Multidomain Object Detection Framework Using Feature Domain Knowledge Distillation.

Object detection techniques have been widely studied, utilized in various works, and have exhibited robust performance on images with sufficient luminance. However, these approaches typically struggle to extract valuable features from low-luminance images, which often exhibit blurriness and dim appearence, leading to detection failures. To overcome this issue, we introduce an innovative unsupervised feature domain knowledge distillation (KD) framework. The proposed framework enhances the generalization capability of neural networks across both low-and high-luminance domains without incurring additional computational costs during testing. This improvement is made possible through the integration of generative adversarial networks and our proposed unsupervised KD process. Furthermore, we introduce a region-based multiscale discriminator designed to discern feature domain discrepancies at the object level rather than from the global context. This bolsters the joint learning process of object detection and feature domain distillation tasks. Both qualitative and quantitative assessments shown that the proposed method, empowered by the region-based multiscale discriminator and the unsupervised feature domain distillation process, can effectively extract beneficial features from low-luminance images, outperforming other state-of-the-art approaches in both low-and sufficient-luminance domains.

A freshwater algae classification system based on machine learning with StyleGAN2-ADA augmentation for limited and imbalanced datasets.

Automated algae classification using machine learning is a more efficient and effective solution compared to manual classification, which can be tedious and time-consuming. However, the practical application of such a classification approach is restricted by the scarcity of labeled freshwater algae datasets, especially for rarer algae. To overcome these challenges, this study proposes to generate artificial algal images with StyleGAN2-ADA and use both the generated and real images to train machine-learning-driven algae classification models. This approach significantly enhances the performance of classification models, particularly in their ability to identify rare algae. Overall, the proposed approach improves the F1-score of lightweight MobileNetV3 classification models covering all 20 freshwater algae covered in this research from 88.4% to 96.2%, while for the models that cover only the rarer algae, the experiments show an improvement from 80% to 96.5% in terms of F1-score. The results show that the approach enables the trained algae classification systems to effectively cover algae with limited image data.

The effect of assessor professional background on interview evaluation during residency selection: A mixed-methods study.

BACKGROUND: Residency interviewer scores are greatly variable and seems to be influenced by personal characteristics of assessors, although factors contributing to variability remain unclear. The study sought to determine how different professional backgrounds influence assessors' scores. METHODS: Fifty-five general surgery applicants rotated through an interview station assessing teamwork. They were scored by surgeons, human-resource managers, pilots, athletes. Pearson's correlation and a repeated-measures ANOVA were used to determine correlations between professions. Structured interviews were used to probe for scoring rationale. RESULTS: Interview scores differed significantly between professions (F (3, 159) = 11.12, p < 0.001. Qualitative analysis revealed that due to the challenge of distinguishing between similarly performing candidates, assessors rely on global impressions informed by personal values. CONCLUSION: Assessor variability is ubiquitous, in part due to the subjective nature of interviews and is associated with personal values. When selecting assessors, programs should choose diverse assessors to assess to ensure a reliable selection process.

Self-Assessment in General Surgery Applicants: An Insight Into Interview Performance.

INTRODUCTION: Selecting medical students for residency is a competitive process, with a narrow range of scores separating middle-ranked applicants. Self-assessment is a fundamental skill for any competent physician with a demonstrated correlation to diagnostic ability, examination scores, and technical skills, but has yet to be investigated in residency selection. The objective of this study was to investigate the relationship between self-assessment and interview performance as a potential adjunct to discriminate between applicants. METHODS: At the University of Ottawa in 2020, 55 applicants completed a 9-station interview circuit assessing different characteristics or skills important for a career in general surgery, followed by a self-assessment questionnaire evaluating their perceived performance at each station. Pearson's correlation was used to determine the relationship between self-assessment scores (SASs) and interviewer scores (ISs). RESULTS: There was a negative correlation between SASs and ISs for all interview stations. High performers underestimated their interview performance, and low performers overestimated their performance. Seven of the nine stations reached statistical significance (r = 0.60-0.73, P < 0.001). There was significant variability in the SAS of middle-ranked applicants, with a range three times greater than the range of ISs and demonstrating distinct self-assessment skills in candidates with very similar scores. CONCLUSIONS: Although we strive to select applicants who will succeed in residency to become competent physicians, self-assessment skills may be a useful adjunct during the interview process to assist in discriminating between applicants with similar scores.

