Additional Margin Research Articles

Intraoperative Fluorescence Guidance for Breast Cancer Lumpectomy Surgery

BackgroundAlthough lumpectomy and mastectomy provide equivalent survival for patients with breast cancer, local recurrence after lumpectomy increases breast cancer mortality. Positive lumpectomy margins, which imply incomplete tumor removal, are the strongest predictor of local recurrence and are identified days after surgery, necessitating a second surgery.MethodsIn this prospective trial, we assessed margin status with or without pegulicianine fluorescence-guided surgery (pFGS) for stages 0 to 3 breast cancers. To prevent surgeons from performing smaller than standard lumpectomies in anticipation of pFGS assistance, patients were randomly assigned 10:1 to pFGS or control groups, thus randomization was not designed to provide a control group for evaluating device performance. In patients undergoing pFGS, additional pFGS-guided cavity margins were excised at sites of pegulicianine signal. We evaluated three coprimary end points: the percentage of patients for whom pFGS-guided margins contained cancer, sensitivity, and specificity.ResultsOverall, 406 patients received 1.0 mg/kg intravenous pegulicianine followed by lumpectomy. Among 392 patients randomly assigned, 316 had invasive cancers, and 76 had in situ cancers. In 27 of 357 patients undergoing pFGS, pFGS-guided margins removed tumor left behind after standard lumpectomy, 22 from cavity orientations deemed negative on standard margin evaluation. Second surgeries were avoided by pFGS in 9 of 62 patients with positive margins. On per-margin analysis, pFGS specificity was 85.2%, and sensitivity was 49.3%. Pegulicianine administration was stopped for adverse events in six patients. Two patients had grade 3 serious adverse events related to pegulicianine.ConclusionsThe use of pFGS in breast cancer surgery met prespecified thresholds for removal of residual tumor and specificity but did not meet the prespecified threshold for sensitivity. (Funded by Lumicell, Inc. and the National Institutes of Health; Clinicaltrials.gov number, NCT03686215.)

Safety of Reinforced Concrete Columns: Effect of Initial Imperfections and Material Deterioration under Emergency Actions

The effect of (1) initial imperfections and (2) material degradation of reinforced concrete columns on their safety in emergency situations was investigated. The research was limited to low- and medium-flexibility columns. Numerical modeling and proven regulatory methods of analysis were applied to determine the ultimate bearing capacity, taking into account supplementary dynamic loading by a longitudinal force and a bending moment in case of emergency. The numerical model, describing the column structure, has 3D elements simulating concrete, and rebars simulating reinforcement frames (cages). Imperfections are simulated by (1) the physical loss of elements, (2) unzip of nodal elements, and (3) unzip and further zip using nonlinear elements simulating gaps and cohesion between concrete and reinforcement. Implicit dynamics and an incremental method were employed to make computations. Within the framework of this computational scheme, a nonlinear problem was solved using the Newton–Raphson method with nodal forces convergence. The effect of imperfections, such as geometrical deviations and deterioration of mechanical characteristics, on the bearing capacity of compressed bending elements was identified under emergency actions. Risks of mechanical safety loss were analyzed to find that columns in the frame structures of highly hazardous, technically complex, and unique buildings and structures, subjected to supplementary loading, need an additional safety margin in the range of 3–21%. Rectangular cross-sections of columns are the most effective in terms of the safety criterion.

CAMRI Loss: Improving the Recall of a Specific Class without Sacrificing Accuracy

In real-world applications of multi-class classification models, misclassification in an important class (e.g., stop sign) can be significantly more harmful than in other classes (e.g., speed limit). In this paper, we propose a loss function that can improve the recall of an important class while maintaining the same level of accuracy as the case using cross-entropy loss. For our purpose, we need to make the separation of the important class better than the other classes. However, existing methods that give a class-sensitive penalty for cross-entropy loss do not improve the separation. On the other hand, the method that gives a margin to the angle between the feature vectors and the weight vectors of the last fully connected layer corresponding to each feature can improve the separation. Therefore, we propose a loss function that can improve the separation of the important class by setting the margin only for the important class, called Class-sensitive Additive Angular Margin Loss (CAMRI Loss). CAMRI loss is expected to reduce the variance of angles between features and weights of the important class relative to other classes due to the margin around the important class in the feature space by adding a penalty to the angle. In addition, concentrating the penalty only on the important classes hardly sacrifices the separation of the other classes. Experiments on CIFAR-10, GTSRB, and AwA2 showed that the proposed method could improve up to 9% recall improvement on cross-entropy loss without sacrificing accuracy.

