Volumetric Overlap Research Articles

Liver Extraction Using Residual Convolution Neural Networks From Low-Dose CT Images.

An efficient and precise liver extraction from computed tomography (CT) images is a crucial step for computer-aided hepatic diseases diagnosis and treatment. Considering the possible risk to patient's health due to X-ray radiation of repetitive CT examination, low-dose CT (LDCT) is an effective solution for medical imaging. However, inhomogeneous appearances and indistinct boundaries due to additional noise and streaks artifacts in LDCT images often make it a challenging task. This study aims to extract a liver model from LDCT images for facilitating medical expert in surgical planning and post-operative assessment along with low radiation risk to the patient. Our method carried out liver extraction by employing residual convolutional neural networks (LER-CN), which is further refined by noise removal and structure preservation components. After patch-based training, our LER-CN shows a competitive performance relative to state-of-the-art methods for both clinical and publicly available MICCAI Sliver07 datasets. We have proposed training and learning algorithms for LER-CN based on back propagation gradient descent. We have evaluated our method on 150 abdominal CT scans for liver extraction. LER-CN achieves dice similarity coefficient up to 96.5[Formula: see text], decreased volumetric overlap error up to 4.30[Formula: see text], and average symmetric surface distance less than 1.4 [Formula: see text]. These findings have shown that LER-CN is a favorable method for medical applications with high efficiency allowing low radiation risk to patients.

A Model-Based Method for Assessment of Salivary Gland and Planning Target Volume Dosimetry in Volumetric-Modulated Arc Therapy Planning on Head-and-Neck Cancer.

This study examined the relationship of achievable mean dose and percent volumetric overlap of salivary gland with the planning target volume (PTV) in volumetric-modulated arc therapy (VMAT) plan in radiotherapy for a patient with head-and-neck cancer. The aim was to develop a model to predict the viability of planning objectives for both PTV coverage and organs-at-risk (OAR) sparing based on overlap volumes between PTVs and OARs, before the planning process. Forty patients with head-and-neck cancer were selected for this retrospective plan analysis. The patients were treated using 6 MV photons with 2-arc VMAT plan in prescriptions with simultaneous integrated boost in dose of 70 Gy, 63 Gy, and 58.1 Gy to primary tumor sites, high-risk nodal regions, and low-risk nodal regions, respectively, over 35 fractions. A VMAT plan was generated using Varian Eclipse (V13.6), in optimization with biological-based generalized equivalent uniform dose (gEUD) objective for OARs and targets. Target dose coverage (D95, Dmax, conformity index) and salivary gland dose (Dmean and Dmax) were evaluated in those plans. With a range of volume overlaps between salivary glands and PTVs and dose constraints applied, results showed that dose D95 for each PTV was adequate to satisfy D95 >95% of the prescription. Mean dose to parotid <26 Gy could be achieved with <20% volumetric overlap with PTV58 (parotid-PTV58). On an average, the Dmean was seen at 15.6 Gy, 21.1 Gy, and 24.2 Gy for the parotid-PTV58 volume at <5%, <10%, and <20%, respectively. For submandibular glands (SMGs), an average Dmean of 27.6 Gy was achieved in patients having <10% overlap with PTV58, and 36.1 Gy when <20% overlap. Mean doses on parotid and SMG were linearly correlated with overlap volume (regression R2 = 0.95 and 0.98, respectively), which were statistically significant (P < 0.0001). This linear relationship suggests that the assessment of the structural overlap might provide prospective for achievable planning objectives in the head-and-neck plan.

