Tumor Contour Research Articles

Can Augmented Reality Be Helpful in Pelvic Bone Cancer Surgery? An In Vitro Study.

Application of surgical navigation for pelvic bone cancer surgery may prove useful, but in addition to the fact that research supporting its adoption remains relatively preliminary, the actual navigation devices are physically large, occupying considerable space in already crowded operating rooms. To address this issue, we developed and tested a navigation system for pelvic bone cancer surgery assimilating augmented reality (AR) technology to simplify the system by embedding the navigation software into a tablet personal computer (PC). Using simulated tumors and resections in a pig pelvic model, we asked: Can AR-assisted resection reduce errors in terms of planned bone cuts and improve ability to achieve the planned margin around a tumor in pelvic bone cancer surgery? We developed an AR-based navigation system for pelvic bone tumor surgery, which could be operated on a tablet PC. We created 36 bone tumor models for simulation of tumor resection in pig pelves and assigned 18 each to the AR-assisted resection group and conventional resection group. To simulate a bone tumor, bone cement was inserted into the acetabular dome of the pig pelvis. Tumor resection was simulated in two scenarios. The first was AR-assisted resection by an orthopaedic resident and the second was resection using conventional methods by an orthopaedic oncologist. For both groups, resection was planned with a 1-cm safety margin around the bone cement. Resection margins were evaluated by an independent orthopaedic surgeon who was blinded as to the type of resection. All specimens were sectioned twice: first through a plane parallel to the medial wall of the acetabulum and second through a plane perpendicular to the first. The distance from the resection margin to the bone cement was measured at four different locations for each plane. The largest of the four errors on a plane was adopted for evaluation. Therefore, each specimen had two values of error, which were collected from two perpendicular planes. The resection errors were classified into four grades: ≤ 3 mm; 3 to 6 mm; 6 to 9 mm; and > 9 mm or any tumor violation. Student's t-test was used for statistical comparison of the mean resection errors of the two groups. The mean of 36 resection errors of 18 pelves in the AR-assisted resection group was 1.59 mm (SD, 4.13 mm; 95% confidence interval [CI], 0.24-2.94 mm) and the mean error of the conventional resection group was 4.55 mm (SD, 9.7 mm; 95% CI, 1.38-7.72 mm; p < 0.001). All specimens in the AR-assisted resection group had errors < 6 mm, whereas 78% (28 of 36) of errors in the conventional group were < 6 mm. In this in vitro simulated tumor model, we demonstrated that AR assistance could help to achieve the planned margin. Our model was designed as a proof of concept; although our findings do not justify a clinical trial in humans, they do support continued investigation of this system in a live animal model, which will be our next experiment. The AR-based navigation system provides additional information of the tumor extent and may help surgeons during pelvic bone cancer surgery without the need for more complex and cumbersome conventional navigation systems.

Detection of Lung Contour with Closed Principal Curve and Machine Learning.

Radiation therapy plays an essential role in the treatment of cancer. In radiation therapy, the ideal radiation doses are delivered to the observed tumor while not affecting neighboring normal tissues. In three-dimensional computed tomography (3D-CT) scans, the contours of tumors and organs-at-risk (OARs) are often manually delineated by radiologists. The task is complicated and time-consuming, and the manually delineated results will be variable from different radiologists. We propose a semi-supervised contour detection algorithm, which firstly uses a few points of region of interest (ROI) as an approximate initialization. Then the data sequences are achieved by the closed polygonal line (CPL) algorithm, where the data sequences consist of the ordered projection indexes and the corresponding initial points. Finally, the smooth lung contour can be obtained, when the data sequences are trained by the backpropagation neural network model (BNNM). We use the private clinical dataset and the public Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) dataset to measure the accuracy of the presented method, respectively. To the private dataset, experimental results on the initial points which are as low as 15% of the manually delineated points show that the Dice coefficient reaches up to 0.95 and the global error is as low as 1.47 × 10-2. The performance of the proposed algorithm is also better than the cubic spline interpolation (CSI) algorithm. While on the public LIDC-IDRI dataset, our method achieves superior segmentation performance with average Dice of 0.83.

