A convolutional neural network for contouring metastatic lymph nodes on diffusion-weighted magnetic resonance images for assessment of radiotherapy response.

BACKGROUND AND PURPOSE: Retrieving quantitative parameters from magnetic resonance imaging (MRI), e.g. for early assessment of radiotherapy treatment response, necessitates contouring regions of interest, which is time-consuming and prone to errors. This becomes more pressing for daily imaging on MRI-guided radiotherapy systems. Therefore, we trained a deep convolutional neural network to automatically contour involved lymph nodes on diffusion-weighted (DW) MRI of head and neck cancer (HNC) patients receiving radiotherapy. MATERIALS AND METHODS: DW-images from 48 HNC patients (18 induction-chemotherapy + chemoradiotherapy; 30 definitive chemoradiotherapy) with 68 involved lymph nodes were obtained on a diagnostic 1.5 T MR-scanner prior to and 2-3 timepoints throughout treatment. A radiation oncologist delineated the lymph nodes on the b = 50 s/mm2 images. A 3D U-net was trained to contour involved lymph nodes. Its performance was evaluated in all 48 patients using 8-fold cross-validation and calculating the Dice similarity coefficient (DSC) and the absolute difference in median apparent diffusion coefficient (ΔADC) between the manual and generated contours. Additionally, the performance was evaluated in an independent dataset of three patients obtained on a 1.5 T MR-Linac. RESULTS: In the definitive chemoradiotherapy patients (n = 96 patients/lymphnodes/timepoints) the DSC was 0.87 (0.81-0.91) [median (1st-3rd quantiles)] and ΔADC was 1.9% (0.8-3.4%) and both remained stable throughout treatment. The network performed worse in the patients receiving induction-chemotherapy (n = 65), with DSC = 0.80 (0.71-0.87) and ΔADC = 3.3% (1.6-8.0%). The network performed well on the MR-Linac data (n = 8) with DSC = 0.80 (0.75-0.82) and ΔADC = 4.0% (0.6-9.1%). CONCLUSIONS: We established accurate automatic contouring of involved lymph nodes for HNC patients on diagnostic and MR-Linac DW-images.

MRI-based Assessment of 3D Intrafractional Motion of Head and Neck Cancer for Radiation Therapy.

PURPOSE: To determine the 3-dimensional (3D) intrafractional motion of head and neck squamous cell carcinoma (HNSCC). METHODS AND MATERIALS: Dynamic contrast-enhanced magnetic resonance images from 56 patients with HNSCC in the treatment position were analyzed. Dynamic contrast-enhanced magnetic resonance imaging consisted of 3D images acquired every 2.9 seconds for 4 minutes 50 seconds. Intrafractional tumor motion was studied in the 3 minutes 43 seconds of images obtained after initial contrast enhancement. To assess tumor motion, rigid registration (translations only) was performed using a region of interest (ROI) mask around the tumor. The results were compared with bulk body motion from registration to all voxels. Motion was split into systematic motion and random motion. Correlations between the tumor site and random motion were tested. The within-subject coefficient of variation was determined from 8 patients with repeated baseline measures. Random motion was also assessed at the end of the first week (38 patients) and second week (25 patients) of radiation therapy to investigate trends of motion. RESULTS: Tumors showed irregular occasional rapid motion (eg, swallowing or coughing), periodic intermediate motion (respiration), and slower systematic drifts throughout treatment. For 95% of the patients, displacements due to systematic and random motion were <1.4 mm and <2.1 mm, respectively, 95% of the time. The motion without an ROI mask was significantly (P<.0001, Wilcoxon signed rank test) less than the motion with an ROI mask, indicating that tumors can move independently from the bony anatomy. Tumor motion was significantly (P=.005, Mann-Whitney U test) larger in the hypopharynx and larynx than in the oropharynx. The within-subject coefficient of variation for random motion was 0.33. The average random tumor motion did not increase notably during the first 2 weeks of treatment. CONCLUSIONS: The 3D intrafractional tumor motion of HNSCC is small, with systematic motion <1.4 mm and random motion <2.1 mm 95% of the time.

Changes in multimodality functional imaging parameters early during chemoradiation predict treatment response in patients with locally advanced head and neck cancer.