Capsule Network Extension Based on Metric Learning

A capsule network (CapsNet) is a deep learning model for image classification that provides robustness to changes in the poses of objects in the images. A capsule is a vector whose direction represents the presence, position, size, and pose of an object. However, with CapsNet, the distribution of capsules is concentrated in a class, and the number of capsules increases with the number of classes. In addition, learning is computationally expensive for a CapsNet. We proposed a method to increase the diversity of capsule directions and decrease the computational cost of CapsNet training by allowing a single capsule to represent multiple object classes. To determine the distance between classes, we used an additive angular margin loss called ArcFace. To validate the proposed method, the distribution of the capsules was determined using principal component analysis to validate the proposed method. In addition, using the MNIST, fashion-MNIST, EMNIST, SVHN, and CIFAR-10 datasets, as well as the corresponding affine-transformed datasets, we determined the accuracy and training time of the proposed method and original CapsNet. The accuracy of the proposed method improved by 8.91% on the CIFAR-10 dataset, and the training time reduced by more than 19% for each dataset compared with those of the original CapsNets.

Liquidity Impact of Novel Market Surveillance Measures—An Evidence from India

Stock price manipulation can be defined as trading behaviour that distorts stock prices with the intent of personal gains. Securities market regulators around the world undertake market surveillance to prevent market manipulation protect investors and ensure market integrity. The Securities Exchange Board of India along with stock exchanges has implemented a novel additional surveillance measure (ASM) which tracks stocks on a daily basis for abnormal price and volume activity accompanied by the participation of market-dominating investors in trading activity. Stocks that breach predefined limits of the above parameters are categorized as ASM category and subject to overall additional margins and selective additional margins for the dominating investor. There are mainly two types of ASM, Long term additional surveillance measures (LTASM) and short-term additional surveillance measures (STASM). STASM consider a shorter duration time (5/15 days) for share price observation LTASM considers a longer duration (1 month/1 year) for screening of the stock for surveillance measures as per ASM criteria. This study is based on STASM. The mechanism, on the one hand, is expected to deter manipulators and on the other, warn investors about suspicious activity in the stock, which is not supported by fundamentals. This article analyses the impact of surveillance actions on a price return and liquidity. The article also looks at the degree of speculative activity surrounding the event. This study used regression analysis for impact assessment. The study results show that the prices of the stock stabilize after inclusion and there is an overall fall in liquidity post-inclusion. However, the liquidity impact is different for stocks with positive abnormal returns as compared to those with negative abnormal returns. The result of the study gives insight into the effectiveness of STASM which will be useful for regulators and investors to prevent market manipulations.

Mohs Surgery for Periocular Basal Cell Carcinoma Without a Mohs Surgeon: The First Series in Hong Kong.