Patient-specific model-based segmentation of brain tumors in 3D intraoperative ultrasound images.

Intraoperative ultrasound (iUS) imaging is commonly used to support brain tumor operation. The tumor segmentation in the iUS images is a difficult task and still under improvement because of the low signal-to-noise ratio. The success of automatic methods is also limited due to the high noise sensibility. Therefore, an alternative brain tumor segmentation method in 3D-iUS data using a tumor model obtained from magnetic resonance (MR) data for local MR-iUS registration is presented in this paper. The aim is to enhance the visualization of the brain tumor contours in iUS. A multistep approach is proposed. First, a region of interest (ROI) based on the specific patient tumor model is defined. Second, hyperechogenic structures, mainly tumor tissues, are extracted from the ROI of both modalities by using automatic thresholding techniques. Third, the registration is performed over the extracted binary sub-volumes using a similarity measure based on gradient values, and rigid and affine transformations. Finally, the tumor model is aligned with the 3D-iUS data, and its contours are represented. Experiments were successfully conducted on a dataset of 33 patients. The method was evaluated by comparing the tumor segmentation with expert manual delineations using two binary metrics: contour mean distance and Dice index. The proposed segmentation method using local and binary registration was compared with two grayscale-based approaches. The outcomes showed that our approach reached better results in terms of computational time and accuracy than the comparative methods. The proposed approach requires limited interaction and reduced computation time, making it relevant for intraoperative use. Experimental results and evaluations were performed offline. The developed tool could be useful for brain tumor resection supporting neurosurgeons to improve tumor border visualization in the iUS volumes.

Quantitative Contour Analysis as an Image-based Discriminator Between Benign and Malignant Renal Tumors

To investigate whether morphologic analysis can differentiate between benign and malignant renal tumors on clinically acquired imaging. Between 2009 and 2014, 3-dimensional tumor volumes were manually segmented from contrast-enhanced computerized tomography (CT) images from 150 patients with predominantly solid, nonmacroscopic fat-containing renal tumors: 100 renal cell carcinomas and 50 benign lesions (eg, oncocytoma and lipid-poor angiomyolipoma). Tessellated 3-dimensional tumor models were created from segmented voxels using MATLAB code. Eleven shape descriptors were calculated: sphericity, compactness, mean radial distance, standard deviation of the radial distance, radial distance area ratio, zero crossing, entropy, Feret ratio, convex hull area and convex hull perimeter ratios, and elliptic compactness. Morphometric parameters were compared using the Wilcoxon rank-sum test to investigate whether malignant renal masses demonstrate more morphologic irregularity than benign ones. Only CHP in sagittal orientation (median 0.96 vs 0.97) and EC in coronal orientation (median 0.92 vs 0.93) differed significantly between malignant and benign masses (P = .04). When comparing these 2 metrics between coronal and sagittal orientations, similar but nonsignificant trends emerged (P = .07). Other metrics tested were not significantly different in any imaging plane. Computerized image analysis is feasible using shape descriptors that otherwise cannot be visually assessed and used without quantification. Shape analysis via the transverse orientation may be reasonable, but encompassing all 3 planar dimensions to characterize tumor contour can achieve a more comprehensive evaluation. Two shape metrics (CHP and EC) may help distinguish benign from malignant renal tumors, an often challenging goal to achieve on imaging and biopsy.

Automated breast tumor detection and segmentation with a novel computational framework of whole ultrasound images.

Due to the low contrast and ambiguous boundaries of the tumors in breast ultrasound (BUS) images, it is still a challenging task to automatically segment the breast tumors from the ultrasound. In this paper, we proposed a novel computational framework that can detect and segment breast lesions fully automatic in the whole ultrasound images. This framework includes several key components: pre-processing, contour initialization, and tumor segmentation. In the pre-processing step, we applied non-local low-rank (NLLR) filter to reduce the speckle noise. In contour initialization step, we cascaded a two-step Otsu-based adaptive thresholding (OBAT) algorithm with morphologic operations to effectively locate the tumor regions and initialize the tumor contours. Finally, given the initial tumor contours, the improved Chan-Vese model based on the ratio of exponentially weighted averages (CV-ROEWA) method was utilized. This pipeline was tested on a set of 61 breast ultrasound (BUS) images with diagnosed tumors. The experimental results in clinical ultrasound images prove the high accuracy and robustness of the proposed framework, indicating its potential applications in clinical practice. Graphical abstract ᅟ.

