Accurate Tumor Delineation Research Articles

Changes in treatment intent and target definition for preoperative radiotherapy after 18F-Fluorodeoxyglucose positron emission tomography in rectal cancer: A Meta-analysis

PurposeTo evaluate the impact of 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) on changes in treatment plan and target definition for preoperative radiotherapy in patients with rectal cancer. MethodsEmbase, PubMed, and Cochrane Library were searched up to November 2020 for all studies investigating the role of preoperative FDG PET in patients who underwent neoadjuvant radiotherapy before curative-intent surgery. The proportion of patients whose treatment plan (curative vs. palliative intent) or target definition was changed after FDG PET was analyzed. A random-effects model was used for pooled analysis. The change in target definition was compared between conventional radiological imaging-based target volume [gross tumor volume (GTV) or planning target volume (PTV)] and PET-based target volume (GTV or PTV) using the standardized mean difference (SMD) and 95% confidence interval (CI). ResultsA total of 336 patients from twelve studies were included. In eight studies, PET changed either the treatment intent or target definition in 24.8% of patients (95% CI 15.1% to 37.9%, I2 = 69%). In ten studies, the PET-based GTV was lower than the conventional imaging-based target volume (SMD −7.0, 95% CI −1.39 to −0.01). However, there was no significant difference between conventional imaging-based and PET-based PTV (SMD −0.07, 95% CI −0.75 to 0.62). In six studies evaluating the initial staging based on PET, the initial staging (nodal or metastasis status) was changed in 53 of 229 patients (23.1%). Newly detected or additional distant metastases were identified in 22 patients (9.6%) after FDG PET. ConclusionThe use of FDG PET influences radiotherapy planning in a fourth of patients with rectal cancer. FDG PET can provide additive information for accurate tumor delineation, although PET-based PTV did not significantly change. These findings suggest that FDG PET may be beneficial to patients with rectal cancer before establishing a radiotherapy plan.

Accurate Tumor Delineation vs. Rough Volume of Interest Analysis for 18F-FDG PET/CT Radiomics-Based Prognostic Modeling inNon-Small Cell Lung Cancer.

BackgroundThe aim of this work was to investigate the ability of building prognostic models in non-small cell lung cancer (NSCLC) using radiomic features from positron emission tomography and computed tomography with 2-deoxy-2-[fluorine-18]fluoro-d-glucose (18F-FDG PET/CT) images based on a “rough” volume of interest (VOI) containing the tumor instead of its accurate delineation, which is a significant time-consuming bottleneck of radiomics analyses.MethodsA cohort of 138 patients with stage II–III NSCLC treated with radiochemotherapy recruited retrospectively (n = 87) and prospectively (n = 51) was used. Two approaches were compared: firstly, the radiomic features were extracted from the delineated primary tumor volumes in both PET (using the automated fuzzy locally adaptive Bayesian, FLAB) and CT (using a semi-automated approach with 3D Slicer™) components. Both delineations were carried out within previously manually defined “rough” VOIs containing the tumor and the surrounding tissues, which were exploited for the second approach: the same features were extracted from this alternative VOI. Both sets for features were then combined with the clinical variables and processed through the same machine learning (ML) pipelines using the retrospectively recruited patients as the training set and the prospectively recruited patients as the testing set. Logistic regression (LR), random forest (RF), and support vector machine (SVM), as well as their consensus through averaging the output probabilities, were considered for feature selection and modeling for overall survival (OS) prediction as a binary classification (either median OS or 6 months OS). The resulting models were compared in terms of balanced accuracy, sensitivity, and specificity.ResultsOverall, better performance was achieved using the features from delineated tumor volumes. This was observed consistently across ML algorithms and for the two clinical endpoints. However, the loss of performance was not significant, especially when a consensus of the three ML algorithms was considered (0.89 vs. 0.88 and 0.78 vs. 0.77).ConclusionOur findings suggest that it is feasible to achieve similar levels of prognostic accuracy in radiomics-based modeling by relying on a faster and easier VOI definition, skipping a time-consuming tumor delineation step, thus facilitating automation of the whole radiomics workflow. The associated cost is a loss of performance in the resulting models, although this loss can be greatly mitigated when a consensus of several models is relied upon.

