Jamie A. Dean

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.

Neurocognitive Function After (Chemo)-Radiotherapy for Head and Neck Cancer

Radical radiotherapy has a pivotal role in the treatment of head and neck cancer (HNC) and cures a significant proportion of patients while simultaneously sparing critical normal organs. Some patients treated with radical radiotherapy for HNC receive significant radiation doses to large volumes of brain tissue. In fact, intensity-modulated radiotherapy techniques for HNC have been associated with a net increase in irradiated brain volumes. The increasing use of chemoradiotherapy for HNC has additionally exposed this patient population to potential neurotoxicity due to cytotoxic drugs. Patients with HNC may be particularly at risk for adverse late brain effects after (chemo)-radiotherapy, such as impaired neurocognitive function (NCF), as risk factors for the development of HNC, such as smoking, excess alcohol consumption and poor diet, are also associated with impaired NCF. The relatively good survival rates with modern treatment for HNC, and exposure to multiple potentially neurotoxic factors, means that it is important to understand the impact of (chemo)-radiotherapy for HNC on NCF, and to consider what measures can be taken to minimise treatment-related neurotoxicity. Here, we review evidence relating to the late neurotoxicity of radical (chemo)-radiotherapy for HNC, with a focus on studies of NCF in this patient population.

A novel method for delineation of oral mucosa for radiotherapy dose-response studies

There is currently no standard method for delineating the oral mucosa and most attempts are oversimplified. A new method to obtain anatomically accurate contours of the oral mucosa surfaces was developed and applied to 11 patients. This is expected to represent an opportunity for improved toxicity modelling of oral mucositis.

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.

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.

Brain-Sparing Methods for IMRT of Head and Neck Cancer

Purpose Radical radiotherapy for head and neck cancer (HNC) may deliver significant doses to brain structures. There is evidence that this may cause a decline in neurocognitive function (NCF). Radiation dose to the medial temporal lobes, and particularly to the hippocampi, seems to be critical in determining NCF outcomes. We evaluated the feasibility of two alternative intensity-modulated radiotherapy (IMRT) techniques to generate hippocampus- and brain-sparing HNC treatment plans to preserve NCF. Methods and Materials A planning study was undertaken for ten patients with HNC whose planning target volume (PTV) included the nasopharynx. Patients had been previously treated using standard (chemo)-IMRT techniques. Bilateral hippocampi were delineated according to the RTOG atlas, on T1w MRI co-registered to the RT planning CT. Hippocampus-sparing plans (HSRT), and whole-brain/hippocampus-sparing fixed-field non-coplanar IMRT (BSRT) plans, were generated. DVHs and dose difference maps were used to compare plans. NTCP calculations for NCF impairment, based on hippocampal dosimetry, were performed for all plans. Results Significant reductions in hippocampal doses relative to standard plans were achieved in eight of ten cases for both HSRT and BSRT. EQD2 D40% to bilateral hippocampi was significantly reduced from a mean of 23.5 Gy (range 14.5–35.0) in the standard plans to a mean of 8.6 Gy (4.2–24.7) for HSRT (p = 0.001) and a mean of 9.0 Gy (4.3–17.3) for BSRT (p < 0.001). Both HSRT and BSRT resulted in a significant reduction in doses to the whole brain, brain stem, and cerebellum. Conclusion We demonstrate that IMRT plans for HNC involving the nasopharynx can be successfully optimised to significantly reduce dose to the bilateral hippocampi and whole brain. The magnitude of the achievable dose reductions results in significant reductions in the probability of radiation-induced NCF decline. These results could readily be translated into a future clinical trial.