A Model-Based Method for Assessment of OAR and PTV Dosimetry in VMAT Planning on Head-and-Neck Cancer

To develop a model to predict the possibility of achieving planning objectives for both PTV coverage and OAR dose-sparing based on the overlap volumes between PTVs and OARs, prior to the planning process. Forty VMAT plans on head-and-neck cancer patients were analyzed in this study. Patient was treated using 2-arc VMAT plan in prescriptions of 70Gy, 63Gy and 58.1Gy to primary tumor, high-risk nodal regions and low-risk nodal regions in bilateral areas of the neck, respectively, over 35 fractions. A VMAT plan was generated using a commercially available treatment planning system, with the use of the biological optimization objective (gEUD) for OARs and targets during optimization. Plans were evaluated by target dose coverage (D95, Dmax, conformity index) and mean or maximum doses on parotid, submandibular gland, larynx, spinal cord and brain stem. With a range of volume overlaps between OARs and PTVs and dose constraints applied, results showed that each PTV dose (D95) met dose coverage requirement. D95 for PTV70 ranged from 69Gy to 70.1Gy. The conformity index (CI) for PTV70 ranged from 81% to 96.6%. D95 for PTV63 was from 60.7Gy to 62.7Gy. CI was from 79% to 94.4%. Similarly, for PTV58, D95 was from 56.4Gy to 58.2Gy, and CI was from 76.5% to 95%. Dose conformity index at above 70% was acceptable and consistent with other reports. Parotid mean dose (Dmean) and its volumetric overlap with PTV58 was the main focus of the study. The relationship of the mean dose and volume of overlap can be fit in a linear regress formula: Dmean = A ∗ Voverlap + B (cGy) Where Dmean is the mean dose of the OAR, and Voverlap is the percentage volume overlapped with PTV58. A and B are regression constants. On average, the Dmean was seen at 15.6Gy, 21.1Gy and 24.6Gy for the parotid overlap volume of <5%, <10% and <20%, respectively. As expected, parotids dose-sparing became more difficult as the increase of overlap volume. When the overlap exceeded 20%, mean dose of 26Gy or less was no longer achievable. For parotids, the constant A in above is 54.8 and B is 1584.4 with regression R2 = 0.95. For submandibular glands (SMG), an average Dmean 27.6Gy was found in those having <10% overlap with PTV58, and 36.1Gy in the glands with <20% overlap. For SMG, the constant A is 41 and B is 2775.1 with regression R2 = 0.97. The mean doses to larynx (average Dmean =26.6Gy), and maximum doses to spinal cord (average Dmax=34.9Gy) and brain stem (average Dmax=31.5Gy) were also examined in those plans. The results demonstrate a strong relationship between the mean dose to parotid and SMG and its volume overlapped with the PTV. The linear model of the regression may serve as the predicting factors for early determinations of achievable dosimetric outcomes for these plans.

Automatic lung segmentation for the inclusion of juxtapleural nodules and pulmonary vessels using curvature based border correction

Lung field segmentation is a significant step in the development of a Computer-Aided Diagnosis system and efficiently used for the quantitative analysis of lung CT images. However, this task is complicated especially in the presence of juxtapleural nodules, pulmonary vessels. In this paper, a more robust and accurate approach for lung segmentation is proposed. Our algorithm determines the dominant points indicating the concave and convex region along the lung boundary and these dominant points are connected by applying dominant point marching algorithm. The lung segmentation algorithm was tested on 36 subjects from the Lung Imaging Database Consortium. Results indicate that our proposed method can successfully re-include juxtapleural nodules and pulmonary vessels and achieves the average volumetric overlap fraction of 96.97%, average under segmentation rate of 2.324% and average over segmentation rate of 0.79%. The average processing time required for each subject is 182.448 s and for each slice is 0.953 s, which is comparatively faster than the manual segmentation done by radiologists. Experimental results suggest that our proposed method is faster and more robust than other state of the art methods. Hence proposed method can be used as an efficient tool for the lung segmentation in clinical practice.

Robust extraction for low-contrast liver tumors using modified adaptive likelihood estimation.

Liver tumor extraction is essential for liver ablation surgery planning and treatment. For accurate and robust tumor segmentation, we propose a semiautomatic method using adaptive likelihood classification with modified likelihood model. First, a minimal ellipse (or quasi-ellipsoid) that encloses a liver tumor is generated for initialization. Then, a hybrid intensity likelihood modification based on nonparametric density estimation is proposed to enhance local likelihood contrast and reduce its inhomogeneity. A prior elliptical (or quasi-ellipsoid) shape constraint is directly integrated into the likelihood to further prevent leakage of the algorithm into adjacent tissues with similar intensity. Finally, an adaptive likelihood classification is proposed for accurate segmentation of tumors with low contrast, high noise or heterogeneous densities. Experiments were performed on 3Dircadb and LiTS datasets. The average volumetric overlap errors of the 3Dircadb and LiTS datasets were 27.05 and 35.72%, respectively. The algorithm's robustness was validated by comparing results of 5 operators with multiple selections on different tumors. The proposed method achieved good results in different tumors, even in low-contrast tumors with blurred boundaries. Reliable results can still be achieved over different initializations by different operators using the proposed method.