Prospective non-comparative interventional case series. 20 consecutive Chinese patients (10 male, age 78.5+10.4 years, range 55-91 years) with primary periocular basal cell carcinoma (pBCC) referred to the university oculoplastic unit between October 2007 and August 2013. MMS were conducted according to a streamlined standard operating procedure emphasizing surgeon-driven mapping, specimen orientation, and on-site clinico-histological correlation with the dermatopathologist at the frozen-section laboratory. Clinical and histological characteristics of tumors, layers of Mohs procedures, complications, and biopsy-confirmed recurrence at the same location. Results All 20 patients received MMS as planned. Sixteen pBCCs (80%) were diffusely pigmented, and three (15%) were focally pigmented. Sixteen were also nodular. The average tumor diameter was 7+3 (3-15) mm.Seven (35%) were within 2 mm of the punctum. Histologically, 11 (55%) were nodules, and four (20%) were superficial. An average of 1.8+0.8 Mohs levels were performed. Apart from the initial two patients, who required four and three levels, respectively, seven (35%) patients were cleared after the first level of MMS using a 1mm clinical margin. The remaining 11 patients required two levels with an additional 1-2mm margin, but only focally as guided histologically.Defects in 16 patients (80%)were reconstructed by local flaps, two by direct closure, and two with pentagon closure. Among the seven patients with pericanalicular BCC, three patients had their remaining canaliculi successfully intubated, while two developed stenotic upper and two lower punctae postoperatively. One patient had prolonged wound healing. Three patients had lid margin notching, two had medial ectropion, one had medial canthal rounding, and two had lateral canthal dystopia.No recurrence was detected at a mean follow-up of 80+23 months (43 to 113 months) in all patients. Conclusions MMS was successfully introduced in Hong Kong without a Mohs surgeon. Providing complete microscopic margin control and preserving tissues, it was proven to be a valuable treatment option for pBCC.Our multidisciplinary protocol demonstrated that these merits are possible and warrant validation in other resource-limited healthcare settings.

Abstract P4-07-16: B-IMMUNE final analysis: a phase Ib/II study of durvalumab combined with dose-dense EC in a neoadjuvant setting for patients with locally advanced luminal B HER2(-) or triple negative breast cancers

Abstract Background: Neoadjuvant association of immune checkpoints inhibitors (ICI) and dose dense chemotherapy is promising for triple negative breast cancers (TNBC). However, response rates vary from one study to another. Timing, best chemotherapy partner and efficacy in less immunogenic breast cancer (BC), like luminal B tumors, should be further investigated. This study evaluates for TNBC and luminal B HER2(-) BC the neoadjuvant treatment with paclitaxel followed by a short combination of an anti-PD-L1 antibody with anthracyclines. Method B-IMMUNE (NCT03356860), a multicentric phase Ib/II prospective trial, included patients with stage I to III luminal B HER2(-) or TNBC treated with paclitaxel 80mg/m2 weekly from week 1 to 12 followed by 4 cycles of epirubicine 90mg/m2 and cyclophosphamide 600 mg/m2 (EC) Q2W in a neoadjuvant setting. Phase Ib evaluated a single infusion of durvalumab (anti-PD-L1) combined with the 3rd cycle of EC. Phase II evaluated infusions of durvalumab with the 1st and 3rd EC cycles. Surgery was planned 3 weeks after the last EC cycle. Primary objectives were safety and pathological complete response (pCR) rate compared to a historical control. Secondary endpoint was the overall response rate (ORR) based on breast MRI. Eleven patients were enrolled in a control arm without durvalumab, exclusively for translational research purposes. Based on a 2-stage Simon design with an α = 0.1 and β = 0.1, 22 TNBC patients were needed in the phase II to test a null hypothesis of 30% pCR rate against a one-side alternative of 60%, and 24 luminal B BC patients to test a null hypothesis of 15% pCR rate against a one-side alternative of 40% (including an additional accrual margin of 10% for eventual dropouts). At least 9 pCRs had to be observed among the first 20 evaluable TNBC patients and 6 among the first 22 evaluable luminal B patients to rule out the null hypothesis. Results This analysis concerns the 50 patients treated with the experimental treatment, 3 from the phase Ib and 47 from the phase II part. Median age was 51 y-old (31 to 72y), tumor subtypes were 24 TNBC, 25 Luminal B and one sarcoma excluded from the efficacy analysis. Seven (14%) patients had a stage I tumor, 17 (34%) a stage IIA, 13 (26%) a stage IIB, 8 (16%) a stage IIIA, 4 (8%) a stage IIIB and 1 (2%) a stage IIIC. Concerning safety, 232 AEs were reported on 39/50 patients and 34 (14,6%) were graded ≥ 3. The 5 most frequent all-grade AEs were fatigue (8,2%), diarrhea (5,6%), neutropenia (5,2%), anemia and nausea (4,3%). Most frequent grade 3 AEs were anemia and neutropenia (14,7%). Among 4 immune-related adverse events, all were thyroid disorders. One patient died 10 months after the end of treatment due to progressive disease in the liver. Forty-six of the 47 phase II patients were evaluable for efficacy. pCR was reported in 12/22 TNBC patients (55%) and 8/24 luminal B HER2(-) patients (33%). Subgroup analyses based on PD-L1 expression and TILs score are planned. Conclusions The B-IMMUNE study met its primary objective showing a significant improvement in pCR versus the historical control in both TNBC and in Luminal B HER2(-) BC cohorts with the addition of only 2 doses of durvalumab to the anthracyclines. The safety profile is comparable to those previously described with reported immune related adverse events limited to thyroid endocrine disorders. Citation Format: Alix Devaux, Gabriela Beniuga, Claire Quaghebeur, Stéphanie Henry, Mieke Van Bockstal, Christine Galant, Paul Delrée, Jean-Luc Canon, Brigitte Honhon, Dominique Korman, Vincent Verschaeve, Christophe Lonchay, Sarah Lefevre, Lionel D’Hondt, Martine Berlière, Sophie Delmarcelle, Jean-Michel Mine, Timour Willems, Gebhard Müller, Nathalie Myant, Isabelle Bar, Sandy Haussy, Pierre G. Coulie, François P. Duhoux, Javier Carrasco. B-IMMUNE final analysis: a phase Ib/II study of durvalumab combined with dose-dense EC in a neoadjuvant setting for patients with locally advanced luminal B HER2(-) or triple negative breast cancers. [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P4-07-16.