Adaptive kernelized evidential clustering for automatic 3D tumor segmentation in FDG–PET images

Automatically and reliably delineating tumor contours in noisy and blurring PET images is a challenging work in clinical oncology. In this paper, we introduce a specific unsupervised learning method to this end. More specifically, a robust clustering algorithm with spatial knowledge enhancement is developed in the framework of belief functions, a formal and powerful tool for modeling and reasoning with uncertain and/or imprecise information. Diverse patch-based image features are extracted to comprehensively describe PET image voxels. Then, informative input features are iteratively selected to learn an adaptive kernel-induced metric in an unsupervised way, so as to precisely grouping voxels into different clusters. The effectiveness of the proposed method has been evaluated on FDG–PET images for lung tumor patients.

Корреляция между радиотерапевтической дозой на лимфоидные опухоли средостения и дозой на сердечно-сосудистую систему при использовании системы контроля за дыханием пациента

Purpose: To find a correlation between prescribed dose on the irradiation area and dose on heart and ascending aorta, using the "breath holding" method in classical radiotherapy course for patients with Hodgkin and non-Hodgkin mediastinal lymphomas.&#x0D; Material and methods: For patients of reproductive age with diagnose Hodgkin, non-Hodgkin mediastinal lymphomas we conducted radiotherapy with contouring of tumor, and critical structures, include heart and ascending aorta. Radiotherapy was conducted on breath holding (ABC) for shielding radiation dose from heart, ascending aorta and lungs. We did calculations of length and diameter of the contoured aorta and dose that it and heart took. Also, we did correlation calculation of relationship between prescribed dose on the tumor and doses cover to the heart and aorta on 6 and 10 MV nominal energies.&#x0D; Results: According to the protocol RTOG 1005 for the heart zone, exceeding the threshold level for V20 &lt; 5 % occurred in 11 cases out 21, and 8 cases from them are for 10 MV energy. For V10 &lt; 30 % it happened in nine cases from 21, and seven cases from them are for 10 MV energy too. The correlation coefficient between the prescribed dose and the received dose for protocol RTOG 1005 V10 &lt; 30 % was 0.71 and it showed the highest value. This index is lower on 13 % for protocol RTOG 0623 (V67) – 0..58, and it has the weakest coefficient correlation (V33) – 0.45. The value of the correlation coefficient for the aorta decreased with an increase in the volume to which the prescribed dose of the irradiation area falls and, consequently a 10 and 20 % of the aorta volume is getting a maximum value from the prescribed dose.&#x0D; Conclusion: Our research showed what contouring of ascending aorta is obligatorily procedure, because the average dose on it was 10.73 Gy for 10 MV and 6.50 Gy for 6 MV energies. It is obvious that using 6 MV energy is more organ-preserving method of thea radiotherapy treatment when using a such techniques as ABC. It is a very important to study the issue of the permissible dose per volume of ascending aorta without critical consequences for cardiovascular system. In addition, it particularly important if we take into account that all patients had a chemotherapy course before radiation therapy which also caused cardiotoxicity.

Interobserver variability in tumor contouring affects the use of radiomics to predict mutational status.

Radiomic features characterize tumor imaging phenotype. Nonsmall cell lung cancer (NSCLC) tumors are known for their complexity in shape and wide range in density. We explored the effects of variable tumor contouring on the prediction of epidermal growth factor receptor (EGFR) mutation status by radiomics in NSCLC patients treated with a targeted therapy (Gefitinib). Forty-six early stage NSCLC patients (EGFR mutant:wildtype = 20:26) were included. Three experienced radiologists independently delineated the tumors using a semiautomated segmentation software on a noncontrast-enhanced baseline and three-week post-therapy CT scan images that were reconstructed using 1.25-mm slice thickness and lung kernel. Eighty-nine radiomic features were computed on both scans and their changes (radiomic delta-features) were calculated. The highest area under the curves (AUCs) were 0.87, 0.85, and 0.80 for the three radiologists and the number of significant features ([Formula: see text]) was 3, 5, and 0, respectively. The AUCs of a single feature significantly varied among radiologists (e.g., 0.88, 0.75, and 0.73 for run-length primitive length uniformity). We conclude that a three-week change in tumor imaging phenotype allows identifying the EGFR mutational status of NSCLC. However, interobserver variability in tumor contouring translates into a significant variability in radiomic metrics accuracy.