Deteksi Tumor Hati dengan Graph Cut dan Taksiran Volume Tumornya

Liver is one of the most important organs in the human body. One of the dangerous diseases of the liver is tumor. In the CT scan image, the tumor has different texture, color, shape, and position, according to patient's condition. In this study, a tumor detection was carried out by tree stages: firstly some steps of preprocessing, such as filtering, edge detection, and erotion; secondly, finding the liver among organs in abdomen using segmentation and checking the liver position in the right abdomen; and thirdly performing the tumor detection in the liver using graph cut and push relabel algorithm. Usually, segmentation using graph cut needs two interactive inputs, namely sample of object area and sample of background area. In this paper, the interactive inputs on graph cut were replaced by deviation standard calculation. Testing using three sets of CT image and the ground truth produces average of the dice similarity coefficient (DSC), volumetric overlap error (VOE), and absolute volume difference (AVD) parameters of 78.15%, 25.72%, 19.30%, respectively. Furthermore, volume of liver tumor is approximated by utilizing area of tumor in each slice of CT image, then displayed in 3D view.

Fully automatic liver segmentation in CT images using modified graph cuts and feature detection

PurposeLiver segmentation from CT images is a fundamental step in trajectory planning for computer-assisted interventional surgery. In this paper, we present a fully automatic procedure using modified graph cuts and feature detection for accurate and fast liver segmentation. MethodsThe initial slice and seeds of graph cuts are automatically determined using an intensity-based method with prior position information. A contrast term based on the similarities and differences of local organs across multi-slices is proposed to enhance the weak boundaries of soft organs and to prevent over-segmentation. The term is then integrated into the graph cuts for automatic slice segmentation. Patient-specific intensity and shape constraints of neighboring slices are also used to prevent leakage. Finally, a feature detection method based on vessel anatomical information is proposed to eliminate the adjacent inferior vena cava with similar intensities. ResultsWe performed experiments on 20 Sliver07, 20 3Dircadb datasets and local clinical datasets. The average volumetric overlap error, volume difference, symmetric surface distance and volume processing time were 5.3%, −0.6%, 1.0 mm, 17.8 s for the Sliver07 dataset and 8.6%, 0.7%, 1.6 mm, 12.7 s for the 3Dircadb dataset, respectively. ConclusionsThe proposed method can effectively extract the liver from low contrast and complex backgrounds without training samples. It is fully automatic, accurate and fast for liver segmentation in clinical settings.

A variational approach to liver segmentation using statistics from multiple sources

Medical image segmentation plays an important role in digital medical research, and therapy planning and delivery. However, the presence of noise and low contrast renders automatic liver segmentation an extremely challenging task. In this study, we focus on a variational approach to liver segmentation in computed tomography scan volumes in a semiautomatic and slice-by-slice manner. In this method, one slice is selected and its connected component liver region is determined manually to initialize the subsequent automatic segmentation process. From this guiding slice, we execute the proposed method downward to the last one and upward to the first one, respectively. A segmentation energy function is proposed by combining the statistical shape prior, global Gaussian intensity analysis, and enforced local statistical feature under the level set framework. During segmentation, the shape of the liver shape is estimated by minimization of this function. The improved Chan–Vese model is used to refine the shape to capture the long and narrow regions of the liver. The proposed method was verified on two independent public databases, the 3D-IRCADb and the SLIVER07. Among all the tested methods, our method yielded the best volumetric overlap error (VOE) of , the best root mean square symmetric surface distance (RMSD) of mm, the best maximum symmetric surface distance (MSD) of mm in 3D-IRCADb dataset, and the best average symmetric surface distance (ASD) of mm, the best RMSD of mm in SLIVER07 dataset, respectively. The results of the quantitative comparison show that the proposed liver segmentation method achieves competitive segmentation performance with state-of-the-art techniques.