Deep Learning Segmentation of Triple-Negative Breast Cancer (TNBC) Patient Derived Tumor Xenograft (PDX) and Sensitivity of Radiomic Pipeline to Tumor Probability Boundary.

Simple SummaryCo-clinical trials are an emerging area of investigation in which a clinical trial is coupled with a corresponding preclinical trial to inform the corresponding clinical trial. The preclinical arm aids in assessing therapeutic efficacy, patient stratification, and designing optimal imaging strategies. There is much interest in harmonizing preclinical and clinical quantitative imaging pipelines. Radiomics is widely explored in clinical imaging to predict response to therapy. In preclinical imaging, high-throughput radiomic analysis is limited by manual delineation of tumor boundaries, which is labor intensive with poor reproducibility. Our proposed deep-learning-based system was trained to automatically segment tumors from multi-contrast MR images and extract radiomic features. The proposed method is highly reproducible with significant correlation in radiomic features. The deployment of this pipeline in the preclinical arm would provide high throughput and reproducible radiomic analysis.Preclinical magnetic resonance imaging (MRI) is a critical component in a co-clinical research pipeline. Importantly, segmentation of tumors in MRI is a necessary step in tumor phenotyping and assessment of response to therapy. However, manual segmentation is time-intensive and suffers from inter- and intra- observer variability and lack of reproducibility. This study aimed to develop an automated pipeline for accurate localization and delineation of TNBC PDX tumors from preclinical T1w and T2w MR images using a deep learning (DL) algorithm and to assess the sensitivity of radiomic features to tumor boundaries. We tested five network architectures including U-Net, dense U-Net, Res-Net, recurrent residual UNet (R2UNet), and dense R2U-Net (D-R2UNet), which were compared against manual delineation by experts. To mitigate bias among multiple experts, the simultaneous truth and performance level estimation (STAPLE) algorithm was applied to create consensus maps. Performance metrics (F1-Score, recall, precision, and AUC) were used to assess the performance of the networks. Multi-contrast D-R2UNet performed best with F1-score = 0.948; however, all networks scored within 1–3% of each other. Radiomic features extracted from D-R2UNet were highly corelated to STAPLE-derived features with 67.13% of T1w and 53.15% of T2w exhibiting correlation ρ ≥ 0.9 (p ≤ 0.05). D-R2UNet-extracted features exhibited better reproducibility relative to STAPLE with 86.71% of T1w and 69.93% of T2w features found to be highly reproducible (CCC ≥ 0.9, p ≤ 0.05). Finally, 39.16% T1w and 13.9% T2w features were identified as insensitive to tumor boundary perturbations (Spearman correlation (−0.4 ≤ ρ ≤ 0.4). We developed a highly reproducible DL algorithm to circumvent manual segmentation of T1w and T2w MR images and identified sensitivity of radiomic features to tumor boundaries.

CBSN: Comparative measures of normalization techniques for brain tumor segmentation using SRCNet

The segmentation of tumors in the brain MRI scans is a difficult job for doctors and radiologists. The segmentation done by different medical experts may also have differences in their opinion for the segmented region, which is popularly known as regions of interest (ROIs). To date, researchers and academicians have proposed several approaches and frameworks for semi- and full-automatic segmentation techniques to identify ROIs accurately. It is prevalent that automatic segmentation gives comparable or even better results compared to human experts for several publicly known and privately collected datasets. Additionally, these are beneficial in those areas where doctors and radiologists’ availability is either uneven or scarce because of geographical dispersion. The convolutional neural networks (CNN) are considered for segmentation for ROIs due to their wide popularity. They have outperformed humans over tasks like object identification and image classification. The publicly available datasets or those collected from different medical institutions may have different statistics, resolution, and properties. Therefore, pre-processing has an essential role in achieving better and accurate delineation and segmentation of tumors. In the proposed work, CBSN, we consider well-known normalization techniques such as Gaussian Mixture Models (GMM), Fuzzy C-Means (FCM), and Z-score normalization for pre-processing the BraTS (Brain Tumor Segmentation Challenge) 2018 dataset. We utilized three variants of U-Net architecture, convolutional block attention module (CBAM), squeeze and excitation module (SEM), and refinement module (RM) for the segmentation of the ROIs. Utilizing Z-score performs better than other normalization techniques for tumor core (TC) and whole tumor (WT) segmentation. In contrast, FCM performs superior to the other two normalization techniques on enhancement tumor (ET) segmentation.