Computer-Aided tumor diagnosis in 3-D breast elastography

Background and objectiveBreast cancer is the major cause of cancer-related mortality in women. However, the death rate can be effectively decreased if the breast cancer can be detected early and treated appropriately. In recent years, many studies have indicated that the elastography has the better diagnosis performance than conventional ultrasound (US). MethodIn this study, the 3-D tumor contour is obtained by using the proposed segmentation methods and then the features containing texture information, shape information, ellipsoid fitting information are extracted respectively by using the segmented 3-D tumor contour and B-mode images, and the features containing elasticity information are calculated using the same contour and elastographic images. ResultsIn this experiment, totally 40 biopsy-proved lesions containing 20 benign tumors and 20 malignant tumors are used to evaluate the proposed computer-aided diagnosis (CAD) system. From the experimental results, the combination of shape, ellipsoid fitting and elastographic features has the best performance with accuracy 90.50% (36/40), sensitivity 85.00% (17/20), specificity 95.00% (19/20), and the area under the ROC curve Az 0.987. ConclusionThe result shows that tumors can be diagnosed more precisely by using the elastography images.

Quantification of Volumetric Changes Occurring During Radiation Therapy for Esophageal Cancer

Many patients undergoing radiation therapy for esophageal cancer have marked anatomic changes during their course of treatment including shrinking of the primary tumor, edematous changes of the treated esophagus, and changes in body habitus. Changes in contour, volume, and location of the esophageal tumor over the total course of radiation therapy could have a potential dosimetric impact if highly conformal treatment techniques are used. We conducted a retrospective repeated measures analysis on daily cone-beam CT imaging to quantify the magnitude of these volumetric changes. Utilizing an institutional retrospective database, fourteen patients were identified for this study. All patients were treated using a linear accelerator with integrated on-board cone beam CT imaging that allows CT imaging during the daily treatment sessions while the patient is immobilized in the treatment position. Gross tumor volumes of the primary tumor were contoured on every axial slice on the primary planning CT and the CT acquired on treatment days 1, 6,11,16,21, and 25. The changes in primary tumor volume over time were analyzed with a repeated measures ANOVA analysis. Gross tumor volumes generally decreased at an average rate of roughly 6% volume loss per week which ultimately resulted in a 23% total average reduction in volume over the total course of therapy. Analysis of variance revealed a statistically significant reduction in tumor volume occurring by cone beam CT 11 with a mean volume of 91% of original (p=0.0011 95% CI 0.89-0.98).The volumes measured on cone beam CTs taken on treatment days 16, 21, and 25 were also statistically significant (p<0.0001) with mean residual volumes of 89% (95% CI = 0.83-0.96), 83% (95% CI=0.79-0.89), and 77% of original volume (95% CI= 0.72-0.83) respectively. Statistically significant volume reductions were noted to begin by radiation treatment day 11 with meaningful dosimetric implications. These data may, therefore, be useful in the implementation of an adaptive radiation therapy planning scheme that periodically adjusts the conformal treatment plan to match these anatomic

2D-Sigmoid Enhancement Prior to Segment MRI Glioma Tumour

Tumour identification has always been a topic that interested researchers around the world. The most challenging phase in tumour identification based on brain MR image is the segmentation of the tumour contour which may contain many unwanted details. Intensity inhomogeneities often occur in real world images and may cause the difficulties in image segmentation. In order to overcome the difficulties caused by intensity inhomogeneity, the study presented pre-processing prior to a region based active contour model with modification of Region Scalable Fitting (MRF) method for image segmentation. Region based active contour model that draw upon intensity information in local regions. The pre-processing is a kind of image enhancement which applies the 2D-sigmoid function at tumour boundary. 2D-sigmoid function enhances the contrast in the brain MRI image for pre-processing steps. Enhanced pixel value, F(x, y), is the ‘S’ shape function of intensity I (x, y) of the image at the point (x, y) , width of the gradient magnitude around brain image (α) and gradient magnitude around brain image (β). Experimental results show desirable of MRF method in terms of computation efficiency.

