RESUMEN
BACKGROUND: Over half a million individuals are diagnosed with head and neck cancer each year globally. Radiotherapy is an important curative treatment for this disease, but it requires manual time to delineate radiosensitive organs at risk. This planning process can delay treatment while also introducing interoperator variability, resulting in downstream radiation dose differences. Although auto-segmentation algorithms offer a potentially time-saving solution, the challenges in defining, quantifying, and achieving expert performance remain. OBJECTIVE: Adopting a deep learning approach, we aim to demonstrate a 3D U-Net architecture that achieves expert-level performance in delineating 21 distinct head and neck organs at risk commonly segmented in clinical practice. METHODS: The model was trained on a data set of 663 deidentified computed tomography scans acquired in routine clinical practice and with both segmentations taken from clinical practice and segmentations created by experienced radiographers as part of this research, all in accordance with consensus organ at risk definitions. RESULTS: We demonstrated the model's clinical applicability by assessing its performance on a test set of 21 computed tomography scans from clinical practice, each with 21 organs at risk segmented by 2 independent experts. We also introduced surface Dice similarity coefficient, a new metric for the comparison of organ delineation, to quantify the deviation between organ at risk surface contours rather than volumes, better reflecting the clinical task of correcting errors in automated organ segmentations. The model's generalizability was then demonstrated on 2 distinct open-source data sets, reflecting different centers and countries to model training. CONCLUSIONS: Deep learning is an effective and clinically applicable technique for the segmentation of the head and neck anatomy for radiotherapy. With appropriate validation studies and regulatory approvals, this system could improve the efficiency, consistency, and safety of radiotherapy pathways.
Asunto(s)
Aprendizaje Profundo , Neoplasias de Cabeza y Cuello , Algoritmos , Neoplasias de Cabeza y Cuello/diagnóstico por imagen , Neoplasias de Cabeza y Cuello/radioterapia , Humanos , Tomografía Computarizada por Rayos XRESUMEN
Diffusion MRI characteristics assessed by apparent diffusion coefficient (ADC) histogram analysis in head and neck squamous cell carcinoma (HNSCC) have been reported as helpful in classifying tumours based on diffusion characteristics. There is little reported on HNSCC lymph nodes classification by diffusion characteristics. The aim of this study was to determine whether pretreatment nodal microstructural diffusion MRI characteristics can classify diseased nodes of patients with HNSCC from normal nodes of healthy volunteers. Seventy-nine patients with histologically confirmed HNSCC prior to chemoradiotherapy, and eight healthy volunteers, underwent diffusion-weighted (DW) MRI at a 1.5-T MR scanner. Two radiologists contoured lymph nodes on DW (b = 300 s/m2 ) images. ADC, distributed diffusion coefficient (DDC) and alpha (α) values were calculated by monoexponential and stretched exponential models. Histogram analysis metrics of drawn volume were compared between patients and volunteers using a Mann-Whitney test. The classification performance of each metric between the normal and diseased nodes was determined by receiver operating characteristic (ROC) analysis. Intraclass correlation coefficients determined interobserver reproducibility of each metric based on differently drawn ROIs by two radiologists. Sixty cancerous and 40 normal nodes were analysed. ADC histogram analysis revealed significant differences between patients and volunteers (p ≤0.0001 to 0.0046), presenting ADC distributions that were more skewed (1.49 for patients, 1.03 for volunteers; p = 0.0114) and 'peaked' (6.82 for patients, 4.20 for volunteers; p = 0.0021) in patients. Maximum ADC values exhibited the highest area under the curve ([AUC] 0.892). Significant differences were revealed between patients and volunteers for DDC and α value histogram metrics (p ≤0.0001 to 0.0044); the highest AUC were exhibited by maximum DDC (0.772) and the 25th percentile α value (0.761). Interobserver repeatability was excellent for mean ADC (ICC = 0.88) and the 25th percentile α value (ICC = 0.78), but poor for all other metrics. These results suggest that pretreatment microstructural diffusion MRI characteristics in lymph nodes, assessed by ADC and α value histogram analysis, can identify nodal disease.