Dosimetric impact of contrast-enhanced 4d computed tomography for stereotactic body radiation therapy of hepatocelluar carcinoma.

A purpose of the study was to investigate the dosimetric impact of contrast media on dose calculation using average 4D contrast-enhanced computed tomography (4D-CECT) and delayed 4D-CT (d4D-CT) images caused by CT simulation contrast agents for stereotactic body radiation therapy (SBRT) of liver cases. Fifteen patients of liver SBRT treated using the volumetric modulated arc therapy (VMAT) technique were selected retrospectively. 4D-CECT, and d4D-CT were acquired with the Anzai gating system and GE CT. For all patients, gross target volume (GTV) was contoured on the ten phases after rigid registration of both the contrast and delayed scans and merged to generate internal target volume (ITV) on average CT images. Region of interest (ROI) was drawn on contrast images and then copied to the delayed images after rigid registration of two average CT datasets. The treatment plans were generated for contrast enhanced average CT, delayed average CT and contrast enhanced average CT with electron density of the heart overridden. No significant dosimetric difference was observed in plans parameters (mean HU value of the liver, total monitor units, total control points, degree of modulation and average segment area) except mean HU value of the aorta amongst the three arms. All the OARs were evaluated and resulted in statistically insignificant variation (p > 0.05) using one way ANOVA analysis. Contrast enhanced 4D-CT is advantageous in accurate delineation of tumors and assessing accurate ITV. The treatment plans generated on average 4D-CECT and average d4D-CT have a clinically insignificant effect on dosimetric parameters.

Combining Images and T-Staging Information to Improve the Automatic Segmentation of Nasopharyngeal Carcinoma Tumors in MR Images

The accurate and reproducible delineation of tumors from uninvolved tissue is essential for radiation oncology. However, the tumor margin may be challenging to identify from magnetic resonance (MR) images of nasopharyngeal carcinomas (NPCs). Additionally, clinical diagnoses such as T-staging may already provide some information on tumor invasion. To use this information and improve the performance of tumor segmentation, we propose a novel deep learning neural network architecture that can incorporate both T-staging and image information. Based on U-Net, our model adds a T-channel composed of T-staging information and uses the attention mechanism. Since the T-staging information is defined by the extent of tumor invasion, the T-channel using T-staging information can improve the segmentation accuracy at different stages. Additionally, the addition of an attention mechanism allows our model to retain the most valuable pixels of the image, thus further improving the delineation accuracy. In our experiments, the proposed network was trained and validated based on records from 251 clinical patients using 10-fold cross-validation. The dice similarity coefficient (DSC) and average symmetric surface distance (ASSD) were used to evaluate our network's results. The average DSC and ASSD and their standard deviation (SD) values are 0.841 ± 0.011 and 0.747 ± 0.199 mm. The unique T-channel effectively utilizes T-staging information to improve the results. With the combination of the T-channel module and the attention module, we significantly improved NPC tumor delineation performance.

Assessment of Gadobutrol Safety in Combination with Ionizing Radiation Using a Preclinical MRI-Guided Radiotherapy Model.