An Advanced Noise Reduction and Edge Enhancement Algorithm.

Complementary metal-oxide-semiconductor (CMOS) image sensors can cause noise in images collected or transmitted in unfavorable environments, especially low-illumination scenarios. Numerous approaches have been developed to solve the problem of image noise removal. However, producing natural and high-quality denoised images remains a crucial challenge. To meet this challenge, we introduce a novel approach for image denoising with the following three main contributions. First, we devise a deep image prior-based module that can produce a noise-reduced image as well as a contrast-enhanced denoised one from a noisy input image. Second, the produced images are passed through a proposed image fusion (IF) module based on Laplacian pyramid decomposition to combine them and prevent noise amplification and color shift. Finally, we introduce a progressive refinement (PR) module, which adopts the summed-area tables to take advantage of spatially correlated information for edge and image quality enhancement. Qualitative and quantitative evaluations demonstrate the efficiency, superiority, and robustness of our proposed method.

A chance for reform: The environmental impact of travel for general surgery residency interviews.

BACKGROUND: In light of the global climate emergency, it is worth reconsidering the current practice of medical students traveling to interview for residency positions. We sought to estimate carbon dioxide (CO2) emissions associated with travel for general surgery residency interviews in Canada, and the potential avoided emissions if interviews were restructured. METHODS: An eight-item survey was constructed to collect data on cities visited, travel modalities, and costs incurred. Applicants to the University of Ottawa General Surgery Program during the 2019/20 Canadian Resident Matching Service (CaRMS) cycle were invited to complete the survey. Potential reductions in CO2 emissions were modeled using a regionalized interview process with either one or two cities. RESULTS: Of a total of 56 applicants, 39 (70%) completed the survey. Applicants on average visited 10 cities with a mean total cost of $4,866 (95% CI=3,995-5,737) per applicant. Mean CO2 emissions were 1.82 (95% CI=1.50-2.14) tonnes per applicant, and the total CO2 emissions by applicants was estimated to be 101.9 (95% CI=84.0 - 119.8) tonnes. In models wherein interviews are regionalized to one or two cities, emissions would be 57.9 tonnes (43.2% reduction) and 84.2 tonnes (17.4% reduction), respectively. Overall, 74.4% of respondents were concerned about the environmental impact of travel and 46% would prefer to interview by videoconference. CONCLUSION: Travel for general surgery residency interviews in Canada is associated with a considerable environmental impact. These findings are likely generalizable to other residency programs. Given the global climate crisis, the CaRMS application process must consider alternative structures.

CONTEXTE: Compte tenu de la situation d'urgence climatique mondiale, il convient de reconsidérer l'usage actuel selon lequel les étudiants en médecine se déplacent pour se présenter aux entrevues en vue d'obtenir un poste de résidence. Nous avons tenté d'estimer les émissions de dioxyde de carbone (CO2) causées par les déplacements pour les entretiens de résidence en chirurgie générale au Canada, et les émissions potentielles évitées si les entretiens étaient organisés autrement. MÉTHODES: Un sondage comportant huit questions a été élaboré pour recueillir les données sur les villes visitées, les modalités de voyage et les coûts encourus. Les candidats au programme de chirurgie générale de l'Université d'Ottawa au cours du cycle 2019/20 du Service canadien de jumelage des résidents (CaRMS) ont été invités à y répondre. Les réductions potentielles des émissions de CO2 ont été modélisées à l'aide d'un processus d'entrevue régionalisé avec une ou deux villes. RÉSULTATS: Sur un total de 56 candidats, 39 (70 %) ont répondu au sondage. Les candidats ont visité en moyenne 10 villes, pour un coût total moyen de 4 866 dollars (IC 95 % = 3 995-5 737) par candidat. Les émissions moyennes de CO2 étaient de 1,82 (IC 95 % = 1,50-2,14) tonne par candidat, et le total des émissions de CO2 pour l'ensemble des candidats était estimé à 101,9 (IC 95 % = 84,0 - 119,8) tonnes. D'après les modèles où les entrevues sont régionalisées avec une ou deux villes, les émissions seraient respectivement de 57,9 tonnes (43,2 % de réduction) et 84,2 tonnes (17,4 % de réduction). Dans l'ensemble, 74,4 % des personnes interrogées se disent préoccupées par l'impact environnemental des déplacements et 46 % préféreraient que l'entretien se fasse par vidéoconférence. CONCLUSION: Les déplacements pour les entrevues de résidence en chirurgie générale au Canada ont un impact environnemental considérable. Ces conclusions sont probablement généralisables à d'autres programmes de résidence. Compte tenu de la crise climatique mondiale, il conviendrait d'envisager d'autres modalités d'organisation des entrevues pour le processus de candidatures du CaRMS.