Asunto(s)
Imagen de Difusión por Resonancia Magnética , Neoplasias de Cabeza y Cuello/diagnóstico por imagen , Voluntarios Sanos , Ganglios Linfáticos/diagnóstico por imagen , Cuello/diagnóstico por imagen , Carcinoma de Células Escamosas de Cabeza y Cuello/diagnóstico por imagen , Adulto , Anciano , Femenino , Humanos , Modelos Logísticos , Masculino , Persona de Mediana Edad , Variaciones Dependientes del Observador , Curva ROCRESUMEN
PURPOSE: About 40-60% of patients treated with post-operative radiotherapy for parotid cancer experience ipsilateral sensorineural hearing loss. Intensity-modulated radiotherapy (IMRT) can reduce radiation dose to the cochlea. COSTAR, a phase III trial, investigated the role of cochlear-sparing IMRT (CS-IMRT) in reducing hearing loss. METHODS: Patients (pT1-4 N0-3 M0) were randomly assigned (1:1) to 3-dimensional conformal radiotherapy (3DCRT) or CS-IMRT by minimisation, balancing for centre and radiation dose of 60Gy or 65Gy in 30 daily fractions. The primary end-point was proportion of patients with sensorineural hearing loss in the ipsilateral cochlea of ≥10 dB bone conduction at 4000 Hz 12 months after radiotherapy compared using Fisher's exact test. Secondary end-points included hearing loss at 6 and 24 months, balance assessment, acute and late toxicity, patient-reported quality of life, time to recurrence and survival. RESULTS: From Aug 2008 to Feb 2013, 110 patients (54 3DCRT; 56 CS-IMRT) were enrolled from 22 UK centres. Median doses to the ipsilateral cochlea were 3DCRT: 56.2Gy and CS-IMRT: 35.7Gy (p < 0.0001). 67/110 (61%) patients were evaluable for the primary end-point; main reasons for non-evaluability were non-attendance at follow-up or incomplete audiology assessment. At 12 months, 14/36 (39%) 3DCRT and 11/31 (36%) CS-IMRT patients had ≥10 dB loss (p = 0.81). No statistically significant differences were observed in hearing loss at 6 or 24 months or in other secondary end-points including patient-reported hearing outcomes. CONCLUSION: CS-IMRT reduced the radiation dose below the accepted tolerance of the cochlea, but this did not lead to a reduction in the proportion of patients with clinically relevant hearing loss.
Asunto(s)
Neoplasias de la Parótida/radioterapia , Radioterapia de Intensidad Modulada/métodos , Adulto , Anciano , Anciano de 80 o más Años , Femenino , Humanos , Masculino , Persona de Mediana Edad , Adulto JovenRESUMEN
BACKGROUND: Outcomes for patients in UK with locally advanced non-small cell lung cancer (LA NSCLC) are amongst the worst in Europe. Assessing outcomes is important for analysing the effectiveness of current practice. However, data quality is inconsistent and regular large scale analysis is challenging. This project investigates the use of routine healthcare datasets to determine progression free survival (PFS) and overall survival (OS) of patients treated with primary radical radiotherapy for LA NSCLC. METHODS: All LA NSCLC patients treated with primary radical radiotherapy in a 2 year period were identified and paired manual and routine data generated for an initial pilot study. Manual data was extracted information from hospital records and considered the gold standard. Key time points were date of diagnosis, recurrence, death or last clinical encounter. Routine data was collected from various data sources including, Hospital Episode Statistics, Personal Demographic Service, chemotherapy data, and radiotherapy datasets. Relevant event dates were defined by proxy time points and refined using backdating and time interval optimization. Dataset correlations were then tested on key clinical outcome indicators to establish if routine data could be used as a reliable proxy measure for manual data. RESULTS: Forty-three patients were identified for the pilot study. The manual data showed a median age of 67 years (range 46- 89 years) and all patients had stage IIIA/B disease. Using the manual data, the median PFS was 10.78 months (range 1.58-37.49 months) and median OS was 16.36 months (range 2.69-37.49 months). Based on routine data, using proxy measures, the estimated median PFS was 10.68 months (range 1.61-31.93 months) and estimated median OS was 15.38 months (range 2.14-33.71 months). Overall, the routine data underestimated the PFS and OS of the manual data but there was good correlation with a Pearson correlation coefficient of 0.94 for PFS and 0.97 for OS. CONCLUSIONS: This is a novel approach to use routine datasets to determine outcome indicators in patients with LA NSCLC that will be a surrogate to analysing manual data. The ability to enable efficient and large scale analysis of current lung cancer strategies has a huge potential impact on the healthcare system.