OBJECTIVE: To assess the optimal timing and predictive value of early intra-treatment changes in multimodality functional and molecular imaging (FMI) parameters as biomarkers for clinical remission in patients receiving chemoradiation for head and neck squamous cell carcinoma (HNSCC). METHODS: Thirty-five patients with stage III-IVb (AJCC 7th edition) HNSCC prospectively underwent 18F-FDG-PET/CT, and diffusion-weighted (DW), dynamic contrast-enhanced (DCE) and susceptibility-weighted MRI at baseline, week 1 and week 2 of chemoradiation. Patients with evidence of persistent or recurrent disease during follow-up were classed as non-responders. Changes in FMI parameters at week 1 and week 2 were compared between responders and non-responders with the Mann-Whitney U test. The significance threshold was set at a p value of <0.05. RESULTS: There were 27 responders and 8 non-responders. Responders showed a greater reduction in PET-derived tumor total lesion glycolysis (TLG40%; p = 0.007) and maximum standardized uptake value (SUVmax; p = 0.034) after week 1 than non-responders but these differences were absent by week 2. In contrast, it was not until week 2 that MRI-derived parameters were able to discriminate between the two groups: larger fractional increases in primary tumor apparent diffusion coefficient (ADC; p < 0.001), volume transfer constant (Ktrans; p = 0.012) and interstitial space volume fraction (Ve; p = 0.047) were observed in responders versus non-responders. ADC was the most powerful predictor (∆ >17%, AUC 0.937). CONCLUSION: Early intra-treatment changes in FDG-PET, DW and DCE MRI-derived parameters are predictive of ultimate response to chemoradiation in HNSCC. However, the optimal timing for assessment with FDG-PET parameters (week 1) differed from MRI parameters (week 2). This highlighted the importance of scanning time points for the design of FMI risk-stratified interventional studies.

The emerging potential of magnetic resonance imaging in personalizing radiotherapy for head and neck cancer: an oncologist's perspective.

Head and neck cancer (HNC) is a challenging tumour site for radiotherapy delivery owing to its complex anatomy and proximity to organs at risk (OARs) such as the spinal cord and optic apparatus. Despite significant advances in radiotherapy planning techniques, radiation-induced morbidities remain substantial. Further improvement would require high-quality imaging and tailored radiotherapy based on intratreatment response. For these reasons, the use of MRI in radiotherapy planning for HNC is rapidly gaining popularity. MRI provides superior soft-tissue contrast in comparison with CT, allowing better definition of the tumour and OARs. The lack of additional radiation exposure is another attractive feature for intratreatment monitoring. In addition, advanced MRI techniques such as diffusion-weighted, dynamic contrast-enhanced and intrinsic susceptibility-weighted MRI techniques are capable of characterizing tumour biology further by providing quantitative functional parameters such as tissue cellularity, vascular permeability/perfusion and hypoxia. These functional parameters are known to have radiobiological relevance, which potentially could guide treatment adaptation based on their changes prior to or during radiotherapy. In this article, we first present an overview of the applications of anatomical MRI sequences in head and neck radiotherapy, followed by the potentials and limitations of functional MRI sequences in personalizing therapy.

Noninvasive Imaging of Cycling Hypoxia in Head and Neck Cancer Using Intrinsic Susceptibility MRI.

Purpose: To evaluate intrinsic susceptibility (IS) MRI for the identification of cycling hypoxia, and the assessment of its extent and spatial distribution, in head and neck squamous cell carcinoma (HNSCC) xenografts and patients.Experimental Design: Quantitation of the transverse relaxation rate, R2*, which is sensitive to paramagnetic deoxyhemoglobin, using serial IS-MRI acquisitions, was used to monitor temporal oscillations in levels of paramagnetic deoxyhemoglobin in human CALR xenografts and patients with HNSCC at 3T. Autocovariance and power spectrum analysis of variations in R2* was performed for each imaged voxel, to assess statistical significance and frequencies of cycling changes in tumor blood oxygenation. Pathologic correlates with tumor perfusion (Hoechst 33342), hypoxia (pimonidazole), and vascular density (CD31) were sought in the xenografts, and dynamic contrast-enhanced (DCE) MRI was used to assess patient tumor vascularization. The prevalence of fluctuations within patient tumors, DCE parameters, and treatment outcome were reported.Results: Spontaneous R2* fluctuations with a median periodicity of 15 minutes were detected in both xenografts and patient tumors. Spatially, these fluctuations were predominantly associated with regions of heterogeneous perfusion and hypoxia in the CALR xenografts. In patients, R2* fluctuations spatially correlated with regions of lymph nodes with low Ktrans values, typically in the vicinity of necrotic cores.Conclusions: IS-MRI can be used to monitor variations in levels of paramagnetic deoxyhemoglobin, associated with cycling hypoxia. The presence of such fluctuations may be linked with impaired tumor vasculature, the presence of which may impact treatment outcome. Clin Cancer Res; 23(15); 4233-41. ©2017 AACR.