MR-linac technology enhances the precision of therapeutic radiation by clarifying the tumor-normal tissue interface and provides the potential for adaptive treatment planning. Accurate delineation of tumors on diagnostic magnetic resonance imaging (MRI) frequently requires gadolinium-based contrast agents (GBCAs). Despite generally being considered safe, previous literature suggests that GBCAs are capable of contrast-induced acute kidney injury (AKI). It is unclear if the risk for AKI is enhanced when GBCAs are administered concurrently with ionizing radiotherapy. During irradiation, gadolinium may be liberated from its chelator which may induce AKI. The goal of this work was to determine if radiation combined with GBCAs increased the incidence of AKI. Using a preclinical MRI-guided irradiation system, where MRI acquisitions and radiation delivery are performed in rapid succession, tumor-bearing mice with normal kidney function were injected with GBCA and treated with 2, 8 or 18 Gy irradiation. Renal function was assessed on days three and seven postirradiation to assess for AKI. No clinically relevant changes in blood urea nitrogen and creatinine were observed in any combination of GBCA and radiation dose. From these data, we conclude that GBCA in combination with radiation does not increase the risk for AKI in mice. Additional investigation of multiple doses of GBCA administered concurrently with irradiation is warranted to evaluate the risk of chronic kidney injury.

Magnetic Resonance Molecular Imaging of Extradomain B Fibronectin Improves Imaging of Pancreatic Cancer Tumor Xenografts.

The survival of pancreatic cancer patients can be greatly improved if their disease is detected at an early, potentially curable stage. Magnetic resonance molecular imaging (MRMI) of oncoproteins is a promising strategy for accurate, early detection of the disease. Here, we test the hypothesis that MRMI of extradomain-B fibronectin (EDB-FN), an abundant oncoprotein in the tumor extracellular matrix, can overcome the stromal barriers of pancreatic cancer to facilitate effective molecular imaging and detection of small tumors. Specimens of normal, premalignant, and malignant human pancreatic tissues were stained with a peptide-fluorophore conjugate (ZD2-Cy5.5) to assess EDB-FN binding and expression. MRMI with ZD2-N3-Gd(HP-DO3A) (MT218) specific to EDB-FN and MRI with Gd(HP-DO3A) were performed in three murine models bearing human pancreatic cancer xenografts, including a Capan-1 flank model, a BxPC3-GFP-Luc and a PANC-1-GFP-Luc intrapancreatic xenograft model. Tumor enhancement of the contrast agents was analyzed and compared. Staining of human tissue samples with ZD2-Cy5.5 revealed high EDB-FN expression in pancreatic tumors, moderate expression in premalignant tissue, and little expression in normal tissue. MRMI with MT218 generated robust intratumoral contrast, clearly detected and delineated small tumors (smallest average size: 6.1 mm2), and out-performed conventional contrast enhanced MRI with Gd(HP-DO3A). Quantitative analysis of signal enhancement revealed that MT218 produced 2.7, 2.1, and 1.6 times greater contrast-to-noise ratio (CNR) than the clinical agent in the Capan-1 flank, BxPC3-GFP-Luc and PANC-1-GFP-Luc intrapancreatic models, respectively (p < 0.05). MRMI of the ECM oncoprotein EDB-FN with MT218 is able to generate superior contrast enhancement in small pancreatic tumors and provide accurate tumor delineation in animal models. Early, accurate detection and delineation of pancreatic cancer with high-resolution MRMI has the potential to guide timely treatment and significantly improve the long-term survival of pancreatic cancer patients.

Concurrent Injection of Unlabeled Antibodies Allows Positron Emission Tomography Imaging of Programmed Cell Death Ligand 1 Expression in an Orthotopic Pancreatic Tumor Model.