Abstract P1-09-05: Effect of Wire vs Magnetic Seed Localization on Lumpectomy Cavity Size

Abstract Purpose/Objectives To assess whether an association exists between surgical localization technique and size of lumpectomy cavity on radiation (RT) planning CT scan. Adjuvant RT with boost to lumpectomy cavity has been shown to improve local control compared to adjuvant whole breast radiation alone, but larger cavity sizes can lead to worse cosmetic outcomes following boost administration, which could lead providers to omit boost. Therefore, decreasing cavity size could increase guideline-concordant boost administration and minimize adverse cosmetic outcomes. Materials/Methods A retrospective review was conducted of all patients undergoing breast conserving surgery with either wire- or magnetic seed -guided lumpectomy followed by adjuvant RT at a single institution from 2018 to 2021. Data were collected from pre-surgical work-up, surgical pathology, and radiation planning. Women undergoing bracketed wire localization and patients treated by a surgeon who only performed wire localized procedures were excluded. The primary outcome was lumpectomy cavity size as measured on planning CT. We first conducted an overlapping weights propensity score analysis to account for imbalance between groups in age, BMI, breast size as measured on planning CT scan, pre-operative imaging tumor volume, neoadjuvant therapies, lumpectomy histology (DCIS alone vs DCIS + invasive vs invasive alone vs invasive with lobular features vs no residual), and multifocality. Multivariable analysis (MVA) of CT cavity volume included the above weighted variables as well as surgeon and radiation oncologist. Secondary analyses included MVA of total pathologic volume, bivariable analysis of boost delivery, bivariable analysis of electron vs photon boost, stratification by surgeon, and fixed effect model for year of surgery. Results Of 617 women who received lumpectomy during the study period, 387 were included in final analysis. Patients who underwent seed localization were less likely to have multifocal disease, less likely to have calcifications on mammogram, more likely to have ultrasound measurements for pre-op imaging, and had smaller tumor size on pre-op imaging. Four surgeons performed all cases, with rates of seed use per surgeon ranging from 27.7% to 70.7% but generally increasing throughout the study period. There was no difference between wire and seed localization in the need for additional margins based on intra-operative margin analysis (58.4 vs 62.7%, p = 0.5). There was no difference in positive margins (6.4 vs 5.4%, p = 0.81) or second surgeries (9.4 vs 8.1%, p = 0.79). Rates of close margins were the same for DCIS (23.4 vs 17.3%, p = 0.35) and invasive carcinoma (7.6 vs 6.8%, p = 0.97) between techniques. Initial uncorrected bivariable analysis shows wire localization has a non-significant trend toward increased CT cavity volume (4.56cc, p = 0.15) and a significant association with total pathology volume (21.7cc, p = 0.004). For the primary outcome, breast size, time from surgery to simulation, and surgeon were all significantly associated with CT cavity volume but there was no significant difference by localization technique (p=0.38). For pathology volume, there was a non-significant trend toward increased specimen volume with wire localization (p = 0.07), and significant associations with BMI, histology, and pre-op imaging volume. When stratified by surgeon, there was no surgeon for whom one localization technique led to significantly different CT or pathology volume over the other. There were no significant changes of the treatment effect over time (p = 0.79). There was no significant difference between wire and seed localization in indicated boost delivery (85% vs 79%, p = 0.14) or electron boost (42% vs 56%, p = 0.13). Conclusion There was no significant difference in CT cavity size between wire localization and magnetic seed localization, suggesting that the choice between these surgical techniques does not impede RT boost delivery. Citation Format: Michael Dykstra, Jessica Thompson, Jessica Aldous, Shannon Jiang, Tasha Hughes, James Hayman, Aleksandar Dragovic, Jennifer Shah, Alfred Chang, Corey W. Speers, Michael Sabel, Lesly Dossett, Matthew Schipper, Reshma Jagsi. Effect of Wire vs Magnetic Seed Localization on Lumpectomy Cavity Size [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P1-09-05.