Liver CBCT Reconstruction by Prior-Knowledge Guided Motion Modeling and Biomechanical Modeling

We developed a new liver CBCT reconstruction technique, which solves a deformation vector field (DVF) to deform a prior high-quality liver CT/CBCT to the new CBCT. The DVF also enables fast on-board liver tumor tracking, dose warping, and adaptive radiation therapy. The new liver CBCT reconstruction technique is based on the foundation of “2D-3D deformation,” which solves the DVF by matching 2D projections simulated from the deformed prior image to 2D on-board CBCT projections. The accuracy of 2D-3D deformation is limited by the low-contrast liver tumors, which lack sufficient intensity variations from liver parenchyma in 2D projections. The new technique applies two strategies to enhance the accuracy of 2D-3D deformation: 1) prior-knowledge guided motion modeling; and 2) finite-element-analysis based biomechanical modeling. To improve the DVF accuracy at the liver boundary, especially for the inferior liver boundary, the liver is contoured on a prior liver 4D-CT and the contoured liver volume at each phase is then density-overridden to enhance the boundary contrasts. Deformable registration is then performed between the 4D-CT phase images to study the liver boundary motion pattern to extract a motion model. Based on the patient-specific liver motion model, an initial DVF is generated and input into 2D-3D deformation for refinement. Based on the refined DVF, biomechanical modeling is further applied to improve the intra-liver DVF accuracy. Biomechanical modeling uses the DVF at the liver boundary to drive finite element analysis, which fine-tunes the DVFs within the liver volume, including those of the tumor. The fine-tuned DVF is fed back into 2D-3D deformation to form an iterative loop until convergence. The efficacy of the reconstruction algorithm was tested on 3 liver cancer patients. Each patient had a prior CT and a new CT. On-board cone-beam projections were simulated from the new CT for reconstruction. The low-contrast liver tumors were contoured on the prior and new CT images. The new tumor contour (“ground-truth”) was compared against those propagated from the prior tumor contour by the solved DVFs. Their DICE coefficients were calculated to evaluate the reconstruction accuracy. Compared with the other techniques, the new technique generated CBCTs with the highest accuracy (Table 1) for the liver tumor.Abstract 206; Table 1The DICE coefficient between the deformed tumor volume and the “ground-truth” tumor volumePatient No.No. of projections for reconstruction2D-3D deformation2D-3D deformation w/ biomechanical modelingThe new technique150.6020.8510.924100.6220.8750.926200.6490.8920.927250.6580.8350.885100.6710.8690.889200.6710.8820.897350.4110.5630.772100.4110.5980.780200.4110.6610.802 Open table in a new tab The new technique substantially improved the accuracy of the reconstructed liver tumor volume. It can also reduce 3D/4D-CBCT imaging dose by acquiring fewer projections for reconstruction.

Radiogenomic Analysis of a Peritumoral CT Image Feature and Its Prognostic Value in Early Stage NSCLC