Segmentation of liver and vessels from CT images and classification of liver segments for preoperative liver surgical planning in living donor liver transplantation

Background and objectiveThe present study developed an effective surgical planning method consisting of a liver extraction stage, a vessel extraction stage, and a liver segment classification stage based on abdominal computerized tomography (CT) images. MethodsAn automatic seed point identification method, customized level set methods, and an automated thresholding method were applied in this study to extraction of the liver, portal vein (PV), and hepatic vein (HV) from CT images. Then, a semi-automatic method was developed to separate PV and HV. Lastly, a local searching method was proposed for identification of PV branches and the nearest neighbor approximation method was applied to classifying liver segments. ResultsOnsite evaluation of liver segmentation provided by the SLIVER07 website showed that the liver segmentation method achieved an average volumetric overlap accuracy of 95.2%. An expert radiologist evaluation of vessel segmentation showed no false positive errors or misconnections between PV and HV in the extracted vessel trees. Clinical evaluation of liver segment classification using 43 CT datasets from two medical centers showed that the proposed method achieved high accuracy in liver graft volumetry (absolute error, AE = 45.2 ± 20.9 ml; percentage of AE, %AE = 6.8% ± 3.2%; percentage of %AE > 10% = 16.3%; percentage of %AE > 20% = none) and the classified segment boundaries agreed with the intraoperative surgical cutting boundaries by visual inspection. ConclusionsThe method in this study is effective in segmentation of liver and vessels and classification of liver segments and can be applied to preoperative liver surgical planning in living donor liver transplantation.

Application of 3D liver microtissues for assessment of drug-induced liver injury (DILI) and for studying liver steatosis

We sought to evaluate a new regional segmentation method for use with three-dimensional (3D) non-contrast abdominal CT images and to report the preliminary results.The proposed method was evaluated in ten cases. Manually segmented areas were used as the gold standard for evaluation. To compare the standard and the extracted liver regions, the degree of coincidence R% was redefined by transforming a volumetric overlap error. We also evaluated the influence of varying the density window size in terms of setting the starting points.We confirmed in ten cases that our method could segment the liver region more precisely than the conventional method. A size of window 15 voxels was optimal as the starting point in all cases.We demonstrated the accuracy of a 3D semiautomatic liver segmentation method for non-contrast CT. This method promises to offer radiologists a time-efficient segmentation aid.

Development of the breast immobilization system in prone setup: The effect of bra in prone position to improve the breast setup error.

Purpose/objective(s)Accurate and reproducible positioning of the breast is difficult due to its deformability and softness; thus, targeting a breast tumor or tumor bed with fractionated radiotherapy using external beam radiation is difficult. The aim of this study was to develop a novel bra to aid in breast immobilization in the prone position.Materials & methodsTo assess the accuracy of prone position fixation of breast tumors, 33 breast cancer patients with 34 lesions were recruited. The bra used in this verification was customized from a commercially available bra. Duplicate MRI were acquired in the prone position, alternating with and without the bra, and for each series, patients were asked to step off the MRI table and re‐set up in the prone position. Patients were also asked to remove and re‐fit the bra for the second MRI. Each pair of images were superimposed to match the shape of the skin surface, and the maximum difference in tumor geometric center in three axes was measured. The required set up margin was calculated as: required margin = mean difference in geometric center + 2.5 standard deviation. The volumetric overlap of the tumor, as well as contouring uncertainties, was evaluated using contour analysis software.ResultsThe median breast size was 498 cc. The required margins for the lateral, vertical, and longitudinal directions were estimated to be 4.1, 4.1, and 5.0 mm, respectively, with the bra, and 5.1, 6.9, and 6.7 mm, respectively, without the bra. These margins covered the dislocation of more than 33 lesions in total. With the bra, 33 lesions had achieved an objective overlap of 95% and 99% with 2 and 4 mm margins, respectively, whereas 4 and 8 mm, respectively, were needed without the bra.ConclusionThe use of an immobilizing bra reduced the setup margin for prone position fixation of breast tumors.