Asunto(s)
Carcinoma de Pulmón de Células no Pequeñas/mortalidad , Neoplasias Pulmonares/mortalidad , Anciano , Anciano de 80 o más Años , Carcinoma de Pulmón de Células no Pequeñas/terapia , Supervivencia sin Enfermedad , Europa (Continente)/epidemiología , Femenino , Humanos , Neoplasias Pulmonares/terapia , Masculino , Persona de Mediana Edad , Recurrencia Local de Neoplasia/mortalidad , Evaluación de Resultado en la Atención de Salud , Proyectos Piloto , Pronóstico , Estudios Prospectivos , Reino Unido/epidemiologíaRESUMEN
PURPOSE: The aim of this study was to assess whether clinically acceptable segmentations of organs at risk (OARs) in head and neck cancer can be obtained automatically and efficiently using the novel "similarity and truth estimation for propagated segmentations" (STEPS) compared to the traditional "simultaneous truth and performance level estimation" (STAPLE) algorithm. METHODS: First, 6 OARs were contoured by 2 radiation oncologists in a dataset of 100 patients with head and neck cancer on planning computed tomography images. Each image in the dataset was then automatically segmented with STAPLE and STEPS using those manual contours. Dice similarity coefficient (DSC) was then used to compare the accuracy of these automatic methods. Second, in a blind experiment, three separate and distinct trained physicians graded manual and automatic segmentations into one of the following three grades: clinically acceptable as determined by universal delineation guidelines (grade A), reasonably acceptable for clinical practice upon manual editing (grade B), and not acceptable (grade C). Finally, STEPS segmentations graded B were selected and one of the physicians manually edited them to grade A. Editing time was recorded. RESULTS: Significant improvements in DSC can be seen when using the STEPS algorithm on large structures such as the brainstem, spinal canal, and left/right parotid compared to the STAPLE algorithm (all p < 0.001). In addition, across all three trained physicians, manual and STEPS segmentation grades were not significantly different for the brainstem, spinal canal, parotid (right/left), and optic chiasm (all p > 0.100). In contrast, STEPS segmentation grades were lower for the eyes (p < 0.001). Across all OARs and all physicians, STEPS produced segmentations graded as well as manual contouring at a rate of 83%, giving a lower bound on this rate of 80% with 95% confidence. Reduction in manual interaction time was on average 61% and 93% when automatic segmentations did and did not, respectively, require manual editing. CONCLUSIONS: The STEPS algorithm showed better performance than the STAPLE algorithm in segmenting OARs for radiotherapy of the head and neck. It can automatically produce clinically acceptable segmentation of OARs, with results as relevant as manual contouring for the brainstem, spinal canal, the parotids (left/right), and optic chiasm. A substantial reduction in manual labor was achieved when using STEPS even when manual editing was necessary.
Asunto(s)
Algoritmos , Neoplasias de Cabeza y Cuello/diagnóstico por imagen , Procesamiento de Imagen Asistido por Computador/métodos , Neoplasias de Cabeza y Cuello/radioterapia , Humanos , Órganos en Riesgo/efectos de la radiación , Planificación de la Radioterapia Asistida por Computador , Radioterapia de Intensidad Modulada/efectos adversos , Tomografía Computarizada por Rayos XRESUMEN
PURPOSE: Multiexponential decay parameters are estimated from diffusion-weighted-imaging that generally have inherently low signal-to-noise ratio and non-normal noise distributions, especially at high b-values. Conventional nonlinear regression algorithms assume normally distributed noise, introducing bias into the calculated decay parameters and potentially affecting their ability to classify tumors. This study aims to accurately estimate noise of averaged diffusion-weighted-imaging, to correct the noise induced bias, and to assess the effect upon cancer classification. METHODS: A new adaptation of the median-absolute-deviation technique in the wavelet-domain, using a closed form approximation of convolved probability-distribution-functions, is proposed to estimate noise. Nonlinear regression algorithms that account for the underlying noise (maximum probability) fit the biexponential/stretched exponential decay models to the diffusion-weighted signal. A logistic-regression model was built from the decay parameters to discriminate benign from metastatic neck lymph nodes in 40 patients. RESULTS: The adapted median-absolute-deviation method accurately predicted the noise of simulated (R(2) = 0.96) and neck diffusion-weighted-imaging (averaged once or four times). Maximum probability recovers the true apparent-diffusion-coefficient of the simulated data better than nonlinear regression (up to 40%), whereas no apparent differences were found for the other decay parameters. CONCLUSIONS: Perfusion-related parameters were best at cancer classification. Noise-corrected decay parameters did not significantly improve classification for the clinical data set though simulations show benefit for lower signal-to-noise ratio acquisitions.
