Clinical Evaluation of Deep Learning and Atlas-Based Auto-Contouring of Bladder and Rectum for Prostate Radiation Therapy.

PURPOSE: Auto-contouring may reduce workload, interobserver variation, and time associated with manual contouring of organs at risk. Manual contouring remains the standard due in part to uncertainty around the time and workload savings after accounting for the review and editing of auto-contours. This preliminary study compares a standard manual contouring workflow with 2 auto-contouring workflows (atlas and deep learning) for contouring the bladder and rectum in patients with prostate cancer. METHODS AND MATERIALS: Three contouring workflows were defined based on the initial contour-generation method including manual (MAN), atlas-based auto-contour (ATLAS), and deep-learning auto-contour (DEEP). For each workflow, initial contour generation was retrospectively performed on 15 patients with prostate cancer. Then, radiation oncologists (ROs) edited each contour while blinded to the manner in which the initial contour was generated. Workflows were compared by time (both in initial contour generation and in RO editing), contour similarity, and dosimetric evaluation. RESULTS: Mean durations for initial contour generation were 10.9 min, 1.4 min, and 1.2 min for MAN, DEEP, and ATLAS, respectively. Initial DEEP contours were more geometrically similar to initial MAN contours. Mean durations of the RO editing steps for MAN, DEEP, and ATLAS contours were 4.1 min, 4.7 min, and 10.2 min, respectively. The geometric extent of RO edits was consistently larger for ATLAS contours compared with MAN and DEEP. No differences in clinically relevant dose-volume metrics were observed between workflows. CONCLUSION: Auto-contouring software affords time savings for initial contour generation; however, it is important to also quantify workload changes at the RO editing step. Using deep-learning auto-contouring for bladder and rectum contour generation reduced contouring time without negatively affecting RO editing times, contour geometry, or clinically relevant dose-volume metrics. This work contributes to growing evidence that deep-learning methods are a clinically viable solution for organ-at-risk contouring in radiation therapy.

Comparing deep learning-based auto-segmentation of organs at risk and clinical target volumes to expert inter-observer variability in radiotherapy planning.

BACKGROUND: Deep learning-based auto-segmented contours (DC) aim to alleviate labour intensive contouring of organs at risk (OAR) and clinical target volumes (CTV). Most previous DC validation studies have a limited number of expert observers for comparison and/or use a validation dataset related to the training dataset. We determine if DC models are comparable to Radiation Oncologist (RO) inter-observer variability on an independent dataset. METHODS: Expert contours (EC) were created by multiple ROs for central nervous system (CNS), head and neck (H&N), and prostate radiotherapy (RT) OARs and CTVs. DCs were generated using deep learning-based auto-segmentation software trained by a single RO on publicly available data. Contours were compared using Dice Similarity Coefficient (DSC) and 95% Hausdorff distance (HD). RESULTS: Sixty planning CT scans had 2-4 ECs, for a total of 60 CNS, 53 H&N, and 50 prostate RT contour sets. The mean DC and EC contouring times were 0.4 vs 7.7 min for CNS, 0.6 vs 26.6 min for H&N, and 0.4 vs 21.3 min for prostate RT contours. There were minimal differences in DSC and 95% HD involving DCs for OAR comparisons, but more noticeable differences for CTV comparisons. CONCLUSIONS: The accuracy of DCs trained by a single RO is comparable to expert inter-observer variability for the RT planning contours in this study. Use of deep learning-based auto-segmentation in clinical practice will likely lead to significant benefits to RT planning workflow and resources.

Magnetic Resonance Imaging Volumetry of Primary Nasopharyngeal Cancer in Patients Treated with Induction Gemcitabine and Cisplatin Followed by Concurrent Cisplatin and Volumetric Modulated Arc Therapy.

Introduction The addition of induction chemotherapy (IC) to the standard concurrent chemoradiotherapy (CCRT) is under consideration in locally advanced nasopharyngeal carcinoma (LANPC). To-date, no studies have reported primary gross tumour volume (GTVp) changes using gemcitabine and cisplatin as the IC phase in LANPC. We investigated the timing and magnitude of GTVp response throughout sequential gemcitabine and cisplatin IC and CCRT for LANPC. Toxicity and tumour control probability (TCP) analyses are also presented Methods Ten patients with LANPC underwent sequential IC and CCRT between 2011 and 2015. All patients had magnetic resonance imaging (MRI) at three time points: before IC (MRI0), after IC (MRI1), and three months after CCRT (MRI3). Five of the 10 patients had an additional MRI four to five weeks into CCRT (MRI2). GTVp contours were delineated retrospectively using contrast-enhanced MRIs, and each GTVp underwent secondary review by a neuroradiologist. Acute toxicities were graded retrospectively via chart review based on the National Cancer Institute Common Terminology for Adverse Events version 4.0 (NCI CTCAE v4.0). Results Mean GTVp reduction between MRI0 - MRI1 was from 68 cc to 47 cc and from 47 cc to 9 cc between MRI1 - MRI3. In patients with MRI2, the mean GTVp reduction between MRI1 - MRI2 was from 57 cc to 32 cc. Tumour control probability estimates increased by 0.11 after IC. Patients tolerated the treatment well with one Grade IV toxicity event. Conclusion The observed GTVp response and improved tumor control probability support further investigation into the use of IC in LANPC.