UnifiedFace: A Uniform Margin Loss Function for Face Recognition

Face recognition has achieved great success due to the development of deep convolutional neural networks (DCNNs) and loss functions based on margin. However, complex DCNNs bring a large number of parameters as well as computational effort, which pose a significant challenge to resource-constrained embedded devices. Meanwhile, the popular margin-based loss functions all introduce only one type of margin and cannot further introduce a larger margin to achieve tighter classification boundary. In contrast to the common approach, we believe that additive and multiplicative margins should be used jointly to introduce larger margins from the margin perspective. Therefore, we propose a new margin-based loss function called UnifiedFace. First, we introduce an additive margin in the target angle activation function. Second, we add a multiplicative margin in the non-target angle. UnifiedFace introduces both additive and multiplicative margins, allowing for the introduction of large margins to achieve more compact intra-class variance and closer separated inter-class variance. In addition, we specifically design efficient face recognition models called GhostFaceNet for resource-constrained embedded devices. Experimental results demonstrate that UnifiedFace achieves state-of-the-art performance or performance competed with popular methods in training datasets of different sizes. UnifiedFace achieves optimal performance in models of varying complexity. Moreover, competitive results are achieved in the large-scale test set IJBB/C, especially the state-of-the-art performance achieved in TAR (FAR=1e−6). GhostFaceNet can significantly improve operational efficiency without significantly degrading recognition performance, making it ideal for embedded devices with limited resources.

The Geodynamic Implications of Passive Margin Subduction in Northwest Turkey

AbstractThe number of subduction zones that facilitated the northward translation of the Anatolide‐Tauride continental terrane derived from Gondwana to the southern margin of Eurasia at the longitude of western Turkey is debated. We hypothesized that if two north dipping subduction zones facilitated incipient collision in western Turkey, a late Cretaceous arc would have formed within the Neotethys and along the southern margin of Eurasia. To determine if an island arc formed within the Neotethys we investigated the sedimentary record of the Central Sakarya basin, which was deposited along the southern margin of Eurasia from 85 to 45 million years ago. Detrital zircon deposited within the lower levels of the Central Sakarya basin (the Değirmenözü Formation) are associated with south to north‐directed paleocurrents and exhibit a unimodal late Cretaceous age peak sourced from isotopically juvenile mantle melts. Zircon maximum depositional ages from the Değirmenözü Formation cluster between 95 and 90 Ma and are 5–10 Myr older than biostratigraphic depositional ages. Between 95 and 80 Ma, a 12‐unit shift from mantle to crustal derived εHf values occurs in the overlying Yenipazar Formation. We explain the absence of Paleozoic, Eurasian‐sourced detrital zircon, the rapid shift from mantle to crustal derived εHf values, and lag time in terms of passive margin subduction within an isolated intra‐oceanic subduction zone, whose island arc was reworked from south to north into the Central Sakarya basin during incipient collision. Thus, widely outcropping late Cretaceous plutonic rocks within Eurasia must have belonged to an additional convergent margin.