A randomised controlled trial of Caphosol mouthwash in management of radiation-induced mucositis in head and neck cancer.

PURPOSE: This phase III, non-blinded, parallel-group, randomised controlled study evaluated the efficacy of Caphosol mouthwash in the management of radiation-induced oral mucositis (OM) in patients with head and neck cancer (HNC) undergoing radical (chemo)radiotherapy. PATIENTS AND METHODS: Eligible patients were randomised at 1:1 to Caphosol plus standard oral care (intervention) or standard oral care alone (control), stratified by radiotherapy technique and use of concomitant chemotherapy. Patients in the intervention arm used Caphosol for 7weeks: 6weeks during and 1-week post-radiotherapy. The primary endpoint was the incidence of severe OM (CTCAE ⩾grade 3) during and up to week 8 post-radiotherapy. Secondary endpoints include pharyngeal mucositis, dysphagia, pain and quality of life. RESULTS: The intervention (n=108) and control (n=107) arms were well balanced in terms of patient demographics and treatment characteristics. Following exclusion of patients with missing data, 210 patients were available for analysis. The incidence of severe OM did not differ between the intervention and control arms (64.1% versus 65.4%, p=0.839). Similarly, no significant benefit was observed for other secondary endpoints. Overall, compliance with the recommended frequency of Caphosol was low. CONCLUSION: Caphosol did not reduce the incidence or duration of severe OM during and after radiotherapy in HNC.

Functional Data Analysis Applied to Modeling of Severe Acute Mucositis and Dysphagia Resulting From Head and Neck Radiation Therapy.

PURPOSE: Current normal tissue complication probability modeling using logistic regression suffers from bias and high uncertainty in the presence of highly correlated radiation therapy (RT) dose data. This hinders robust estimates of dose-response associations and, hence, optimal normal tissue-sparing strategies from being elucidated. Using functional data analysis (FDA) to reduce the dimensionality of the dose data could overcome this limitation. METHODS AND MATERIALS: FDA was applied to modeling of severe acute mucositis and dysphagia resulting from head and neck RT. Functional partial least squares regression (FPLS) and functional principal component analysis were used for dimensionality reduction of the dose-volume histogram data. The reduced dose data were input into functional logistic regression models (functional partial least squares-logistic regression [FPLS-LR] and functional principal component-logistic regression [FPC-LR]) along with clinical data. This approach was compared with penalized logistic regression (PLR) in terms of predictive performance and the significance of treatment covariate-response associations, assessed using bootstrapping. RESULTS: The area under the receiver operating characteristic curve for the PLR, FPC-LR, and FPLS-LR models was 0.65, 0.69, and 0.67, respectively, for mucositis (internal validation) and 0.81, 0.83, and 0.83, respectively, for dysphagia (external validation). The calibration slopes/intercepts for the PLR, FPC-LR, and FPLS-LR models were 1.6/-0.67, 0.45/0.47, and 0.40/0.49, respectively, for mucositis (internal validation) and 2.5/-0.96, 0.79/-0.04, and 0.79/0.00, respectively, for dysphagia (external validation). The bootstrapped odds ratios indicated significant associations between RT dose and severe toxicity in the mucositis and dysphagia FDA models. Cisplatin was significantly associated with severe dysphagia in the FDA models. None of the covariates was significantly associated with severe toxicity in the PLR models. Dose levels greater than approximately 1.0 Gy/fraction were most strongly associated with severe acute mucositis and dysphagia in the FDA models. CONCLUSIONS: FPLS and functional principal component analysis marginally improved predictive performance compared with PLR and provided robust dose-response associations. FDA is recommended for use in normal tissue complication probability modeling.