Global Feature Tracking and Similarity Estimation in Time‐Dependent Scalar Fields

AbstractWe present an algorithm for tracking regions in time‐dependent scalar fields that uses global knowledge from all time steps for determining the tracks. The regions are defined using merge trees, thereby representing a hierarchical segmentation of the data in each time step. The similarity of regions of two consecutive time steps is measured using their volumetric overlap and a histogram difference. The main ingredient of our method is a directed acyclic graph that records all relevant similarity information as follows: the regions of all time steps are the nodes of the graph, the edges represent possible short feature tracks between consecutive time steps, and the edge weights are given by the similarity of the connected regions. We compute a feature track as the global solution of a shortest path problem in the graph. We use these results to steer the – to the best of our knowledge – first algorithm for spatio‐temporal feature similarity estimation. Our algorithm works for 2D and 3D time‐dependent scalar fields. We compare our results to previous work, showcase its robustness to noise, and exemplify its utility using several real‐world data sets.

DeepSite: protein-binding site predictor using 3D-convolutional neural networks.

An important step in structure-based drug design consists in the prediction of druggable binding sites. Several algorithms for detecting binding cavities, those likely to bind to a small drug compound, have been developed over the years by clever exploitation of geometric, chemical and evolutionary features of the protein. Here we present a novel knowledge-based approach that uses state-of-the-art convolutional neural networks, where the algorithm is learned by examples. In total, 7622 proteins from the scPDB database of binding sites have been evaluated using both a distance and a volumetric overlap approach. Our machine-learning based method demonstrates superior performance to two other competitive algorithmic strategies. DeepSite is freely available at www.playmolecule.org. Users can submit either a PDB ID or PDB file for pocket detection to our NVIDIA GPU-equipped servers through a WebGL graphical interface. gianni.defabritiis@upf.edu. Supplementary data are available at Bioinformatics online.

Automated Ischemic Lesion Segmentation in MRI Mouse Brain Data after Transient Middle Cerebral Artery Occlusion.

Magnetic resonance imaging (MRI) has become increasingly important in ischemic stroke experiments in mice, especially because it enables longitudinal studies. Still, quantitative analysis of MRI data remains challenging mainly because segmentation of mouse brain lesions in MRI data heavily relies on time-consuming manual tracing and thresholding techniques. Therefore, in the present study, a fully automated approach was developed to analyze longitudinal MRI data for quantification of ischemic lesion volume progression in the mouse brain. We present a level-set-based lesion segmentation algorithm that is built using a minimal set of assumptions and requires only one MRI sequence (T2) as input. To validate our algorithm we used a heterogeneous data set consisting of 121 mouse brain scans of various age groups and time points after infarct induction and obtained using different MRI hardware and acquisition parameters. We evaluated the volumetric accuracy and regional overlap of ischemic lesions segmented by our automated method against the ground truth obtained in a semi-automated fashion that includes a highly time-consuming manual correction step. Our method shows good agreement with human observations and is accurate on heterogeneous data, whilst requiring much shorter average execution time. The algorithm developed here was compiled into a toolbox and made publically available, as well as all the data sets.

Patterns of failure for patients with glioblastoma following O-(2-[18F]fluoroethyl)-L-tyrosine PET- and MRI-guided radiotherapy

Background and purposeTo evaluate the patterns of failure following clinical introduction of amino-acid O-(2-[18F]fluoroethyl)-L-tyrosine (FET)-PET-guided target definition for radiotherapy (RT) of glioblastoma patients. Materials and methodsThe first 66 consecutive patients with confirmed histology, scanned using FET-PET/CT and MRI were selected for evaluation. Chemo-radiotherapy was delivered to a volume based on both MRI and FET-PET (PETvol). The volume of recurrence (RV) was defined on MRI data collected at the time of progression according to RANO criteria. ResultsFifty patients were evaluable, with median follow-up of 45months. Central, in-field, marginal and distant recurrences were observed for 82%, 10%, 2%, and 6% of the patients, respectively. We found a volumetric overlap of 26%, 31% and 39% of the RV with the contrast-enhancing MR volume, PETvol and the composite MRPETvol, respectively. MGMT-methylation (p=0.03), larger PETvol (p<0.001), and less extensive surgery (p<0.001), were associated with larger PETvol overlap. ConclusionThe combined MRPETvol had a stronger association with the recurrence volume than either of the modalities alone. Larger overlap of PETvol and RV was observed for patients with MGMT-methylation, less extensive surgery, and large PETvol on the RT-planning scans.

