: The automated method for measurement of anatomical metrics of the tibiofemoral joint achieves high correlation with expert radiologist without the need for time consuming and error prone manual selection of landmarks.

Purpose: CMOS x-ray detectors offer small pixel sizes and low electronic noise that may support the development of novel high-resolution imaging applications of cone-beam CT (CBCT). We investigate the effects of CsI scintillator thickness on the performance of CMOS detectors in high resolution imaging tasks, in particular in quantitative imaging of bone microstructure in extremity CBCT. Methods: A scintillator thickness-dependent cascaded systems model of CMOS x-ray detectors was developed. Detectability in low-, high- and ultra-high resolution imaging tasks (Gaussian with FWHM of ~250 μm, ~80 𝜇m and ~40 μm, respectively) was studied as a function of scintillator thickness using the theoretical model. Experimental studies were performed on a CBCT test bench equipped with DALSA Xineos3030 CMOS detectors (99 μm pixels) with CsI scintillator thicknesses of 400 μm and 700 𝜇m, and a 0.3 FS compact rotating anode x-ray source. The evaluation involved a radiographic resolution gauge (0.6-5.0 lp/mm), a 127 μm tungsten wire for assessment of 3D resolution, a contrast phantom with tissue-mimicking inserts, and an excised fragment of human tibia for visual assessment of fine trabecular detail. Results: Experimental studies show ~35% improvement in the frequency of 50% MTF modulation when using the 400 μm scintillator compared to the standard nominal CsI thickness of 700 μm. Even though the high-frequency DQE of the two detectors is comparable, theoretical studies show a 14% to 28% increase in detectability index (d’2) of high- and ultra- high resolution tasks, respectively, for the detector with 400 μm CsI compared to 700 μm CsI. Experiments confirm the theoretical findings, showing improvements with the adoption of 400 μm panel in the visibility of the radiographic pattern (2x improvement in peak-to-through distance at 4.6 lp/mm) and a 12.5% decrease in the FWHM of the tungsten wire. Reconstructions of the tibial plateau reveal enhanced visibility of trabecular structures with the CMOS detector with 400 μm scinitllator. Conclusion: Applications on CMOS detectors in high resolution CBCT imaging of trabecular bone will benefit from using a thinner scintillator than the current standard in general radiography. The results support the translation of the CMOS sensor with 400 μm scinitllator.

Intraoperative mobile C-arm fluoroscopy is widely used for interventional verification in trauma surgery, high flexibility combined with low cost being the main advantages of the method. However, the lack of global device-to- patient orientation is challenging, when comparing the acquired data to other intrapatient datasets. In upper ankle joint fracture reduction accompanied with an unstable syndesmosis, a comparison to the unfractured contralateral site is helpful for verification of the reduction result. To reduce dose and operation time, our approach aims at the comparison of single projections of the unfractured ankle with volumetric images of the reduced fracture. For precise assessment, a pre-alignment of both datasets is a crucial step. We propose a contour extraction pipeline to estimate the joint space location for a prealignment of fluoroscopic C-arm projections containing the upper ankle joint. A quadtree-based hierarchical variance comparison extracts potential feature points and a Hough transform is applied to identify bone shaft lines together with the tibiotalar joint space. By using this information we can define the coarse orientation of the projections independent from the ankle pose during acquisition in order to align those images to the volume of the fractured ankle. The proposed method was evaluated on thirteen cadaveric datasets consisting of 100 projections each with manually adjusted image planes by three trauma surgeons. The results show that the method can be used to detect the joint space orientation. The correlation between angle deviation and anatomical projection direction gives valuable input on the acquisition direction for future clinical experiments.

Orthopedic fractures are often fixed using metal implants. The correct positioning of cylindrical implants such as surgical screws, rods and guide wires is highly important. Intraoperative 3D imaging is often used to ensure proper implant placement. However, 3D image interaction is time-consuming and requires experience. We developed an automatic method that simplifies and accelerates location assessment of cylindrical implants in 3D images. Our approach is composed of three major steps. At first, cylindrical characteristics are detected by analyzing image gradients in small image regions. Next, these characteristics are grouped in a cluster analysis. The clusters represent cylindrical implants and are used to initialize a cylinder-to-image registration. Finally, the two end points are optimized regarding image contrast along the cylinder axis. A total of 67 images containing 420 cylindrical implants were used for testing. Different anatomical regions (calcaneus, spine) and various image sources (two mobile devices, three reconstruction methods) were investigated. Depending on the evaluation set, the detection performance was between 91.7 and 96.1 % true- positive rate with a false-positive rate between 2.0 and 3.2 %. The end point distance errors ranged from $$1.0 \pm 1.2$$ to $$4.3 \pm 2.9$$ mm and the orientation errors from $$1.6 \pm 2.2$$ to $$2.3 \pm 2.2$$ degrees. The average computation time was less than 5 seconds. An automatic method was developed and tested that obviates the need for 3D image interaction during intraoperative assessment of cylindrical orthopedic implants. The required time for working with the viewing software of cone-beam CT device is drastically reduced and leads to a shorter time under anesthesia for the patient.