The Predictive Value of Early Assessment After 1 Cycle of Induction Chemotherapy with 18F-FDG PET/CT and Diffusion-Weighted MRI for Response to Radical Chemoradiotherapy in Head and Neck Squamous Cell Carcinoma.

The objective of this study was to assess the predictive value of early assessment (after 1 cycle of induction chemotherapy [IC]) with 18F-FDG PET/CT and diffusion-weighted (DW) MRI for subsequent response to radical chemoradiotherapy in locally advanced head and neck squamous cell carcinoma (HNSCC). METHODS: Twenty patients with stage III-IVa HNSCC prospectively underwent 18F-FDG PET/CT and DW MRI before and 2 wk after each cycle of IC (first cycle, IC1; second cycle, IC2). Response was assessed 3 mo after completion of chemoradiotherapy with clinical examination, MRI, and 18F-FDG PET/CT. Patients with persistent disease were classed as nonresponders. Changes in functional and molecular imaging parameters after IC1 were compared between responders and nonresponders with the Mann-Whitney U test. The significance threshold was set at a P value of less than 0.05. RESULTS: Responders showed a significantly greater reduction in metabolic tumor volume (P = 0.03) and total lesion glycolysis (P = 0.04) after IC1 than nonresponders. Responders also showed a tendency toward a larger but statistically nonsignificant increase in apparent diffusion coefficient after IC1. There was no significant difference in the changes from baseline between the IC1 and IC2 for all functional and molecular imaging parameters, indicating that most biologic response to IC measured by 18F-FDG PET/CT and DW MRI was observed early after the first cycle of IC. CONCLUSION: Our preliminary data indicate that the 18F-FDG PET/CT-derived metabolic tumor volume or total lesion glycolysis, acquired after IC1, are early predictive biomarkers for ultimate response to subsequent chemoradiotherapy. These early biomarkers enable identification of patients at risk of treatment failure at an early time point, permitting treatment individualization and consideration of alternative strategies such as radiotherapy dose escalation or surgery.

Normal tissue complication probability (NTCP) modelling using spatial dose metrics and machine learning methods for severe acute oral mucositis resulting from head and neck radiotherapy.

BACKGROUND AND PURPOSE: Severe acute mucositis commonly results from head and neck (chemo)radiotherapy. A predictive model of mucositis could guide clinical decision-making and inform treatment planning. We aimed to generate such a model using spatial dose metrics and machine learning. MATERIALS AND METHODS: Predictive models of severe acute mucositis were generated using radiotherapy dose (dose-volume and spatial dose metrics) and clinical data. Penalised logistic regression, support vector classification and random forest classification (RFC) models were generated and compared. Internal validation was performed (with 100-iteration cross-validation), using multiple metrics, including area under the receiver operating characteristic curve (AUC) and calibration slope, to assess performance. Associations between covariates and severe mucositis were explored using the models. RESULTS: The dose-volume-based models (standard) performed equally to those incorporating spatial information. Discrimination was similar between models, but the RFCstandard had the best calibration. The mean AUC and calibration slope for this model were 0.71 (s.d.=0.09) and 3.9 (s.d.=2.2), respectively. The volumes of oral cavity receiving intermediate and high doses were associated with severe mucositis. CONCLUSIONS: The RFCstandard model performance is modest-to-good, but should be improved, and requires external validation. Reducing the volumes of oral cavity receiving intermediate and high doses may reduce mucositis incidence.

Assessment of fully-automated atlas-based segmentation of novel oral mucosal surface organ-at-risk.

BACKGROUND AND PURPOSE: Current oral mucositis normal tissue complication probability models, based on the dose distribution to the oral cavity volume, have suboptimal predictive power. Improving the delineation of the oral mucosa is likely to improve these models, but is resource intensive. We developed and evaluated fully-automated atlas-based segmentation (ABS) of a novel delineation technique for the oral mucosal surfaces. MATERIAL AND METHODS: An atlas of mucosal surface contours (MSC) consisting of 46 patients was developed. It was applied to an independent test cohort of 10 patients for whom manual segmentation of MSC structures, by three different clinicians, and conventional outlining of oral cavity contours (OCC), by an additional clinician, were also performed. Geometric comparisons were made using the dice similarity coefficient (DSC), validation index (VI) and Hausdorff distance (HD). Dosimetric comparisons were carried out using dose-volume histograms. RESULTS: The median difference, in the DSC and HD, between automated-manual comparisons and manual-manual comparisons were small and non-significant (-0.024; p=0.33 and -0.5; p=0.88, respectively). The median VI was 0.086. The maximum normalised volume difference between automated and manual MSC structures across all of the dose levels, averaged over the test cohort, was 8%. This difference reached approximately 28% when comparing automated MSC and OCC structures. CONCLUSIONS: Fully-automated ABS of MSC is suitable for use in radiotherapy dose-response modelling.