3D Liver Tumor Segmentation in CT Images Using Improved Fuzzy C-Means and Graph Cuts.

Three-dimensional (3D) liver tumor segmentation from Computed Tomography (CT) images is a prerequisite for computer-aided diagnosis, treatment planning, and monitoring of liver cancer. Despite many years of research, 3D liver tumor segmentation remains a challenging task. In this paper, an efficient semiautomatic method was proposed for liver tumor segmentation in CT volumes based on improved fuzzy C-means (FCM) and graph cuts. With a single seed point, the tumor volume of interest (VOI) was extracted using confidence connected region growing algorithm to reduce computational cost. Then, initial foreground/background regions were labeled automatically, and a kernelized FCM with spatial information was incorporated in graph cuts segmentation to increase segmentation accuracy. The proposed method was evaluated on the public clinical dataset (3Dircadb), which included 15 CT volumes consisting of various sizes of liver tumors. We achieved an average volumetric overlap error (VOE) of 29.04% and Dice similarity coefficient (DICE) of 0.83, with an average processing time of 45 s per tumor. The experimental results showed that the proposed method was accurate for 3D liver tumor segmentation with a reduction of processing time.

Automatic 3D liver location and segmentation via convolutional neural network and graph cut.

Segmentation of the liver from abdominal computed tomography (CT) images is an essential step in some computer-assisted clinical interventions, such as surgery planning for living donor liver transplant, radiotherapy and volume measurement. In this work, we develop a deep learning algorithm with graph cut refinement to automatically segment the liver in CT scans. The proposed method consists of two main steps: (i) simultaneously liver detection and probabilistic segmentation using 3D convolutional neural network; (ii) accuracy refinement of the initial segmentation with graph cut and the previously learned probability map. The proposed approach was validated on forty CT volumes taken from two public databases MICCAI-Sliver07 and 3Dircadb1. For the MICCAI-Sliver07 test dataset, the calculated mean ratios of volumetric overlap error (VOE), relative volume difference (RVD), average symmetric surface distance (ASD), root-mean-square symmetric surface distance (RMSD) and maximum symmetric surface distance (MSD) are 5.9, 2.7%, 0.91, 1.88 and 18.94mm, respectively. For the 3Dircadb1 dataset, the calculated mean ratios of VOE, RVD, ASD, RMSD and MSD are 9.36, 0.97%, 1.89, 4.15 and 33.14mm, respectively. The proposed method is fully automatic without any user interaction. Quantitative results reveal that the proposed approach is efficient and accurate for hepatic volume estimation in a clinical setup. The high correlation between the automatic and manual references shows that the proposed method can be good enough to replace the time-consuming and nonreproducible manual segmentation method.

Comparative evaluation of respiratory-gated and ungated FDG-PET for target volume definition in radiotherapy treatment planning for pancreatic cancer

ObjectiveThe purpose of this study was to evaluate the usefulness of respiratory-gated positron emission tomography (4D-PET) in pancreatic cancer radiotherapy treatment planning (RTTP). Materials and methodsFourteen patients with 18F-fluorodeoxyglucose (FDG)-avid pancreatic tumours were evaluated between December 2013 and March 2015. Two sets of volumes were contoured for the pancreatic tumour of each patient. The biological target volume in three-dimensional RTTP (BTV3D) was contoured using conventional respiratory un-gated PET. The BTV3D was then expanded using population-based margins to generate a series of internal target volume 3D (ITV3D) values. The ITV 4D (ITV4D) was contoured using 4D-PET. Each of the five phases of 4D-PET was used for 4D contouring, and the ITV4D was constructed by summing the volumes defined on the five individual 4D-PET images. The relative volumes and normalized volumetric overlap were computed between ITV3D and ITV4D. ResultsOn average, the FDG-avid tumour volumes were 1.6 (range: 0.8–2.3) fold greater in the ITV4D than in the BTV3D. On average, the ITV3D values were 2.0 (range: 1.1–3.4) fold larger than the corresponding ITV4D values. ConclusionThe ITV generated from 4D-PET can be used to improve the accuracy or reduce normal tissue irradiation compared with conventional un-gated PET-based ITV.