Normal Tissue Complication Probability (NTCP) Modelling of Severe Acute Mucositis using a Novel Oral Mucosal Surface Organ at Risk

AIMS A normal tissue complication probability (NTCP) model of severe acute mucositis would be highly useful to guide clinical decision making and inform radiotherapy planning. We aimed to improve upon our previous model by using a novel oral mucosal surface organ at risk (OAR) in place of an oral cavity OAR. MATERIALS AND METHODS Predictive models of severe acute mucositis were generated using radiotherapy dose to the oral cavity OAR or mucosal surface OAR and clinical data. Penalised logistic regression and random forest classification (RFC) models were generated for both OARs and compared. Internal validation was carried out with 100-iteration stratified shuffle split cross-validation, using multiple metrics to assess different aspects of model performance. Associations between treatment covariates and severe mucositis were explored using RFC feature importance. RESULTS Penalised logistic regression and RFC models using the oral cavity OAR performed at least as well as the models using mucosal surface OAR. Associations between dose metrics and severe mucositis were similar between the mucosal surface and oral cavity models. The volumes of oral cavity or mucosal surface receiving intermediate and high doses were most strongly associated with severe mucositis. CONCLUSIONS The simpler oral cavity OAR should be preferred over the mucosal surface OAR for NTCP modelling of severe mucositis. We recommend minimising the volume of mucosa receiving intermediate and high doses, where possible.

Incorporating spatial dose metrics in machine learning-based normal tissue complication probability (NTCP) models of severe acute dysphagia resulting from head and neck radiotherapy

Highlights • Machine learning-based NTCP modelling of acute dysphagia was performed.• The models generated performed well on internal and external validation.• Doses of approximately 1 Gy/fraction were most strongly associated with severe dysphagia.• No spatial variation in radiosensitivity was observed for the pharyngeal mucosa.• These results could inform clinical decision-support and radiotherapy planning.

Quality assurance in MRI breast screening: comparing signal-to-noise ratio in dynamic contrast-enhanced imaging protocols

Abstract MRI has been extensively used in breast cancer staging, management and high risk screening. Detection sensitivity is paramount in breast screening, but variations of signal-to-noise ratio (SNR) as a function of position are often overlooked. We propose and demonstrate practical methods to assess spatial SNR variations in dynamic contrast-enhanced (DCE) breast examinations and apply those methods to different protocols and systems. Four different protocols in three different MRI systems (1.5 and 3.0 T) with receiver coils of different design were employed on oil-filled test objects with and without uniformity filters. Twenty 3D datasets were acquired with each protocol; each dataset was acquired in under 60 s, thus complying with current breast DCE guidelines. In addition to the standard SNR calculated on a pixel-by-pixel basis, we propose other regional indices considering the mean and standard deviation of the signal over a small sub-region centred on each pixel. These regional indices include effects of the spatial variation of coil sensitivity and other structured artefacts. The proposed regional SNR indices demonstrate spatial variations in SNR as well as the presence of artefacts and sensitivity variations, which are otherwise difficult to quantify and might be overlooked in a clinical setting. Spatial variations in SNR depend on protocol choice and hardware characteristics. The use of uniformity filters was shown to lead to a rise of SNR values, altering the noise distribution. Correlation between noise in adjacent pixels was associated with data truncation along the phase encoding direction. Methods to characterise spatial SNR variations using regional information were demonstrated, with implications for quality assurance in breast screening and multi-centre trials.

OC-0257: NTCP models for acute dysphagia resulting from (chemo)radiotherapy for head and neck cancer

incontinence and mucus loss in the rectum mapping. For the IG-IMRT patients only mucus loss was significant in the anus mapping, while both mucus loss and proctitis were significant in the rectum mapping. Results for proctitis are illustrated for 3D-CRT (44% incidence) in the anus mapping (Fig. 1e, p=0.01) and for IG-IMRT (27% incidence) in the rectum mapping (Fig. 1f, p<0.01). Note that the largest dose differences are not necessarily most significant. In a region with large dose variations, random permutations can lead to large differences by chance. Figs. 1c and 1e indicate an effect of the extent of the intermediate dose region around the circumference of the anus. Figs. 1b and 1f show the effect of a longer intermediate dose region along the central axis. Conclusions: Significant differences in the local dose to rectal and anal surfaces were found between patients with or without various GI toxicities. The locations of such differences provide clues about variables which are relevant to the clinical outcome, and which may serve as a basis for subsequent dose-effect modeling.