High Capacity Innovations Enabling Scalable Optical Transmission Networks

We review and introduce key optical innovations that enable scalable and efficient high capacity optical networks, such as advanced modulation formats, flexible grid, optical superchannel, 400GE and emerging 400G ZR/ZR+. We show results from recent industry leading technology trial of real-time 2 × 200G Probabilistic Constellation Shaping (PCS) DP-16QAM superchannel transporting 400GE over a live long haul flexible grid link, achieving long term error free performance with a high spectral efficiency of 5.33 bit/s/Hz and significantly enhanced performance over conventional 200G DP-16QAM. We also present results from successful demonstration of emerging 400G ZR/ZR+ technologies and IP over WDM applications, where different network scenarios and use cases are investigated including single span WDM link of 125 km and multi-span WDM link of 252 km as well as unamplified 40 km fiber link. 400G ZR/ZR+ IP over WDM wavelengths have shown robust performance with good optical margin for practical network applications as well as potential for longer optical reach. In comparison, 400G ZR+ has enhanced performance over 400G ZR with additional margin of ∼3 dB. We have also demonstrated streaming telemetry, a new network monitoring approach, for monitoring of 400G ZR/ZR+ IP over WDM wavelengths giving more performance insights. These industry leading trials and demonstration give network operators critical understandings of the performance, maturity and operational viability of emerging optical technologies which will allow them to scale their networks gracefully to meet continuing bandwidth growth.

Permanent section margin concordance after Mohs micrographic surgery with immunohistochemistry for invasive melanoma and melanoma in situ: A retrospective dual-center analysis

Mohs Micrographic Surgery (MMS) for melanoma practices vary among dermatologic surgeons. The implementation of immunohistochemical staining in MMS for melanoma mitigates challenges associated with slide interpretation, however the reliability of Melanoma antigen recognized by T cells 1 (MART-1), the preferred immunostain for melanoma, has yet to be compared to permanent section pathology. To assess concordance rates of MART-1 frozen sections and permanent section pathologic interpretation of melanoma treated with MMS. A dual center retrospective analysis was conducted to collect concordance and demographic data. Chi square tests were performed for group comparisons of categorical variables. Of the 379 permanent sections sent, 367 were concordant with frozen section pathology for an overall concordance rate of 96.8%. Cases were stratified into indeterminately concordant and indisputably concordant. 22 (6%) of cases were indeterminately concordant whereas 345 (94.0%) of cases were indisputably concordant. The concordance rate is derived from a comparison of adjacent tissue margins, an inevitable consequence of utilizing two techniques. To the author's knowledge, this study represents the largest investigation examining concordance rates of MART-1 frozen sections in Mohs for melanoma. High concordance disputes the ongoing need for additional permanent margins when using MART-1 in routine cases.

Radio-Protected Area Estimation Model Using Location-Dependent Gain for a Spectrum Sharing System in the VHF-Band

This paper proposes a radio-protected area (RPA) estimation model that achieves full protection and avoids overprotection by adding an appropriate margin to the extended-Hata (EH) model, for realizing efficient spectrum sharing in the V-High-band. This additional margin, called location gain, is a location-dependent value that accounts for knife-edge losses calculated from topographic information. This study experimentally obtains the location gain by conducting VHF-band-based propagation experiments in urban and suburban areas. Subsequently, we developed an estimation formula for the location gain using topographic data, which include correction terms based on the experimental data. As a result, RPA can be estimated using only the topographic data, excluding the need for propagation experiments. The proposed model resolved the problems of the EH model (incomplete protection) and free-space (FS) model (overprotection). In urban and suburban areas, the proposed model achieved full protection and over 99.7% reduction in the RPA size of the FS model when the threshold of the received power for determining the RPA was set to –80 dBm.

