Uncertainty estimation methods are expected to improve the understanding and quality of computer-assisted methods used in medical applications (e.g., neurosurgical interventions, radiotherapy planning), where automated medical image segmentation is crucial. In supervised machine learning, a common practice to generate ground truth label data is to merge observer annotations. However, as many medical image tasks show a high inter-observer variability resulting from factors such as image quality, different levels of user expertise and domain knowledge, little is known as to how inter-observer variability and commonly used fusion methods affect the estimation of uncertainty of automated image segmentation. In this paper we analyze the effect of common image label fusion techniques on uncertainty estimation, and propose to learn the uncertainty among observers. The results highlight the negative effect of fusion methods applied in deep learning, to obtain reliable estimates of segmentation uncertainty. Additionally, we show that the learned observers' uncertainty can be combined with current standard Monte Carlo dropout Bayesian neural networks to characterize uncertainty of model's parameters.

Background: Automated brain tumor segmentation methods are computational algorithms that yield tumor delineation from, in this case, multimodal magnetic resonance imaging (MRI). We present an automated segmentation method and its results for resection cavity (RC) in glioblastoma multiforme (GBM) patients using deep learning (DL) technologies. Methods: Post-operative, T1w with and without contrast, T2w and fluid attenuated inversion recovery MRI studies of 30 GBM patients were included. Three radiation oncologists manually delineated the RC to obtain a reference segmentation. We developed a DL cavity segmentation method, which utilizes all four MRI sequences and the reference segmentation to learn to perform RC delineations. We evaluated the segmentation method in terms of Dice coefficient (DC) and estimated volume measurements. Results: Median DC of the three radiation oncologist were 0.85 (interquartile range [IQR]: 0.08), 0.84 (IQR: 0.07), and 0.86 (IQR: 0.07). The results of the automatic segmentation compared to the three different raters were 0.83 (IQR: 0.14), 0.81 (IQR: 0.12), and 0.81 (IQR: 0.13) which was significantly lower compared to the DC among raters (chisquare = 11.63, p = 0.04). We did not detect a statistically significant difference of the measured RC volumes for the different raters and the automated method (Kruskal-Wallis test: chi-square = 1.46, p = 0.69). The main sources of error were due to signal inhomogeneity and similar intensity patterns between cavity and brain tissues. Conclusions: The proposed DL approach yields promising results for automated RC segmentation in this proof of concept study. Compared to human experts, the DC are still subpar.

IntroductionA wide variety of fractionation schedules have been employed for the treatment of early glottic cancer. The aim is to report our 10-year experience of using hypofractionated radiotherapy with 55Gy in 20 fractions at 2.75Gy per fraction.MethodsPatients treated between 2004 and 2013 with definitive radiotherapy to a dose of 55Gy in 20 fractions over 4 weeks for T1/2 N0 squamous cell carcinoma of the glottis were retrospectively identified. Patients with prior therapeutic minor surgery (eg. laser stripping, cordotomy) were included. The probabilities of local control, ultimate local control (including salvage surgery), regional control, cause specific survival (CSS) and overall survival (OS) were calculated.ResultsOne hundred thirty-two patients were identified. Median age was 65 years (range 33–89). Median follow up was 72 months (range 7–124). 50 (38 %), 18 (14 %) and 64 (48 %) of patients had T1a, T1b and T2 disease respectively. Five year local control and ultimate local control rates were: overall - 85.6 % and 97.3 % respectively, T1a - 91.8 % and 100 %, T1b - 81.6 and 93.8 %, and T2 - 80.9 % and 95.8 %. Five year regional control, CSS and OS rates were 95.4 %, 95.7 % and 78.8 % respectively. There were no significant associations of covariates (e.g. T-stage, extent of laryngeal extension, histological grade) with local control on univariate analysis. Only increasing age and transglottic extension in T2 disease were significantly associated with overall survival (both p <0.01). Second primary cancers developed in 17 % of patients. 13 (9.8 %) of patients required enteral tube feeding support during radiotherapy; no patients required long term enteral nutrition. One patient required a tracheostomy due to a non-functioning larynx on long term follow up.ConclusionsHypofractionated radiation therapy with a dose of 55Gy in 20 fractions for early stage glottic cancer provides high rates of local control with acceptable toxicity.