Repeatability and sensitivity of T2* measurements in patients with head and neck squamous cell carcinoma at 3T.

PURPOSE: To determine whether quantitation of T2* is sufficiently repeatable and sensitive to detect clinically relevant oxygenation levels in head and neck squamous cell carcinoma (HNSCC) at 3T. MATERIALS AND METHODS: Ten patients with newly diagnosed locally advanced HNSCC underwent two magnetic resonance imaging (MRI) scans between 24 and 168 hours apart prior to chemoradiotherapy treatment. A multiple gradient echo sequence was used to calculate T2* maps. A quadratic function was used to model the blood transverse relaxation rate as a function of blood oxygenation. A set of published coefficients measured at 3T were incorporated to account for tissue hematocrit levels and used to plot the dependence of fractional blood oxygenation (Y) on T2* values, together with the corresponding repeatability range. Repeatability of T2* using Bland-Altman analysis, and calculation of limits of agreement (LoA), was used to assess the sensitivity, defined as the minimum difference in fractional blood oxygenation that can be confidently detected. RESULTS: T2* LoA for 22 outlined tumor volumes were 13%. The T2* dependence of fractional blood oxygenation increases monotonically, resulting in increasing sensitivity of the method with increasing blood oxygenation. For fractional blood oxygenation values above 0.11, changes in T2* were sufficient to detect differences in blood oxygenation greater than 10% (Δ T2* > LoA for ΔY > 0.1). CONCLUSION: Quantitation of T2* at 3T can detect clinically relevant changes in tumor oxygenation within a wide range of blood volumes and oxygen tensions, including levels reported in HNSCC. J. Magn. Reson. Imaging 2016;44:72-80.

Prospective, longitudinal, multi-modal functional imaging for radical chemo-IMRT treatment of locally advanced head and neck cancer: the INSIGHT study.

BACKGROUND: Radical chemo-radiotherapy (CRT) is an effective organ-sparing treatment option for patients with locally advanced head and neck cancer (LAHNC). Despite advances in treatment for LAHNC, a significant minority of these patients continue to fail to achieve complete response with standard CRT. By constructing a multi-modality functional imaging (FI) predictive biomarker for CRT outcome for patients with LAHNC we hope to be able to reliably identify those patients at high risk of failing standard CRT. Such a biomarker would in future enable CRT to be tailored to the specific biological characteristics of each patients' tumour, potentially leading to improved treatment outcomes. METHODS/DESIGN: The INSIGHT study is a single-centre, prospective, longitudinal multi-modality imaging study using functional MRI and FDG-PET/CT for patients with LAHNC squamous cell carcinomas receiving radical CRT. Two cohorts of patients are being recruited: one treated with, and another treated without, induction chemotherapy. All patients receive radical intensity modulated radiotherapy with concurrent chemotherapy. Patients undergo functional imaging before, during and 3 months after completion of radiotherapy, as well as at the time of relapse, should that occur within the first two years after treatment. Serum samples are collected from patients at the same time points as the FI scans for analysis of a panel of serum markers of tumour hypoxia. DISCUSSION: The primary aim of the INSIGHT study is to acquire a prospective multi-parametric longitudinal data set comprising functional MRI, FDG PET/CT, and serum biomarker data from patients with LAHNC undergoing primary radical CRT. This data set will be used to construct a predictive imaging biomarker for outcome after CRT for LAHNC. This predictive imaging biomarker will be used in future studies of functional imaging based treatment stratification for patients with LAHNC. Additional objectives are: defining the reproducibility of FI parameters; determining robust methods for defining FI based biological target volumes for IMRT planning; creation of a searchable database of functional imaging data for data mining. The INSIGHT study will help to establish the role of FI in the clinical management of LAHNC. TRIAL REGISTRATION: NCRI H&N CSG ID 13860.