Purpose: Among the treatment options for pancreatic ductal adenocarcinoma (PDAC) are antibodies against the programmed cell death receptor 1 (PD-1)/programmed cell death ligand 1 (PD-L1) pathway. Positron emission tomography (PET) has been successfully used to assess PD-1/PD-L1 signaling in subcutaneous tumor models, but orthotopic tumor models are increasingly being recognized as a better option to accurately recapitulate human disease. However, when PET radiotracers have high uptake in the liver and spleen, it can obscure signals from the adjacent pancreas, making visualization of the response in orthotopic pancreatic tumors technically challenging. In this study, we first investigated the impact of radioisotope chelators on the biodistribution of 64Cu-labeled anti-PD-1 and anti-PD-L1 antibodies and compared the distribution profiles of anti-PD-1 and anti-PD-L1 antibodies. We then tested the hypothesis that co-injection of unlabeled antibodies reduces uptake of 64Cu-labeled anti-PD-L1 antibodies in the spleen and thereby permits accurate delineation of orthotopic pancreatic tumors in mice. Procedures: We established subcutaneous and orthotopic mouse models of PDAC using KRAS* murine pancreatic cancer cells with a doxycycline-inducible mutation of KRASG12D. We then (1) compared the biodistribution of 64Cu-labeled anti-PD-1 with 2-(4-isothiocyanatobenzyl)-1,4,7,10-tetraazacyclododecane tetraacetic acid (p-SCN-Bn-DOTA) and 2-(4-isothiocyanatobenzyl)-1,4,7-triazacyclononane-1,4,7-triacetic acid (p-SCN-Bn-NOTA) used as the chelators in the orthotopic model; (2) compared the biodistribution of [64Cu]Cu-NOTA-anti-PD-1 and [64Cu]Cu-NOTA-anti-PD-L1 in the orthotopic model; and (3) imaged subcutaneous and orthotopic KRAS* tumors with [64Cu]Cu-NOTA-anti-PD-L1 with and without co-injection of unlabeled anti-PD-L1 as the blocking agent. Results: [64Cu]Cu-NOTA-anti-PD-L1 was a promising imaging probe. By co-injection of an excess of unlabeled anti-PD-L1, background signals of [64Cu]Cu-NOTA-anti-PD-L1 from the spleen were significantly reduced, leading to a clear delineation of orthotopic pancreatic tumors. Conclusions: Co-injection with unlabeled anti-PD-L1 is a useful method for PET imaging of PD-L1 expression in orthotopic pancreatic cancer models.

Gradient-based generation of intermediate images for heterogeneous tumor segmentation within hybrid PET/MRI scans

Segmentation of tumors from hybrid PET/MRI scans plays an essential role in accurate diagnosis and treatment planning. However, when treating tumors, several challenges, notably heterogeneity and the problem of leaking into surrounding tissues with similar high uptake, have to be considered. To address these issues, we propose an automated method for accurate delineation of tumors in hybrid PET/MRI scans. The method is mainly based on creating intermediate images. In fact, an automatic detection technique that determines a preliminary Interesting Uptake Region (IUR) is firstly performed. To overcome the leakage problem, a separation technique is adopted to generate the final IUR. Then, smart seeds are provided for the Graph Cut (GC) technique to obtain the tumor map. To create intermediate images that tend to reduce heterogeneity faced on the original images, the tumor map gradient is combined with the gradient image. Lastly, segmentation based on the GCsummax technique is applied to the generated images. The proposed method has been validated on PET phantoms as well as on real-world PET/MRI scans of prostate, liver and pancreatic tumors. Experimental comparison revealed the superiority of the proposed method over state-of-the-art methods. This confirms the crucial role of automatically creating intermediate images in addressing the problem of wrongly estimating arc weights for heterogeneous targets.

Label-free multiphoton imaging allows brain tumor recognition based on texture analysis-a study of 382 tumor patients.

BackgroundLabel-free multiphoton microscopy has been suggested for intraoperative recognition and delineation of brain tumors. For any future clinical application, appropriate approaches for image acquisition and analysis have to be developed. Moreover, an evaluation of the reliability of the approach, taking into account inter- and intrapatient variability, is needed.MethodsCoherent anti-Stokes Raman scattering (CARS), two-photon excited fluorescence (TPEF), and second-harmonic generation were acquired on cryosections of brain tumors of 382 patients and 28 human nontumor brain samples. Texture parameters of those images were calculated and used as input for linear discriminant analysis.ResultsThe combined analysis of texture parameters of the CARS and TPEF signal proved to be most suited for the discrimination of nontumor brain versus brain tumors (low- and high-grade astrocytoma, oligodendroglioma, glioblastoma, recurrent glioblastoma, brain metastases of lung, colon, renal, and breast cancer and of malignant melanoma) leading to a correct rate of 96% (sensitivity: 96%, specificity: 100%). To approximate the clinical setting, the results were validated on 42 fresh, unfixed tumor biopsies. 82% of the tumors and, most important, all of the nontumor samples were correctly recognized. An image resolution of 1 µm was sufficient to distinguish brain tumors and nontumor brain. Moreover, the vast majority of single fields of view of each patient’s sample were correctly classified with high probabilities, which is important for clinical translation.ConclusionLabel-free multiphoton imaging might allow fast and accurate intraoperative delineation of primary and secondary brain tumors in combination with endoscopic systems.