We present an algorithm for automatic anatomical measurements in tomographic datasets of the knee. The algorithm uses a set of atlases, each consisting of a knee image, surface segmentations of the bones, and locations of landmarks required by the anatomical metrics. A multistage volume-to-volume and surface-to-volume registration is performed to transfer the landmarks from the atlases to the target volume. Manual segmentation of the target volume is not required in this approach. Metrics were computed from the transferred landmarks of a best-matching atlas member (different for each bone), identified based on a mutual information criterion. Leave-one-out validation of the algorithm was performed on 24 scans of the knee obtained using extremity cone-beam computed tomography. Intraclass correlation (ICC) between the algorithm and the expert who generated atlas landmarks was above 0.95 for all metrics. This compares favorably to inter-reader ICC, which varied from 0.19 to 0.95, depending on the metric. Absolute agreement with the expert was also good, with median errors below 0.25 deg for measurements of tibial slope and static alignment, and below 0.2 mm for tibial tuberosity-trochlear groove distance and medial tibial depth. The automatic approach is anticipated to improve measurement workflow and mitigate the effects of operator experience and training on reliability of the metrics.

The assessment of intra-operatively acquired volumetric data is a difficult and often time-consuming task, which demands a new set of skills from the surgeons. In the case of orthopedic surgeries such as the treatment of calcaneal fractures, the correctness of the reduction of the bone fragments can be verified with the help of C-arm CT volumetric images. For an accurate intra-operative assessment of the displaced fragments, an automatic segmentation of the articular surfaces and color-coded visualization was developed. Our automatic approach consists of three major steps: first, using adjusted standard planes intersecting the articular region, the joint space is localized with an intensity profile-based method. In a second step, the localized joint space is segmented on the Laplacian of Gaussian filtered volumetric image by a modified binary flood fill algorithm. Finally, a 3D surface model of the segmented joint space is analyzed and visualized with focus on critical displacements of the surface. A specifically designed human cadaver study consisting of ten lower legs of ten different donors was conducted to acquire 48 realistic C-arm CT images of misaligned bone fragments (steps of varying sizes) in the posterior talar articular surface of the calcaneus. The proposed algorithmic pipeline was verified by the acquired image data and showed very good results with no false positives and an overall correct displacement assessment of $$93.8\,\%$$ . The proposed algorithmic pipeline can be easily integrated into the clinical workflow and qualifies for intra-operative usage. It showed very good results on the reference data set of the cadaver study. With the help of such an assistance system, the time-consuming process of 2D view adjustment and visual assessment of the gray value images can be greatly simplified.

Calcaneal fractures are commonly treated by open reduction and internal fixation. An anatomical reconstruction of involved joints is mandatory to prevent cartilage damage and premature arthritis. In order to avoid intraarticular screw placements, the use of mobile C-arm CT devices is required. However, for analyzing the screw placement in detail, a time-consuming human-computer interaction is necessary to navigate through 3D images and therefore to view a single screw in detail. Established interaction procedures of repeatedly positioning and rotating sectional planes are inconvenient and impede the intraoperative assessment of the screw positioning. To simplify the interaction with 3D images, we propose an automatic screw segmentation that allows for an immediate selection of relevant sectional planes. Our algorithm consists of three major steps. At first, cylindrical characteristics are determined from local gradient structures with the help of RANSAC. In a second step, a DBScan clustering algorithm is applied to group similar cylinder characteristics. Each detected cluster represents a screw, whose determined location is then refined by a cylinder-to-image registration in a third step. Our evaluation with 309 screws in 50 images shows robust and precise results. The algorithm detected 98% (303) of the screws correctly. Thirteen clusters led to falsely identified screws. The mean distance error for the screw tip was 0.8 ± 0.8 mm and for the screw head 1.2 ± 1 mm. The mean orientation error was 1.4 ± 1.2 degrees.

Frakturen am Fersenbein werden mit Hilfe offener Reduktion und interner Fixation korrigiert. Eine anatomisch korrekte Rekonstruktion beteiligter Gelenke ist notwendig, um Knorpelschaden und verfruhte Arthrose vorzubeugen. Um intraartikulare Schraubenplatzierungen zu vermeiden wird der mobile 3D C-Bogen eingesetzt. Die detaillierte Analyse der Schraubenlage anhand des erzeugten 3D Bildes ist jedoch auf eine zeitaufwandige Mensch-Computer-Interaktion angewiesen. Etablierte Interaktionsprozeduren basieren auf wiederholtem Positionieren und Rotieren von Schnittebenen, wodurch die intraoperative Kontrolle der Schraubenplatzierung die Dauer der Operation wesentlich verlangert. Um die Interaktion mit 3D C-Bogen Daten zu erleichtern schlagen wir eine automatische Schraubendetektion vor, mit der eine direkte Anwahl relevanter Schnittebenen moglich wird. Unser Ansatz setzt sich aus zwei Schritten zusammen. Im ersten Schritt werden zylindrische Charakteristiken anhand lokaler Gradientstrukturen mit Hilfe von RANSAC ermittelt. Diese Charakteristiken werden dann durch die Anwendung des DBScan Clustering Algorithmus im zweiten Schritt gruppiert. Jedes detektierte Cluster reprasentiert abschliesend eine Schraube. Unsere Evaluation mit 309 Schrauben in 50 Bildern zeigt robuste Ergebnisse. Der Algorithmus detektierte 97.4% der Schrauben korrekt.