An Automatic Detection Framework for Electrical Anomalies in Electrified Rail Transit System

Automatic electrical anomaly detection for high-speed rail (HSR) is essential to ensure the safe and efficient operation of HSR. In this article, a novel anomaly detection method is proposed to rapidly identify typical anomalies and detect unknown anomalies in train-grid electrical coupling system (TGECS). The framework of the method consists of feature extraction, pre-classification and re-identification. First, the raw electrical signal is processed by Hilbert transform (HT) to obtain the instantaneous envelope. Second, the one-dimensional (1-D) convolutional neural network (CNN) is adopted to learn deeper features and perform pre-classification. To reduce the number of parameters, the global average pooling layer is used instead of the fully connected layers. To further improve the recognition accuracy, the additive angular margin named arcface is used to enhance the discriminative power of the model. Finally, the re-identification operation determines the final result by calculating the distances between the features. The accuracy of the proposed method for typical electrical anomaly recognition is beyond 99% and for unknown anomaly detection exceeds 86%. The experimental results demonstrate the effectiveness and the superior performance of the proposed method in terms of recognition accuracy, unknown anomalies detection and computational complexity, compared to the NN, 2-D CNN, LSTM and CNN-LSTM.

Ripple Excitation-Based Adaptive Sensorless Control of IPMSM in Full Speed Range

This paper proposes an adaptive sensorless finite-control-set model predictive control (FCS-MPC) method for the interior permanent magnet synchronous motor (IPMSM). The method is feasible in the full speed range from zero speed to the flux-weakening region above rated speed, without the method transition between the low and the high speeds. A ripple excitation-based position estimation method is proposed to extract the rotor-saliency-based position information from the inherent voltage ripples of the FCS-MPC. Therefore, the additional voltage margin consumption, the special sampling timing, or the interference in the fundamental control are all avoided. Furthermore, to alleviate the parameter dependency, an adaptive predictive model with the online estimation of the lumped voltage terms and the equivalent dynamic susceptances is proposed. In this way, the main machine parameters and the unmodeled disturbances in the d-q voltage equations of the IPMSM are all considered. The position deviation due to the cross-coupling effect is also analyzed, and a compensation method is proposed. Finally, the effectiveness of the proposed method is validated by the experimental results and the comparison with existing methods.

KappaFace: Adaptive Additive Angular Margin Loss for Deep Face Recognition

KappaFace: Adaptive Additive Angular Margin Loss for Deep Face Recognition

Natural resources: Cost of capital and discounting – Risk and uncertainty

Notable contention remains around what discount rate should be used in the determination of the net present value (NPV) of a mining operation's discounted cash flows (DCF). It is typically deemed simpler to apply a single “corporate discount rate” that satisfies the targeted project's return associated with the investment capital outlay and some unquantified element of risk. This targeted return will ideally incorporate the actual cost of capital, being the weighted average cost of capital (WACC), incorporating the capital asset pricing model (CAPM) for the equity component and the cost of debt (the inter-bank lending rate with an additional lending margin). In addition, it will also incorporate some gut-feel or indeterminate additional return factor, to account for risk, to appease the project owners and its associated shareholders. This additional project risk being captured in a discount rate is contentious, as it generally is not quantified. In addition, while there may be some argument to justify why a single, all-encompassing discount rate should be used, notably for a short-life operation, can a single discount rate applied over the life of an operating asset actually be justified? Can a single rate be equally justified for medium and long-term operations as it is for short-life operations? Importantly and accepting that the WACC and hence the discount rate vary over time, how does the risk and the uncertainty associated with a mining project get factored into the valuation, and should these factors remain static, or should they be considered separately and also be varied over on operation's life?