The induced membrane technique for the management of long bone defects.

AIMS: The purpose of this study was to: review the efficacy of the induced membrane technique (IMT), also known as the Masquelet technique; and investigate the relationship between patient factors and technique variations on the outcomes of the IMT. METHODS: A systematic search was performed in CINAHL, The Cochrane Library, Embase, Ovid MEDLINE, and PubMed. We included articles from 1 January 1980 to 30 September 2019. Studies with a minimum sample size of five cases, where the IMT was performed primarily in adult patients (≥ 18 years old), in a long bone were included. Multivariate regression models were performed on patient-level data to determine variables associated with nonunion, postoperative infection, and the need for additional procedures. RESULTS: A total of 48 studies were included, with 1,386 cases treated with the IMT. Patients had a mean age of 40.7 years (4 to 88), and the mean defect size was 5.9 cm (0.5 to 26). In total, 82.3% of cases achieved union after the index second stage procedure. The mean time to union was 6.6 months (1.4 to 58.7) after the second stage. Our multivariate analysis of 450 individual patients showed that the odds of developing a nonunion were significantly increased in those with preoperative infection. Patients with tibial defects, and those with larger defects, were at significantly higher odds of developing a postoperative infection. Our analysis also demonstrated a trend towards the inclusion of antibiotics in the cement spacer having a protective effect against the need for additional procedures. CONCLUSION: The IMT is an effective management strategy for complex segmental bone defects. Standardized reporting of individual patient data or larger prospective trials is required to determine the optimal implementation of this technique. This is the most comprehensive review of the IMT, and the first to compile individual patient data and use regression models to determine predictors of outcomes. Cite this article: Bone Joint J 2020;102-B(12):1723-1734.

Longer Trials of Non-operative Management for Adhesive Small Bowel Obstruction Are Associated with Increased Complications.

BACKGROUND: Current guidelines for the management of adhesive small bowel obstruction suggest a limited trial of non-operative management, often of 3-5 days. A longer delay to operation may worsen post-operative outcomes in patients who ultimately require operation. Our objective was to evaluate the impact of time to operation on post-operative outcomes in patients who undergo operation following a trial of non-operative management for adhesive small bowel obstruction. METHODS: We used health administrative data to identify patients with adhesive small bowel obstruction who underwent operative management following a trial of non-operative management from 2005 to 2014 in the province of Ontario, Canada. We used multivariable logistic regression to examine the relationship between the time from admission to operation with rates of 30-day mortality, serious complication, and bowel resection. RESULTS: Three thousand five hundred sixty-three patients underwent operation after a trial of non-operative management for adhesive small bowel obstruction. Older patients, patients with a high comorbidity burden, and patients with a lower socioeconomic status were more likely to experience a longer pre-operative period. After adjusting for covariates, each additional day from admission to operation increased odds of serious complication (OR = 1.07, 95% CI = 1.03-1.11) and bowel resection (OR = 1.06, 95% CI = 1.03-1.98). Longer times to operation were not associated with greater adjusted odds of 30-day mortality. CONCLUSION: Each additional day from admission to operation is associated with greater odds of adverse outcomes. Clinical practice guidelines should emphasize strategies that identify patients who will ultimately require operation.

An integrative longitudinal resilience curriculum.

