Uncertainty estimation for deep learning-based pectoral muscle segmentation via Monte Carlo dropout.

Objective. Deep Learning models are often susceptible to failures after deployment. Knowing when your model is producing inadequate predictions is crucial. In this work, we investigate the utility of Monte Carlo (MC) dropout and the efficacy of the proposed uncertainty metric (UM) for flagging of unacceptable pectoral muscle segmentations in mammograms.Approach. Segmentation of pectoral muscle was performed with modified ResNet18 convolutional neural network. MC dropout layers were kept unlocked at inference time. For each mammogram, 50 pectoral muscle segmentations were generated. The mean was used to produce the final segmentation and the standard deviation was applied for the estimation of uncertainty. From each pectoral muscle uncertainty map, the overall UM was calculated. To validate the UM, a correlation between the dice similarity coefficient (DSC) and UM was used. The UM was first validated in a training set (200 mammograms) and finally tested in an independent dataset (300 mammograms). ROC-AUC analysis was performed to test the discriminatory power of the proposed UM for flagging unacceptable segmentations.Main results. The introduction of dropout layers in the model improved segmentation performance (DSC = 0.95 ± 0.07 versus DSC = 0.93 ± 0.10). Strong anti-correlation (r= -0.76,p< 0.001) between the proposed UM and DSC was observed. A high AUC of 0.98 (97% specificity at 100% sensitivity) was obtained for the discrimination of unacceptable segmentations. Qualitative inspection by the radiologist revealed that images with high UM are difficult to segment.Significance. The use of MC dropout at inference time in combination with the proposed UM enables flagging of unacceptable pectoral muscle segmentations from mammograms with excellent discriminatory power.

[Costs of potentially outpatient endoscopic procedures in cases with a 1-day hospital stay versus a longer stay]. / Kosten potenziell ambulant erbringbarer endoskopischer Leistungen in Fällen mit 1-Tages-Verweildauer gegenüber einer längeren Verweildauer.

INTRODUCTION: The transfer of patient care and medical interventions that was previously provided on an inpatient basis to outpatient settings is a stated goal of health politics. It is unclear to what extent costs of an endoscopic procedure and the disease severity depend on the duration of inpatient treatment. We therefore examined whether endoscopic services for cases with a one-day length of stay (VWD) are comparably expensive to cases with a longer VWD. METHODS: Outpatient services were selected from the DGVS service catalog. Day cases with exactly one such gastroenterological endoscopic (GAEN) service were compared with cases with VWD>1 day regarding their patient clinical complexity levels (PCCL) and mean costs. Data from the DGVS-DRG project with §21-KHEntgG cost data from a total of 57 hospitals from 2018 and 2019 served as the basis. Endoscopic costs were taken from cost center group 8 of the InEK cost matrix and plausibility checked. RESULTS: A total of 122,514 cases with exactly one GAEN service were identified. Statistically equal costs were shown in 30 of 47 service groups. In 10 groups, the cost difference was not relevant (<10%). Cost differences >10% existed only for EGD with variceal therapy, insertion of a self-expanding prosthesis, dilatation/bougienage/exchange with PTC/PTCD in place, non-extensive ERCP, endoscopic ultrasound in the upper gastrointestinal tract, and colonoscopy with submucosal or full thickness resection, or foreign object removal. PCCL differed in all but one group. CONCLUSION: Gastroenterology endoscopy services provided as part of inpatient care but potentially performable on an outpatient basis are predominantly equally expensive for day cases as for patients with a length of stay greater than one day. The disease severity is lower. Calculated §21-KHEntgG cost data thus form a reliable basis for the calculation of appropriate reimbursement for hospital services to be provided as outpatient services under the AOP in the future.

Willingness to pay for a cure of low-risk melanoma patients in Germany.

Malignant melanoma is potentially life-threatening but in most cases curable if detected early. Willingness to pay (WTP) is a preference-based construct that reflects burden of disease by assessment of the monetary value for a hypothetical cure from disease. Since WTP (directly as total amount of money) has not been assessed so far in patients with low risk melanoma, it was interesting to gain insights in this patient population and then, in a second step, compare it directly with the WTP of their treating dermato-oncologists. WTP was assessed in 125 patients with low-risk melanoma and additionally in 105 treating physicians, asking for the one-time and continuous payments they would be willing to make for a sustainable cure, both as absolute sums and as percentages of monthly income. The median WTP based on one-time payment was €10,000 for patients and €100,000 for physicians; relative numbers were 100% versus 300% of monthly income. For continuous monthly payments, WTP was €500 for patients and €1000 for physicians, relative numbers 25% and 50% of income, respectively. Even after controlling for income differences, there was a significantly higher WTP in physicians for all four questions. Compared to patients with chronic skin diseases such as vitiligo, rosacea, atopic eczema and psoriasis, patients with low-risk melanoma showed a significantly higher WTP. Our data suggest that there is a relevant burden of disease even in patients with low-risk tumors. Higher WTP of physicians underlines the prevalence of differences in disease perception.