Re-irradiation volumetric modulated arc therapy optimization based on cumulative biologically effective dose objectives.

The objective of this note is to introduce a clinical tool that generates ideal base plan dose distributions to enable re-irradiation volumetric modulated arc therapy (VMAT) optimization based on cumulative biological effective dose objectives for specific organs at risk (OARs). The tool is demonstrated with a lung cancer case that required re-irradiation at our clinic. First, previous treatment dose is deformed onto the retreatment computed tomography (CT) using commercial software. Then, the in-house Matlab tool alters the deformed previous dose using radiobiological concepts on a voxel-by-voxel manner to generate an ideal base plan dose distribution. Ideal base plans that were generated using the in-house Matlab tool were compatible with the Varian Eclipse™ treatment planning system. The tool enabled optimization of VMAT re-irradiation plans using cumulative dose limits for OARs and all OAR cumulative dose objectives were met on the first optimization for the recurrent lung cancer case tested.

Improved Volumetric Modulated Arc Therapy Field Junctions Using In Silico Base Plans: Application to Craniospinal Irradiation.

INTRODUCTION: Field junctions present a major challenge for planning craniospinal irradiation (CSI) using volumetric modulated arc therapy (VMAT). In this study, the feasibility of using in silico base dose distributions for planning junctioned VMAT fields for CSI is assessed. METHODS: An in-house computer program was created to generate strategic base plans with controlled linear dose gradients across the junction. The algorithm was generalized to allow user-defined parameters such as number of junctions and junction length. In silico base plans were used to optimize junctioned VMAT CSI plans for a pediatric case and an adult case. Throughout optimization, dose to the eyes, kidneys, lungs, heart, and liver were minimized. Final plan quality was evaluated using the percent of planning target volume receiving at least 95% prescription dose (V95%), homogeneity index, and conformity number. Final plan robustness to setup error was evaluated using changes in near-minimum, median, and near-maximum doses defined as the most exposed 98%, 50%, and 2% of the planning target volume (D98%, D50%, D2%) within the junction region before and after setup errors of ±3, ±5, and ±7 mm in the craniocaudal direction. RESULTS: The program generated ideal in silico dose distributions that were compatible with a commercial treatment planning system for use as base doses during VMAT optimization. VMAT plans, that were optimized with the in silico base plans, had complementary linear dose profiles across the junction. Final pediatric and adult VMAT CSI plans both had V95% ≥98.1% and 98.1%, homogeneity index: 0.09 and 0.10, and conformity number: 0.86, 0.84, respectively. In addition, dose to surrounding organs at risk was acceptably low for both cases. For ±3 mm setup errors, small variations in the junction dose were recorded with ΔD98% ≤2.3%, ΔD50% ≤2.3%, and ΔD2% ≤2.8%. CONCLUSIONS: This is the first demonstration of junctioned VMAT field optimization with a controlled linear dose gradient across the junction without the use of any extra planning contours. Planning junctioned VMAT using in silico base plans is feasible and capable of generating high-quality plans that are robust to clinically expected setup variations.

Topiramate induces acute intracellular acidification in glioblastoma.

Reversal of the intracellular/extracellular pH gradient is a hallmark of malignant tumors and is an important consideration in evaluating tumor growth potential and the effectiveness of anticancer therapies. Glioblastoma multiforme (GBM) brain tumors have increased expression of the carbonic anhydrase (CA) isozymes CAII, CAIX and CAXII that contribute to the altered regulation of intracellular pH (pHi). The anti-epileptic drug topiramate (TPM) inhibits CA action and may acidify the tumor intracellular compartment. In-vivo detection of acute tumor acidification could aid in cancer diagnosis and monitoring treatment response. Chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) has been used to measure tissue pH. Using a recently developed CEST-MRI method called amine/amide concentration independent detection (AACID), we have previously shown intracellular acidification caused by single dose of lonidamine. The current study aims to evaluate the intracellular acidification induced by a single dose of the clinically approved drug TPM. Brain tumors were induced in NU/NU mice by injecting 105 U87 human glioblastoma multiforme cells into the right frontal lobe. Using a 9.4T MRI scanner AACID measurements were acquired, before and after administration of TPM (dose: 120 mg/kg, intraperitoneal), 15 ± 2 days after tumor cell implantation. TPM administration induced acute intracellular acidification (average ± SD: baseline AACID = 1.14 ± 0.05; post AACID = 1.19 ± 0.05, paired ttest p = 0.02) in implanted brain tumors. In contrast, contralateral tissue showed no change in AACID value. These results suggest that topiramate can rapidly induce a tumor specific physiological change detectable by AACID CEST. This pH challenge paradigm could be exploited to aid in tumor detection and monitoring treatment response.