The prevalence of early stage non-small cell lung cancer (NSCLC) is expected to increase with the implementation of CT screening programs. Prognostic biomarkers are critically needed that can predict which patients are at a high risk of relapse and thus require additional systemic therapy. The purpose of this work is to identify the molecular basis and assess prognostic value of a novel peritumoral CT image feature in early stage NSCLC. We retrospectively analyzed seven independent NSCLC cohorts (n = 1069). First, a 3D quantitative pleural contact index (PCI) was defined on CT images as tumor-pleura contact area normalized by the total tumor surface area. Based on PCI, we dichotomized patients into two groups in a discovery cohort (n = 117) and tested on an external cohort (n = 88). Both cohorts consisted of stage I NSCLC patients treated with radiation therapy. Next, we analyzed another cohort of 89 NSCLC patients with both imaging and gene expression data to identify the molecular basis of PCI and build a gene expression-based surrogate (i.e., radiogenomic classifier). To further test its prognostic relevance, we identified four datasets totaling 775 stage I NSCLC patients with publically available gene expression data and linked survival information. PCI showed a high degree of reproducibility against variations in tumor contours with an intra-class correlation of 0.89. At a cutoff of 0.25, PCI significantly stratified patients for overall survival in both discovery (log-rank test P = 0.0076) and validation (log-rank test P = 0.0304) imaging cohorts. PCI also stratified patients for distant metastasis in the discovery cohort (log-rank test P = 0.0060). Gene set enrichment analysis showed that PCI was significantly associated with genes related to extracellular matrix organization/disassembly and collagen catabolic process (Spearmen test P < 0.05). The radiogenomic classifier stratified patients for overall survival in each of the four gene expression cohorts (log-rank test P range: <0.0001 to 0.019) and remained an independent predictor when adjusting for age, gender, and tumor stage (Wald test P = 0.0007). Pleural contact index can serve as a noninvasive prognostic marker and might be used to guide risk-adapted therapy in early stage NSCLC. We envision that the radiogenomic strategy can be used to identify biological meaningful and clinical relevant imaging biomarkers in other cancer types.

The influence of automation on tumor contouring

Fully or semi-automatic contouring tools are increasingly being used in the tumor contouring task for radiotherapy. While the fully automatic contouring tools have not reached sufficient efficiency, the semi-automatic contouring tools balance more effectively between the human interaction and automation. This study evaluates the influences of a semi-automation contouring tool, called between-slice interpolation, on the resulting contours and the contouring process. The tumor contouring study was conducted on three patient cases with five physicians in a naturalistic setting. The contouring task consisted of initiating the 2D contour manually or with the interpolation tool and correcting that initial contour. The similarity of the resulting contours was pairwise measured within the manual or the interpolated category. Interactions with the software were recorded, and variations in the contouring workflows steps were compared. Results indicated that using the between-slice interpolation tool for creating the initial contour, instead of initiating it manually, influenced both the contouring process and outcomes. First, it was identified that contours initiated by the interpolation tool showed an increased similarity among themselves compared to the manually initiated contours. At the same time, influences to the resulting contours were below clinical relevance, and toward the desired direction—improved consistency of contours. Second, when interpolation was used, in two cases out of three, the average contouring time also decreased significantly. Therefore, the use of such an automation tool can be encouraged.

Segmentation of breast ultrasound images based on active contours using neutrosophic theory.

Ultrasound imaging is an effective approach for diagnosing breast cancer, but it is highly operator-dependent. Recent advances in computer-aided diagnosis have suggested that it can assist physicians in diagnosis. Definition of the region of interest before computer analysis is still needed. Since manual outlining of the tumor contour is tedious and time-consuming for a physician, developing an automatic segmentation method is important for clinical application. The present paper represents a novel method to segment breast ultrasound images. It utilizes a combination of region-based active contour and neutrosophic theory to overcome the natural properties of ultrasound images including speckle noise and tissue-related textures. First, due to inherent speckle noise and low contrast of these images, we have utilized a non-local means filter and fuzzy logic method for denoising and image enhancement, respectively. This paper presents an improved weighted region-scalable active contour to segment breast ultrasound images using a new feature derived from neutrosophic theory. This method has been applied to 36 breast ultrasound images. It generates true-positive and false-positive results, and similarity of 95%, 6%, and 90%, respectively. The purposed method indicates clear advantages over other conventional methods of active contour segmentation, i.e., region-scalable fitting energy and weighted region-scalable fitting energy.

Association between textural and morphological tumor indices on baseline PET-CT and early metabolic response on interim PET-CT in bulky malignant lymphomas.