Asunto(s)
Carcinoma de Células Escamosas/patología , Imagen de Difusión por Resonancia Magnética/métodos , Neoplasias de Cabeza y Cuello/patología , Metástasis Linfática , Adulto , Anciano , Algoritmos , Humanos , Aumento de la Imagen/métodos , Interpretación de Imagen Asistida por Computador/métodos , Persona de Mediana Edad , Modelos Teóricos , Relación Señal-RuidoRESUMEN
Head and neck cancer refers to a group of malignancies that affects the epithelium of the upper aereodigestive tract, primarily the lip and mouth, pharynx and larynx. Head and neck cancer is strongly associated with tobacco smoking, alcohol consumption and betel nut chewing, and indeed a reduction in the exposure to these risk factors has determined a recent decrease in incidence rates in many countries. There remains, however, a significant increase in head and neck cancer rates in those regions where tobacco epidemic continues, as well as in the number of oral cancers related to HPV infection (in particular cancer of the oropharynx, tonsil, and base of the tongue), which typically affect young adults with no history of exposure to tobacco or alcohol. Treatment of head and neck cancer has significantly changed during the last few decades, and an increasing number of individuals are currently offered combined chemoradiotherapy as single treatment modality for organ preservation or in association with surgery to improve prognosis. Unfortunately, the majority of head and neck cancer patients eventually succumb to their disease, with inoperable locoregional recurrences and lack of response to chemoradiotherapy representing the main causes of death. There is an urgent need of novel molecular-targeted therapeutics that could overcome the limitations of current treatment modalities. This paper reviews the characteristics of a novel group of promising antineoplastic agents, poly (ADP-ribose) polymerases (PARP) inhibitors, which cleverly target one of the mechanisms cancer cells use to escape the toxic effect of chemoradiation, and describe the potential benefits of their addition to current limited range of head and neck cancer antineoplastic agents.
Asunto(s)
Antineoplásicos/farmacología , Neoplasias de Cabeza y Cuello/tratamiento farmacológico , Inhibidores de Poli(ADP-Ribosa) Polimerasas , Consumo de Bebidas Alcohólicas/efectos adversos , Animales , Areca/efectos adversos , Neoplasias de Cabeza y Cuello/etiología , Neoplasias de Cabeza y Cuello/patología , Humanos , Terapia Molecular Dirigida , Factores de Riesgo , Fumar/efectos adversosRESUMEN
BACKGROUND: The effectiveness of ABC has been traditionally measured as the reduction in internal margin (IM) within the planning target volume (PTV). Not to overestimate the benefit of ABC, the effect of patient movement during treatment also needs to be taken into account. We determined the IM and set-up error with ABC and the effect on physical lung parameters compared to standard margins used with free breathing. We also assessed interfraction oesophageal movement to determine a planning organ at risk volume (PRV). MATERIALS AND METHODS: Two sequential studies were performed using ABC in NSCLC patients suitable for radical radiotherapy (RT). Twelve out of 14 patients in Study 1 had tumours visible fluoroscopically and had intrafraction tumour movement assessed with and without ABC. Sixteen patients were recruited to Study 2 and had interfraction tumour movement measured using ABC in a moderate deep inspiration breath-hold, of these 7 patients also had interfraction oesophageal movement recorded. Interfraction movement was assessed by CT scan prior to and in the middle and final week of RT. Displacement of the tumour centre of mass and oesophageal borders relative to the first scan provided a measure of movement. Set-up error was measured in 9 patients treated with an in-house lung board adapted for the ABC device. Combining movement and set-up errors determined PTV and PRV margins with ABC. The effect of ABC on mean lung dose (MLD), lung V20 and V13 was calculated. RESULTS: ABC in a moderate deep inspiration breath-hold was tolerated in 25 out of 30 patients (83%) in Study 1 and 2. The random contribution of periodic tumour motion was reduced by 90% in the y direction with ABC compared to free-breathing. The magnitude of motion reduction was less in the x and z direction. Combining the systematic and random set-up error in quadrature with the systematic and random intrafraction and interfraction tumour variations with ABC results in a PTV margin of 8.3mm in the x direction, 12.0mm in the y direction and 9.8mm in the z direction. There was a relative mean reduction in MLD, lung V20 and V13 of 25%, 21% and 18% with the ABC PTV compared to a free-breathing PTV. Oesophageal movement combined with set-up error resulted in an isotropic PRV of 4.7 mm. CONCLUSIONS: The reduction in PTV size with ABC resulted in an 18-25% relative reduction in physical lung parameters. PTV margin reduction has the potential to spare normal lung and allow dose-escalation if coupled with image-guided RT. The oesophageal PRV needs to be considered when irradiating central disease and is of increasing importance with altered RT fractionation and concomitant chemoradiation schedules. Further reductions in PTV and PRV may be possible if patient set-up error was minimised, confirming that attention to patient immobilisation is as important as attempts to control tumour motion.