Techniques for adaptive prostate radiotherapy.

Variations in the position and shape of the prostate make accurate setup and treatment challenging. Adaptive radiation therapy (ART) techniques seek to alter the treatment plan, at one or more points throughout the treatment course, in response to changes in patient anatomy observed between planning and pre-treatment images. This article reviews existing and developing ART techniques for prostate cancer along with an overview of supporting in-room imaging technologies. Challenges to the clinical implementation of adaptive radiotherapy are also discussed.

Assessment of Adaptive Margins Using a Single Planning Computed Tomography Scan for Bladder Radiotherapy.

INTRODUCTION: A "plan of the day" (PoD) adaptive radiotherapy approach is presented for bladder cancer. The potential reduction in volumes of normal tissue and bowel bag receiving high-dose radiation is evaluated. MATERIALS AND METHODS: Planning computed tomography (pCT) and daily cone beam CT (CBCT) data sets were analyzed for eight previously treated bladder cancer patients. For each patient, a whole bladder clinical target volume (CTV) was delineated on a pCT. Then, the clinical target volume was expanded using five sets of anisotropic or isotropic margins to create five planning target volumes (PTVs). A library of five corresponding treatment PoDs was then created using volumetric modulated arc therapy. Offline PoD selection was performed by three independent radiation therapists (RTs) using daily CBCTs. Dosimetric results were compared between PoD treatments and two conventional treatments using isotropic 1.5- and 2.0-cm margins. RESULTS: The smallest PTV using 1.0-cm isotropic margins was selected most frequently (70%). Three RTs demonstrated good agreement for daily PTV selections, choosing identical PoDs for 51% of all CBCTs. In addition, acceptable dosimetric coverage of the whole bladder was achieved for all PoD selections, similar to standard treatments. The average volume of bowel bag receiving 45 and 50 Gy and normal tissue receiving 95% prescription dose was significantly (P < .01) lower for PoD treatments compared with both conventional treatments. CONCLUSIONS: A PoD strategy using one pCT with isotropic and anisotropic margins can be used to treat bladder cancer and improve sparing of the bowel bag. Minimal dosimetric differences observed between three RTs suggests that PoD strategies are feasible for clinical implementation.

Introduction of Peripheral Carboxylates to Decrease the Charge on Tm(3+) DOTAM-Alkyl Complexes: Implications for Detection Sensitivity and in Vivo Toxicity of PARACEST MRI Contrast Agents.

A series of structurally modified Tm(3+) DOTAM-alkyl complexes as potential PARACEST MRI contrast agents has been synthesized with the aim to decrease the overall positive charge associated with these molecules and increase their biocompatibility. Two types of structural modification have been performed, an introduction of terminal carboxylate arms to the alkyl side chains and a conjugation of one of the alkyl side chains with aspartic acid. Detailed evaluation of the magnetic resonance imaging chemical exchange contrast associated with the structurally modified contrast agents has been performed. In contrast to the acutely toxic Tm(3+) DOTAM-alkyl complexes, the structurally modified compounds were found to be tolerated well during in vivo MRI studies in mice; however, only the aspartic acid modified chelates produced an amide proton-based PARACEST signal.

Imaging chemical exchange saturation transfer (CEST) effects following tumor-selective acidification using lonidamine.

Increased lactate production through glycolysis in aerobic conditions is a hallmark of cancer. Some anticancer drugs have been designed to exploit elevated glycolysis in cancer cells. For example, lonidamine (LND) inhibits lactate transport, leading to intracellular acidification in cancer cells. Chemical exchange saturation transfer (CEST) is a novel MRI contrast mechanism that is dependent on intracellular pH. Amine and amide concentration-independent detection (AACID) and apparent amide proton transfer (APT*) represent two recently developed CEST contrast parameters that are sensitive to pH. The goal of this study was to compare the sensitivity of AACID and APT* for the detection of tumor-selective acidification after LND injection. Using a 9.4-T MRI scanner, CEST data were acquired in mice approximately 14 days after the implantation of 10(5) U87 human glioblastoma multiforme (GBM) cells in the brain, before and after the administration of LND (dose, 50 or 100 mg/kg). Significant dose-dependent LND-induced changes in the measured CEST parameters were detected in brain regions spatially correlated with implanted tumors. Importantly, no changes were observed in T1- and T2-weighted images acquired before and after LND treatment. The AACID and APT* contrast measured before and after LND injection exhibited similar pH sensitivity. Interestingly, LND-induced contrast maps showed increased heterogeneity compared with pre-injection CEST maps. These results demonstrate that CEST contrast changes after the administration of LND could help to localize brain cancer and monitor tumor response to chemotherapy within 1 h of treatment. The LND CEST experiment uses an anticancer drug to induce a metabolic change detectable by endogenous MRI contrast, and therefore represents a unique cancer detection paradigm which differs from other current molecular imaging techniques that require the injection of an imaging contrast agent or tracer.