We investigated whether metabolic, textural, and morphological tumoral indices evaluated on baseline PET-CT were predictive of early metabolic response on interim PET-CT in a cohort of patients with bulky Hodgkin and non-Hodgkin malignant lymphomas. This retrospective study included 57 patients referred for initial PET-CT examination. In-house dedicated software was used to delineate tumor contours using a fixed 30% threshold of SUV max and then to compute tumoral metabolic parameters (SUV max, mean, peak, standard deviation, skewness and kurtosis, metabolic tumoral volume (MTV), total lesion glycolysis, and area under the curve of the cumulative histogram), textural parameters (Moran's and Geary's indices, energy, entropy, contrast, correlation derived from the gray-level co-occurrence matrix, area under the curve of the power spectral density, auto-correlation distance, and granularity), and shape parameters (surface, asphericity, convexity, surfacic extension, and 2D and 3D fractal dimensions). Early metabolic response was assessed on interim PET-CT using the Deauville 5-point scale and patients were ranked according to the Lugano classification as complete or not complete metabolic responders. The impact of the segmentation method (alternate threshold at 41%) and image resolution (Gaussian postsmoothing of 3, 5, and 7mm) was investigated. The association of the proposed parameters with early response was assessed in univariate and multivariate analyses. Their added predictive value was explored using supervised classification by support vector machines (SVM). We evaluated in leave-one-out cross-validation three SVMs admitting as input features (a) MTV, (b) MTV+histological type, and (c) MTV+histology+relevant texture/shape indices. Features associated with complete metabolic response were low MTV (P=0.01), low TLG (P=0.003), high power spectral density AUC (P=0.007), high surfacic extension (P=0.006), low 2D fractal dimension (P=0.007), and low 3D fractal dimension (P=0.003). The prognostic value of these metrics was optimal with the 30% segmentation threshold and overall was progressively altered with decreasing image resolution. In cross-validation, the SVM accounting for texture and shape achieved the highest predictive value with ROC AUC of 0.82 and 80% accuracy (compared with 0.68 and 61% for MTV, and 0.65 and 68% for MTV+histology). The combination of usual prognostic factors with appropriately chosen textural and shape parameters evaluated on baseline PET-CT improves the prediction of early metabolic response in bulky lymphoma.

An Image Fusion System for Estimating the Therapeutic Effects of Radiofrequency Ablation on Hepatocellular Carcinoma.

BackgroundDuring ultrasound-guided radiofrequency ablation (RFA) of hepatocellular carcinoma (HCC), high echoic areas due to RFA-induced microbubbles can help calculate the extent of ablation. However, these areas also decrease visualization of target tumors, making it difficult to assess whether they completely cover the tumors. To estimate the effects of RFA more precisely, we used an image fusion system (IFS).Patients and methodsWe enrolled patients with a single HCC who received RFA with or without the IFS. In the IFS group, we drew a spherical marker along the contour of a target tumor on reference images immediately after administering RFA so that the synchronized spherical marker represented the contour of the target tumor on real-time ultrasound images. When the high echoic area completely covered the marker, we considered the ablation to be complete. We compared outcomes between the IFS and control groups.ResultsWe enrolled 25 patients and 20 controls, and the baseline characteristics were similar between the two groups. The complete ablation rates during the first RFA session were significantly higher in the IFS group compared with those in the control group (88.0% vs. 60.0%, P = 0.041). The number of RFA sessions was significantly smaller in the IFS group compared with that in the control group (1.1 ± 0.3 vs. 1.5 ± 0.7, P = 0.016).ConclusionsThe study suggested that the IFS enables a more precise estimation of the effects of RFA on HCC, contributing to enhanced treatment efficacy and minimized patient burden.

Real-time dynamic MR image reconstruction using compressed sensing and principal component analysis (CS-PCA): Demonstration in lung tumor tracking.