Integration of a knowledge-based constraint into generative models with applications in semi-automatic segmentation of liver tumors

Accurate delineation of liver tumors in medical images is a vital step in diagnosis, treatment planning, and monitoring. In this paper, we utilize a generative model for segmentation of abnormal liver regions. After preprocessing of an input image, the ROI of the tumor is determined, and the boundary of the abnormal region in a single slice is specified. Then, the remaining slices are processed by a generative model that is enhanced by the integration of a constraint. We search for the boundary of the tumor by a probabilistic approach and obtain the solution using the Bayesian inference. The Kullback-Leibler divergence is used to measure the consistency of the results to the model’s constraint. We evaluated the proposed method using synthetic and clinical data. In the public dataset, we achieved a Dice measure of 0.84 ± 0.06, which outperforms state-of-the-art hepatic tumor segmentation algorithms. Concerning all available clinical images, the Dice index of the proposed method is 0.90 ± 0.03.

Segmenting brain tumors from FLAIR MRI using fully convolutional neural networks

Background and ObjectiveMagnetic resonance imaging (MRI) is an indispensable tool in diagnosing brain-tumor patients. Automated tumor segmentation is being widely researched to accelerate the MRI analysis and allow clinicians to precisely plan treatment—accurate delineation of brain tumors is a critical step in assessing their volume, shape, boundaries, and other characteristics. However, it is still a very challenging task due to inherent MR data characteristics and high variability, e.g., in tumor sizes or shapes. We present a new deep learning approach for accurate brain tumor segmentation which can be trained from small and heterogeneous datasets annotated by a human reader (providing high-quality ground-truth segmentation is very costly in practice). MethodsIn this paper, we present a new deep learning technique for segmenting brain tumors from fluid attenuation inversion recovery MRI. Our technique exploits fully convolutional neural networks, and it is equipped with a battery of augmentation techniques that make the algorithm robust against low data quality, and heterogeneity of small training sets. We train our models using only positive (tumorous) examples, due to the limited amount of available data. ResultsOur algorithm was tested on a set of stage II-IV brain-tumor patients (image data collected using MAGNETOM Prisma 3T, Siemens). Rigorous experiments, backed up with statistical tests, revealed that our approach outperforms the state-of-the-art approach (utilizing hand-crafted features) in terms of segmentation accuracy, offers very fast training and instant segmentation (analysis of an image takes less than a second). Building our deep model is 1.3 times faster compared with extracting features for extremely randomized trees, and this training time can be controlled. Finally, we showed that too aggressive data augmentation may lead to deteriorated performance of the model, especially in the fixed-budget training (with maximum numbers of training epochs). ConclusionsOur method yields the better performance when compared with the state of the art method which utilizes hand-crafted features. In addition, our deep network can be effectively applied to difficult (small, imbalanced, and heterogeneous) datasets, offers controllable training time, and infers in real-time.

Deep learning model integrating features and novel classifiers fusion for brain tumor segmentation.

Automatic and precise segmentation and classification of tumor area in medical images is still a challenging task in medical research. Most of the conventional neural network based models usefully connected or convolutional neural networks to perform segmentation and classification. In this research, we present deep learning models using long short term memory (LSTM) and convolutional neural networks (ConvNet) for accurate brain tumor delineation from benchmark medical images. The two different models, that is, ConvNet and LSTM networks are trained using the same data set and combined to form an ensemble to improve the results. We used publicly available MICCAI BRATS 2015 brain cancer data set consisting of MRI images of four modalities T1, T2, T1c, and FLAIR. To enhance the quality of input images, multiple combinations of preprocessing methods such as noise removal, histogram equalization, and edge enhancement are formulated and best performer combination is applied. To cope with the class imbalance problem, class weighting is used in proposed models. The trained models are tested on validation data set taken from the same image set and results obtained from each model are reported. The individual score (accuracy) of ConvNet is found 75% whereas for LSTM based network produced 80% and ensemble fusion produced 82.29% accuracy.