SU‐C‐202‐02: A Comprehensive Evaluation of Adaptive Daily Planning for Cervical Cancer HDR Brachytherapy

Purpose:To evaluate adaptive daily planning for cervical cancer patients who underwent high‐dose‐rate intra‐cavitary brachytherapy (HDR‐ICBT).Methods:This study included 22 cervical cancer patients who underwent 5 fractions of HDR ICBT. Regions of interest (ROIs) including high‐risk clinical tumor volume (HR‐CTV) and organs‐at‐risk (OARs) were manually contoured on daily CT images. All patients were treated with adaptive daily plans, which involved ROI delineation and dose optimization at each treatment fraction. Single treatment plans were retrospectively generated by applying the first treatment fraction's dwell times adjusted for decay and dwell positions of the applicator to subsequent treatment fractions. Various existing similarity metrics were calculated for the ROIs to quantify interfractional organ variations. A novel similarity score (JRARM) was established, which combined both volumetric overlap metrics (DSC, JSC, and RVD) and distance metrics (ASD, MSD, and RMSD). Linear regression was performed to determine a relationship between inter‐fractional organ variations of various similarity metrics and D2cc variations from both plans. Wilcoxon Signed Rank Tests were used to assess adaptive daily plans and single plans by comparing EQD2 D2cc (α/β=3) for OARs.Results:For inter‐fractional organ variations, the sigmoid demonstrated the greatest variations based on the JRARM and DSC similarity metrics. Comparisons between paired ROIs showed differences in JRARM scores and DSCs at each treatment fraction. RVD, MSD, and RMSD were found to be significantly correlated to D2cc variations for bladder and sigmoid. The comparison between plans found that adaptive daily planning provided lower EQD2 D2cc of OARs than single planning, specifically for the sigmoid (p=0.015).Conclusion:Substantial inter‐fractional organ motion can occur during HDR‐BT, which may significantly affect D2cc of OARs. Adaptive daily planning provides improved dose sparing for OARs compared to single planning.

Trade-off between angular and spatial resolutions in in vivo fiber tractography

Tractography is becoming an increasingly popular method to reconstruct white matter connections in vivo. The diffusion MRI data that tractography is based on requires a high angular resolution to resolve crossing fibers whereas high spatial resolution is required to distinguish kissing from crossing fibers. However, scan time increases with increasing spatial and angular resolutions, which can become infeasible in clinical settings. Here we investigated the trade-off between spatial and angular resolutions to determine which of these factors is most worth investing scan time in. We created a unique diffusion MRI dataset with 1.0mm isotropic resolution and a high angular resolution (100 directions) using an advanced 3D diffusion-weighted multi-slab EPI acquisition. This dataset was reconstructed to create subsets of lower angular (75, 50, and 25 directions) and lower spatial (1.5, 2.0, and 2.5mm) resolution. Using all subsets, we investigated the effects of angular and spatial resolutions in three fiber bundles—the corticospinal tract, arcuate fasciculus and corpus callosum—by analyzing the volumetric bundle overlap and anatomical correspondence between tracts. Our results indicate that the subsets of 25 and 50 directions provided inferior tract reconstructions compared with the datasets with 75 and 100 directions. Datasets with spatial resolutions of 1.0, 1.5, and 2.0mm were comparable, while the lowest resolution (2.5mm) datasets had discernible inferior quality. In conclusion, we found that angular resolution appeared to be more influential than spatial resolution in improving tractography results. Spatial resolutions higher than 2.0mm only appear to benefit multi-fiber tractography methods if this is not at the cost of decreased angular resolution.