Quantitative tissue pH measurement during cerebral ischemia using amine and amide concentration-independent detection (AACID) with MRI.

Tissue pH is an indicator of altered cellular metabolism in diseases including stroke and cancer. Ischemic tissue often becomes acidic due to increased anaerobic respiration leading to irreversible cellular damage. Chemical exchange saturation transfer (CEST) effects can be used to generate pH-weighted magnetic resonance imaging (MRI) contrast, which has been used to delineate the ischemic penumbra after ischemic stroke. In the current study, a novel MRI ratiometric technique is presented to measure absolute pH using the ratio of CEST-mediated contrast from amine and amide protons: amine/amide concentration-independent detection (AACID). Effects of CEST were observed at 2.75 parts per million (p.p.m.) for amine protons and at 3.50 p.p.m. for amide protons downfield (i.e., higher frequency) from bulk water. Using numerical simulations and in vitro MRI experiments, we showed that pH measured using AACID was independent of tissue relaxation time constants, macromolecular magnetization transfer effects, protein concentration, and temperature within the physiologic range. After in vivo pH calibration using phosphorus ((31)P) magnetic resonance spectroscopy ((31)P-MRS), local acidosis is detected in mouse brain after focal permanent middle cerebral artery occlusion. In summary, our results suggest that AACID represents a noninvasive method to directly measure the spatial distribution of absolute pH in vivo using CEST MRI.

Simultaneous in vivo pH and temperature mapping using a PARACEST-MRI contrast agent.

Altered tissue temperature and/or pH is a common feature in pathological conditions, where metabolic demand exceeds oxygen supply such as in tumors and following stroke. Therefore, in vivo tissue temperature and pH may become valuable biomarkers for disease detection and the monitoring of disease progression or treatment response in conditions with altered metabolic demand. In this study, pH is measured using the amide protons of a thulium (Tm(3+)) complex with a DOTAM-Glycine-Lysine (ligand: Tm(3+)-DOTAM-Gly-Lys). The pH was uniquely determined from the linewidth of the asymmetry curve of the chemical exchange saturation transfer spectrum, independent of contrast agent concentration, or temperature for a given saturation pulse. pH maps with an inter-pixel standard deviation of less than 0.1 pH units were obtained in 10 mM Tm(3+)-DOTAM-Gly-Lys solutions with pH ranging from 6.0 to 8.0 pH units at 37°C. Temperature maps were simultaneously obtained using the chemical shift of the chemical exchange saturation transfer peak. Temperature and pH maps are demonstrated in the mouse leg (N = 3), where the mean and standard deviation for pH was 7.2 ± 0.2 pH unit and temperature was 37.4 ± 0.5°C.

Water-soluble gold nanoparticles (AuNP) functionalized with a gadolinium(iii) chelate via Michael addition for use as a MRI contrast agent.

A contrast agent suitable for magnetic resonance imaging based on small, water soluble gold nanoparticles (AuNP) conjugated to over 50 Gd3+ chelators has been prepared by using an interfacial Michael addition in aqueous media. The resultant chelator-AuNP conjugates have been successfully characterised by 1H NMR spectroscopy, IR spectroscopy, ICP-OES, ζ-potential analysis, TEM and MRI. T1-weighted in vivo images of mouse kidney were obtained using the agent at 9.4 T. A preliminary in vivo experiment produced no ill effects and the clearance profile of the agent suggests it is suitable for animal testing at clinically relevant concentrations.

Chaotic dynamics in cardiac aggregates induced by potassium channel block.

Chaotic rhythms in deterministic models can arise as a consequence of changes in model parameters. We carried out experimental studies in which we induced a variety of complex rhythms in aggregates of embryonic chick cardiac cells using E-4031 (1.0-2.5 µM), a drug that blocks the hERG potassium channel. Following the addition of the drug, the regular rhythm evolved to display a spectrum of complex dynamics: irregular rhythms, bursting oscillations, doublets, and accelerated rhythms. The interbeat intervals of the irregular rhythms can be described by one-dimensional return maps consistent with chaotic dynamics. A Hodgkin-Huxley-style cardiac ionic model captured the different types of complex dynamics following blockage of the hERG mediated potassium current.