This work presents a real-time dynamic image reconstruction technique, which combines compressed sensing and principal component analysis (CS-PCA), to achieve real-time adaptive radiotherapy with the use of a linac-magnetic resonance imaging system. Six retrospective fully sampled dynamic data sets of patients diagnosed with non-small-cell lung cancer were used to investigate the CS-PCA algorithm. Using a database of fully sampled k-space, principal components (PC's) were calculated to aid in the reconstruction of undersampled images. Missing k-space data were calculated by projecting the current undersampled k-space data onto the PC's to generate the corresponding PC weights. The weighted PC's were summed together, and the missing k-space was iteratively updated. To gain insight into how the reconstruction might proceed at lower fields, 6× noise was added to the 3T data to investigate how the algorithm handles noisy data. Acceleration factors ranging from 2 to 10× were investigated using CS-PCA and Split Bregman CS for comparison. Metrics to determine the reconstruction quality included the normalized mean square error (NMSE), as well as the dice coefficients (DC) and centroid displacement of the tumor segmentations. Our results demonstrate that CS-PCA performed superior than CS alone. The CS-PCA patient averaged DC for 3T and 6× noise added data remained above 0.9 for acceleration factors up to 10×. The patient averaged NMSE gradually increased with increasing acceleration; however, it remained below 0.06 up to an acceleration factor of 10× for both 3T and 6× noise added data. The CS-PCA reconstruction speed ranged from 5 to 20 ms (Intel i7-4710HQ CPU @ 2.5 GHz), depending on the chosen parameters. A real-time reconstruction technique was developed for adaptive radiotherapy using a Linac-MRI system. Our CS-PCA algorithm can achieve tumor contours with DC greater than 0.9 and NMSE less than 0.06 at acceleration factors of up to, and including, 10×. The reconstruction speed for the Split Bregman CS ranged from 200 to 260 ms, whereas the CS-PCA reconstruction speed ranged from 5 to 20 ms implemented using nonoptimized MATLAB code.

Three-dimensional histopathological reconstruction as a reliable ground truth for prostate cancer studies

To validate new imaging modalities for prostate cancer, images must be three-dimensionally correlated with the histological ground truth. In this work, an interpolation algorithm is described to construct a reliable three-dimensional reference from two-dimensional (2D) histological slices. Eight clinically relevant in silico phantoms were designed to represent difficult-to-reconstruct tumour structures. These phantoms were subjected to different slicing procedures. Additionally, controlled errors were added to investigate the impact of varying slicing distance, front-face orientation, and inter-slice misalignment on the reconstruction performance. Using a radial-basis-function interpolation algorithm, the 2D data were reconstructed in three dimensions. Our results demonstrate that slice thicknesses up to 4 mm can be used to reliably reconstruct tumours of clinically significant size; the surfaces lay within a 1.5 mm 90%-error margin from each other and the volume difference between the original and reconstructed tumour structures does not exceed 10%. With these settings, Dice coefficients above 0.85 are obtained. The presented interpolation algorithm is able to reconstruct clinically significant tumour structures from 2D histology slices. Errors occurring are in the order of magnitude of common registration artefacts. The method’s applicability to real histopathological data is also shown in two resected prostates. An inter-slice spacing of 4 mm or less is recommended during histopathology; the use of a 1.5 mm error margin along the tumour contours can then ensure reliable mapping of the ground truth.

Quantitative imaging: Correlating image features with the segmentation accuracy of PET based tumor contours in the lung

The purpose of this study was to investigate the correlation between image features extracted from PET images and the accuracy of manually drawn lesion contours in the lung. Such correlations are interesting in that they could potentially be used in predictive models to help guide physician contouring. In this work, 26 synthetic PET datasets were created using an anthropomorphic phantom and Monte Carlo simulation. Manual contours of simulated lesions were provided by 10 physicians. Contour accuracy was quantified using five commonly used similarity metrics which were then correlated with several features extracted from the images. Features were sub-divided into three groups using intensity, geometry, and texture as categorical descriptors. When averaged among the participants, the results showed relatively strong correlations with complexity and contrastI (r≥0.65, p<0.001), and moderate correlations with several other image features (r≥0.5, p<0.01). The predictive nature of these correlations was improved through stepwise regression and the creation of multi-feature models. Imaging features were also correlated with the standard deviation of contouring error in order to investigate inter-observer variability. Several features were consistently identified as influential including integral of mean curvature and complexity. These relationships further the understanding as to what causes variation in the contouring of PET positive lesions.