Mn-Loaded apolactoferrin dots forin vivoMRI and NIR-II cancer imaging

The dual modal imaging probeH-dotwas prepared forin vivoNIR-II/MR imaging of orthotopic U87MG glioblastoma xenografts.

Assessment of uveal melanomas using advanced diffusion-weighted imaging techniques: value of reduced field of view DWI ("zoomed DWI") and readout-segmented DWI (RESOLVE).

Background Diffusion-weighted imaging (DWI) has become an important tool for lesion characterization. Advanced techniques of DWI may be used to improve image quality. Purpose To evaluate multi-shot segmented DWI (rs-EPI, RESOLVE) and reduced field-of view DWI (rFOV-EPI, “zoomed” EPI) in patients with ocular melanoma and to compare image quality and diagnostic performance in differentiation of melanoma from retinal detachment. Material and Methods In this prospective and IRB-approved trial, we examined 26 patients using methods including conventional single-shot echo-planar DWI (ss-EPI), rs-EPI, and rFOV-EPI. Subjective image quality was compared using a four-point score and the maximum tumor length was measured in all sequences. Apparent diffusion coefficient (ADC) measurements were performed using a region-of interest analysis. Tumor delineation and differences in ADC values between melanomas and retinal detachments were compared. Results Diffusion restriction was markedly reduced in melanomas in all applied image techniques. Subjective image quality was significantly higher for rFOV-EPI (score = 3.5 ± 0.5) compared with rs-EPI (score = 3.3 ± 0.6) and ss-EPI (score = 2.5 ± 0.9). Regarding tumor diameter measurements, rFOV-EPI showed the best agreement compared with high-resolution conventional sequences. ADC measurements of the tumor and retinal detachment differed significantly ( P &lt; 0.001) with the rFOV-EPI performing best (sensitivities and specificities compared with T1-weighted ss-EPI 61%/82%; rFOV-EPI 86%/92%; rs-EPI 79%/92%, respectively). Conclusion rFOV-EPI showed improved image quality compared with ss-EPI and rs-EPI, the most accurate tumor delineation and the best differentiation from retinal detachments in our patients.

Treatment planning based on water density image generated using dual‐energy computed tomography for pancreatic cancer with contrast‐enhancing agent: Phantom and clinical study

A contrast-enhancing agent is imperative for the accurate target delineation of pancreatic tumors. This study demonstrates the potential use of treatment planning for patients with pancreatic tumors based on the water density image (WDI) generated by dual-energy computed tomography (DECT). Tissue characterization and multi-energy phantom scanning were performed through DECT and the physical characteristics of the WDI and a virtual monochromatic image (VMI) were assessed. The measured and the corresponding theoretical electron density relative to water (RED) and mass density (MD) were compared. Treatment plans based on the WDI (TPWDI ) and VMI (TPVMI ) were compared for 22 pancreatic cancer patients who underwent contrast-enhanced DECT scan. The total absolute difference in the HU value between the conventional 120kVp images and the VMI was the smallest at the energy level of 77keV (3.3HU), and the VMI at 77keV was used for subsequent analysis. The difference between the measured and theoretical values of RED and MD for iodine using the VMI (>15%) was larger than that using WDI (<4%). In clinical cases, the maximum difference in the dosimetric parameters between TPWDI and TPWDI for the planning target volume was 3.0% when the doses were calculated using AXB, and for the duodenum, it was 1.7%. The WDI estimated the RED and MD accurately and could form the basis for a new treatment planning approach for pancreatic cancer using contrast-enhancing agent.

Investigating a novel split-filter dual-energy CT technique for improving pancreas tumor visibility for radiation therapy.