Asunto(s)
Carcinoma de Pulmón de Células no Pequeñas/radioterapia , Neoplasias Pulmonares/radioterapia , Respiración , Anciano , Anciano de 80 o más Años , Carcinoma de Pulmón de Células no Pequeñas/diagnóstico por imagen , Carcinoma de Pulmón de Células no Pequeñas/patología , Femenino , Fluoroscopía , Humanos , Inmovilización , Neoplasias Pulmonares/diagnóstico por imagen , Neoplasias Pulmonares/patología , Masculino , Persona de Mediana Edad , Planificación de la Radioterapia Asistida por Computador , Estadísticas no Paramétricas , Tomografía Computarizada por Rayos XRESUMEN
The forward and inverse treatment plans of 10 patients with lung cancer were compared in terms of PTV coverage, sparing of normal lung and time required to generate a plan. The inverse planning produced as good treatment plans as an experienced dosimetrist with considerable reduction in staff time. When translated to other complex sites, inverse non-IMRT planning may have considerable impact on manpower requirements.
Asunto(s)
Carcinoma de Pulmón de Células no Pequeñas/radioterapia , Imagenología Tridimensional/métodos , Neoplasias Pulmonares/radioterapia , Modelos Biológicos , Radiometría/métodos , Planificación de la Radioterapia Asistida por Computador/métodos , Radioterapia Conformacional/métodos , Algoritmos , Carga Corporal (Radioterapia) , Carcinoma de Pulmón de Células no Pequeñas/fisiopatología , Simulación por Computador , Humanos , Neoplasias Pulmonares/fisiopatología , Dosificación Radioterapéutica , Efectividad Biológica Relativa , Reproducibilidad de los Resultados , Sensibilidad y EspecificidadRESUMEN
PURPOSE: To compare static electron therapy, electron arc therapy, and photon intensity-modulated radiation therapy (IMRT) for treatment of extensive scalp lesions and to examine the dosimetric accuracy of the techniques. METHODS AND MATERIALS: A retrospective treatment-planning study was performed to evaluate the relative merits of static electron fields, arcing electron fields, and five-field photon IMRT. Thermoluminescent dosimeters (TLD) were used to verify the accuracy of the techniques. The required thickness of bolus was investigated, and an anthropomorphic phantom was also used to examine the effects of air gaps between the wax bolus used for the IMRT technique and the patient's scalp. RESULTS: Neither static nor arcing electron techniques were able to provide a reliable coverage of the planning target volume (PTV), owing to obliquity of the fields in relation to the scalp. The IMRT technique considerably improved PTV dose uniformity, though it irradiated a larger volume of brain. Either 0.5 cm or 1.0 cm of wax bolus was found to be suitable. Air gaps of up to 1 cm between the bolus and the patient's scalp were correctly handled by the treatment-planning system and had negligible influence on the dose to the scalp. CONCLUSIONS: Photon IMRT provides a feasible alternative to electron techniques for treatment of large scalp lesions, resulting in improved homogeneity of dose to the PTV but with a moderate increase in dose to the brain.