BACKGROUND: It is well documented that student well-being is challenged at medical school and that levels of distress increase as students navigate their training. The Doctor of Medicine (MD) programme at the University of Toronto developed a 4-year resilience curriculum (RC) to encourage students to reach out for help and equip them with resilience-building strategies to manage adversities in a demanding academic and clinical programme. … resilience curriculum (RC) to encourage students to reach out for help and equip them with resilience-building strategies METHODS: Satisfaction surveys, consisting of statements rated by a five-point Likert scale and short-answer questions, were distributed to 518 students; in total, data from four workshops were collected. Two focus groups comprising 12 participants in total were facilitated (n = 6 per group). A thematic content analysis was conducted for the focus group data; open coding was used for transcriptions via an iterative process and inductive analysis. FINDINGS: Preliminary quantitative and qualitative data suggest that students valued the curriculum. The main themes generated from the thematic content analysis were the value of the RC, the delivery of the RC, and developing a resilient community. DISCUSSION: More research must be conducted to assess whether the RC has affected student well-being and resilience. The sustainability of the curriculum depends on the faculty members that support it; faculty development within the areas of wellness and resilience is imperative. INNOVATION AND IMPLICATIONS: The RC, embedded in the core curriculum and integrated within a medical community, is gaining momentum and is valued by students. Further research will assist in the creation of an innovative tool to assess the impact of the RC on medical students.

Learning Stylometric Representations for Authorship Analysis.

Authorship analysis (AA) is the study of unveiling the hidden properties of authors from textual data. It extracts an author's identity and sociolinguistic characteristics based on the reflected writing styles in the text. The process is essential for various areas, such as cybercrime investigation, psycholinguistics, political socialization, etc. However, most of the previous techniques critically depend on the manual feature engineering process. Consequently, the choice of feature set has been shown to be scenario- or dataset-dependent. In this paper, to mimic the human sentence composition process using a neural network approach, we propose to incorporate different categories of linguistic features into distributed representation of words in order to learn simultaneously the writing style representations based on unlabeled texts for AA. In particular, the proposed models allow topical, lexical, syntactical, and character-level feature vectors of each document to be extracted as stylometrics. We evaluate the performance of our approach on the problems of authorship characterization, authorship identification and authorship verification with the Twitter, blog, review, novel, and essay datasets. The experiments suggest that our proposed text representation outperforms the static stylometrics, dynamic n -grams, latent Dirichlet allocation, latent semantic analysis, distributed memory model of paragraph vectors, distributed bag of words version of paragraph vector, word2vec representations, and other baselines.

Quantifying the costs and benefits of privacy-preserving health data publishing.

Cost-benefit analysis is a prerequisite for making good business decisions. In the business environment, companies intend to make profit from maximizing information utility of published data while having an obligation to protect individual privacy. In this paper, we quantify the trade-off between privacy and data utility in health data publishing in terms of monetary value. We propose an analytical cost model that can help health information custodians (HICs) make better decisions about sharing person-specific health data with other parties. We examine relevant cost factors associated with the value of anonymized data and the possible damage cost due to potential privacy breaches. Our model guides an HIC to find the optimal value of publishing health data and could be utilized for both perturbative and non-perturbative anonymization techniques. We show that our approach can identify the optimal value for different privacy models, including K-anonymity, LKC-privacy, and ∊-differential privacy, under various anonymization algorithms and privacy parameters through extensive experiments on real-life data.

Privacy-preserving heterogeneous health data sharing.

OBJECTIVE: Privacy-preserving data publishing addresses the problem of disclosing sensitive data when mining for useful information. Among existing privacy models, ε-differential privacy provides one of the strongest privacy guarantees and makes no assumptions about an adversary's background knowledge. All existing solutions that ensure ε-differential privacy handle the problem of disclosing relational and set-valued data in a privacy-preserving manner separately. In this paper, we propose an algorithm that considers both relational and set-valued data in differentially private disclosure of healthcare data. METHODS: The proposed approach makes a simple yet fundamental switch in differentially private algorithm design: instead of listing all possible records (ie, a contingency table) for noise addition, records are generalized before noise addition. The algorithm first generalizes the raw data in a probabilistic way, and then adds noise to guarantee ε-differential privacy. RESULTS: We showed that the disclosed data could be used effectively to build a decision tree induction classifier. Experimental results demonstrated that the proposed algorithm is scalable and performs better than existing solutions for classification analysis. LIMITATION: The resulting utility may degrade when the output domain size is very large, making it potentially inappropriate to generate synthetic data for large health databases. CONCLUSIONS: Unlike existing techniques, the proposed algorithm allows the disclosure of health data containing both relational and set-valued data in a differentially private manner, and can retain essential information for discriminative analysis.