Das Standardvorgehen bei der Behandlung von Calcaneusfrakturen ist eine Osteosynthese. Mit Hilfe der intraoperativen Bildgebung wie dem mobilen C-Bogen CT kann der Chirurg das Repositionsergebnis noch im Operationssaal verifizieren und wenn notig korrigieren. Die Mobilitat des C-Bogen CT hat jedoch zur Folge, dass Informationen uber die Orientierung des Patienten zum Gerat verloren gehen. Dadurch kann keine Standard-Ausrichtung der dreidimensionalen Daten an die Anatomie erfolgen. Eine manuelle Einstellung des Volumendatensatzes durch den Chirurgen ist damit unabdingbar. Dies ist ein zeitaufwendiger Schritt und kann bei einer unprazisen Einstellung zu Fehlern bei der Beurteilung der Daten fuhren. In diesem Paper stellen wir zwei automatische Methoden zur Einstellung der Standard-Ebenen auf mobilen C-Bogen CT Daten vor. Die automatischen Methoden rekonstruieren die Standard-Ebenen in zwei Schritten: als Erstes werden SURF-Keypoints (2D und neu eingefuhrte Pseudo-3D-Punkte) fur das Bildvolumen berechnet, in einem zweiten Schritt wird eine Atlas-Punktwolke auf diese Merkmale registriert und die Parameter der Standard-Ebenen transformiert. Die Genauigkeit unserer Methoden wurde an 51 klinischen mobilen C-Bogen CT Bildern mit manuell eingestellten Standard-Ebenen evaluiert. Die Referenzdaten wurden von drei Chirurgen mit unterschiedlichem Erfahrungsstand erstellt. Die durchschnittlich benotigte Zeit der Experten (46 s) unterscheidet sich von der des fortgeschrittenen Benutzers (55 s) um neun Sekunden. Die Berechnungszeit des 2D-Surf Ansatzes betragt 10 Sekunden und liefert bei 88% der Ebenen der Referenzdaten eine korrkte Einstellung. Der Pseudo-3D Ansatz liefert die besten Ergebnisse mit einer Genauigkeit von 91% und einer Berechnungszeit von nur 8 Sekunden.

Purpose: Extremity cone-beam CT (CBCT) with an amorphous silicon (aSi) flat-panel detector (FPD) provides low-dose volumetric imaging with high spatial resolution. We investigate the performance of the newer complementary metal-oxide semiconductor (CMOS) detectors to enhance resolution of extremities CBCT to ∼0.1 mm, enabling morphological analysis of trabecular bone. Quantitative in-vivo imaging of bone microarchitecture could present an important advance for osteoporosis and osteoarthritis diagnosis and therapy assessment. Methods: Cascaded systems models of CMOS- and FPD-based extremities CBCT were implemented. Performance was compared for a range of pixel sizes (0.05–0.4 mm), focal spot sizes (0.3–0.6 FS), and x-ray techniques (0.05–0.8 mAs/projection) using detectability of high-, low-, and all-frequency tasks for a nonprewhitening observer. Test-bench implementation of CMOS-based extremity CBCT involved a Teledyne DALSA Xineos3030HR detector with 0.099 mm pixels and a compact rotating anode x-ray source with 0.3 FS (IMD RTM37). Metrics of bone morphology obtained using CMOS-based CBCT were compared in cadaveric specimens to FPD-based system using a Varian PaxScan4030 (0.194 mm pixels). Results: Finer pixel size and reduced electronic noise for CMOS (136 e compared to 2000 e for FPD) resulted in ∼1.9× increase in detectability for high-frequency tasks and ∼1.1× increase for all-frequency tasks. Incorporation of the new x-ray source with reduced focal spot size (0.3 FS vs. 0.5 FS used on current extremities CBCT) improved detectability for CMOS-based CBCT by ∼1.7× for high-frequency tasks. Compared to FPD CBCT, the CMOS detector yielded improved agreement with micro-CT in measurements of trabecular thickness (∼1.7× reduction in relative error), bone volume (∼1.5× reduction), and trabecular spacing (∼3.5× reduction). Conclusion: Imaging performance modelling and experimentation indicate substantial improvements for high-frequency imaging tasks through adoption of the CMOS detector and small FS x-ray source, motivating the use of these components in a new system for quantitative in-vivo imaging of trabecular bone. Financial Support: US NIH grant R01EB018896. Qian Cao is a Howard Hughes Medical Institute International Student Research Fellow. Disclosures: W Zbijewski, J Siewerdsen and A Sisniega receive research funding from Carestream Health.