PurposeTumor delineation using conventional CT images can be a challenge for pancreatic adenocarcinoma where contrast between the tumor and surrounding healthy tissue is low. This work investigates the ability of a split‐filter dual‐energy CT (DECT) system to improve pancreatic tumor contrast and contrast‐to‐noise ratio (CNR) for radiation therapy treatment planning.Materials and methodsMultiphasic scans of 20 pancreatic tumors were acquired using a split‐filter DECT technique with iodinated contrast medium, OMNIPAQUE TM. Analysis was performed on the pancreatic and portal venous phases for several types of DECT images. Pancreatic gross target volume (GTV) contrast and CNR were calculated and analyzed from mixed 120 kVp‐equivalent images and virtual monoenergetic images (VMI) at 57 and 40 keV. The role of iterative reconstruction on DECT images was also investigated. Paired t‐tests were used to assess the difference in GTV contrast and CNR among the different images.ResultsThe VMIs at 40 keV had a 110% greater image noise compared to the mixed 120 kVp‐equivalent images (P < 0.0001). VMIs at 40 keV increased GTV contrast from 15.9 ± 19.9 HU to 93.7 ± 49.6 HU and CNR from 1.37 ± 2.05 to 3.86 ± 2.78 in comparison to the mixed 120 kVp‐equivalent images. The iterative reconstruction algorithm investigated decreased noise in the VMIs by about 20% and improved CNR by about 30%.ConclusionsPancreatic tumor contrast and CNR were significantly improved using VMIs reconstructed from the split‐filter DECT technique, and the use of iterative reconstruction further improved CNR. This gain in tumor contrast may lead to more accurate tumor delineation for radiation therapy treatment planning.

3D FULLY CONVOLUTIONAL NETWORKS FOR CO-SEGMENTATION OF TUMORS ON PET-CT IMAGES.

Positron emission tomography and computed tomography (PET-CT) dual-modality imaging provides critical diagnostic information in modern cancer diagnosis and therapy. Automated accurate tumor delineation is essentially important in computer-assisted tumor reading and interpretation based on PET-CT. In this paper, we propose a novel approach for the segmentation of lung tumors that combines the powerful fully convolutional networks (FCN) based semantic segmentation framework (3D-UNet) and the graph cut based co-segmentation model. First, two separate deep UNets are trained on PET and CT, separately, to learn high level discriminative features to generate tumor/non-tumor masks and probability maps for PET and CT images. Then, the two probability maps on PET and CT are further simultaneously employed in a graph cut based co-segmentation model to produce the final tumor segmentation results. Comparative experiments on 32 PET-CT scans of lung cancer patients demonstrate the effectiveness of our method.

Clinical implementation of contrast-enhanced four-dimensional dual-energy computed tomography for target delineation of pancreatic cancer

Background and purposeThe accurate delineation of pancreatic tumor with respiratory motion is challenging. This study demonstrates the application of contrast-enhanced four-dimensional dual-energy computed tomography (CE-4D-DECT) for tumor delineation and assesses the objective and subjective image quality. Material and methodsTwelve patients underwent CE-4D-DECT, and quantitative spectral analysis was performed on the resulting virtual monochromatic images (VMI) to determine the optimal VMI (O-VMI) with the highest contrast-to-noise ratio (CNR). The objective value of the CNR between pancreatic parenchyma and tumor, and the subjective measurement with five-point scale were compared between O-VMI, standard VMI (S-VMI, 77 keV) and single energy CT (SECT, 120 kVp). ResultsThe CNR was the highest in the VMI at 60 keV, and the corresponding CNR in the O-VMI (3.4) was significantly higher (p < 0.05) than that in the S-VMI (2.4) and the SECT (2.7). The overall mean subjective measurements among 4 radiation oncologists were higher for the O-VMI over the S-VMI and SECT with respect to overall image quality (4.0, 3.3 and 3.7, respectively), tumor enhancement (3.4, 2.6 and 3.2, respectively), and vessel delineation (4.2, 3.6 and 4.2, respectively). ConclusionsThe O-VMI derived from the CE-4D-DECT demonstrated its superiority over the S-VMI and SECT in depicting pancreatic tumor.