Contralateral esophageal sparing technique in definitive radiotherapy for non-small cell lung cancer: dosimetric parameters and normal tissue complication probability modeling.

The purpose of this study was to assess the benefit of the contralateral esophageal sparing technique (CEST) in definitive radiotherapy of non-small cell lung cancer (NSCLC). We retrospectively reviewed radiation plans for 13 patients who underwent definitive chemoradiation for locally advanced NSCLC. Alternative plans were prepared with the use of CEST, with an additional margin of 5 mm from planning treatment volume (PTV). Normal tissue complication probability (NTCP) analyses for the esophagus and tumor control probability (TCP) for the PTV were performed for original and CEST plans using the equivalent uniform dose (EUD)-based mathematical model. In all cases, the CEST plan allowed for the reduction of esophageal dose, with a mean of 3.8 Gy (range, 0.7 to 8.7 Gy). The mean reductions of V40 and V60 to the esophagus were 6.4 Gy (range, 2.1 to 17.2 Gy) and 1.9 Gy (range, 3.4 to 10.0 Gy), respectively. There was no substantial decrease in the maximal dose to the esophagus. Reduction of NTCP was achieved for all patients (range, 5-73%), and TCP was not affected (-1.8 to +6.7%). The application of CEST in definitive radiotherapy of locally advanced NSCLC allows for reducing selected dosimetric parameters to the esophagus without compromising TCP.

Effect of three-dimensional intraoperative imaging on surgical outcomes with breast conservation therapy

BackgroundBreast conservation therapy (BCT) is frequently performed for breast cancer and associated with a significant risk for positive margins. Intraoperative three-dimensional (3-D) tomosynthesis potentially could limit the risk of positive margins. MethodsRetrospective review of an institutional breast cancer registry. Evaluated BCT cases for a two year time period prior to and after the introduction of intraoperative 3-D tomosynthesis. Primary outcome was the effect of 3-D tomosynthesis on margin positivity rates. Secondary measures were the impact of 3-D tomosynthesis on additional margin procurements at the index surgery and operative time. ResultsA total of 228 cases were evaluated with 106 cases utilizing 3-D tomosynthesis and 122 cases with standard imaging. No significant difference in margin positivity rates between the cohorts at 23.9% versus 15.8% for 3-D tomosynthesis and standard imaging respectively (OR 1.53, CI 0.772–3.032, P = 0.221). 3-D tomosynthesis was associated with increased margin procurement rates (OR 2.34, 95%CI 1.303–4.190, P = 0.004) and longer operative times (P < 0.001). ConclusionIntraoperative 3-D tomosynthesis was not found to limit margin positivity rates or improve the performance of the procedure.

An open-set fault diagnosis framework for MMCs based on optimized temporal convolutional network

Reliability of the modular multilevel converters (MMCs) provides a vital guarantee for the uninterrupted operation of a system. Insulated Gate Bipolar Transistors (IGBTs) open circuit fault diagnosis is a common challenge in MMCs applications. In this paper, a novel open-set fault diagnosis framework called Multiscale-AAM-OTCN is proposed to solve both the known and unknown fault diagnosis problems of MMCs by outputting current signals. First, batch normalization and layer normalization are introduced into the original Temporal Convolutional Network (TCN) model to accelerate convergence and promote the generalization ability of the model for different tasks. Second, to strengthen the feature extraction ability of the model, the multiscale coordinate residual attention (MCRA) mechanism is designed for the one-dimensional (1D) current signal to improve the performance and stability of the method. Compared with recently developed attention mechanisms such as convolutional block attention module (CBAM), efficient channel attention (ECA), simple, parameter-free attention module (SimAM) and coordinate attention (CA), the proposed MCRA exhibits better performance in MMCs fault diagnosis tasks. Finally, the additive angular margin (AAM) loss and local outlier factor (LOF) algorithm are integrated into the Multiscale-OTCN framework to identify the density difference between known and unknown fault clusters by controlling the intra-class similarity and inter-class variance of the samples. The experimental results demonstrate the feasibility of the proposed fault diagnosis framework for known and unknown fault diagnoses.