Anonymized Dataset Research Articles

Use of the Multidisciplinary Specialised Case Conference at the Freiburg Orbital Centre

The treatment of tumors increasingly takes place in specialised interdisciplinary centres. Therapeutic decisions are usually made at case conferences. Ophthalmologists, oromaximillofacial surgeons, ENT physicians, neurosurgeons, as well as pediatricians, radiotherapists and radiologists are all involved in the treatment of orbital diseases. The aim of this article is to present the concept of a multidisciplinary case conference for orbital diseases and to analyse case numbers, indications, and the influence on the patient's therapy. We analysed an anonymized data set of patients who presented in the case conference of the University Hospital Freiburg from 2008 to 2018 with regard to clinical diagnosis, histological diagnoses, number of surgical interventions, and number of interdisciplinary therapy decisions. From 2008 to 2018, 545 patients were presented in a weekly conference. Of these, 453 were available for anonymous evaluation. The median age was 56 years (quartiles 41; 69). The most frequent indication was an orbital tumour of unclear malignancy (n = 52; 11%). Further indications included Grave's orbitopathy (n = 39; 9%), orbital pseudotumour (n = 36; 8%), cranial nerve palsy (n = 22; 5%), and orbital lymphoma (n = 22; 5%). The most frequent histological diagnoses were B-cell lymphoma (n = 10; 2%), venous malformation (cavernoma, n = 8; 2%), marginal zone lymphoma (n = 8; 2%), and squamous cell carcinoma (n = 6; 1%). An interdisciplinary therapeutic approach was defined for 174 patients. A high demand for the interdisciplinary case conference was demonstrated. The high rate of primary or secondary interdisciplinary decisions indicates the value of such a conference. Hence, the patient is spared multiple examinations in the individual specialist areas and quick and effective therapy decisions can be achieved.

Micro-journal mining to understand mood triggers

In computational linguistics, binary sentiment analysis methods have been proposed to predict whether a document expresses a positive or a negative opinion. In this paper, we study a unique research problem—identifying environmental stimuli that contribute to different moods (mood triggers). Our analysis is enabled by an anonymous micro-journalling dataset, containing over 700,000 short journals from over 67,000 writers and their self-reported moods at the time of writing. We first build a multinomial logistic regression model to predict the mood (e.g., happy, sad, tired, productive) associated with a micro-journal. We then examine the model to identify predictive words and word trigrams associated with various moods. Our study offers new data-driven insights into public well-being.

Autoregressive Affective Language Forecasting: A Self-Supervised Task.

Human natural language is mentioned at a specific point in time while human emotions change over time. While much work has established a strong link between language use and emotional states, few have attempted to model emotional language in time. Here, we introduce the task of affective language forecasting – predicting future change in language based on past changes of language, a task with real-world applications such as treating mental health or forecasting trends in consumer confidence. We establish some of the fundamental autoregressive characteristics of the task (necessary history size, static versus dynamic length, varying time-step resolutions) and then build on popular sequence models for words to instead model sequences of language-based emotion in time. Over a novel Twitter dataset of 1,900 users and weekly + daily scores for 6 emotions and 2 additional linguistic attributes, we find a novel dual-sequence GRU model with decayed hidden states achieves best results (r = .66). We make our anonymized dataset as well as task setup and evaluation code available for others to build on.

Rapid Re-Identification Risk Assessment for Anonymous Data Set in Mobile Multimedia Scene

Ubiquitous mobile multimedia applications bring great convenience to users. However, when enjoying mobile multimedia services, users provide personal data to service platforms. Although the service platforms always claim that the collected personal data are de-identified, the risk of re-identifying users through linkage attacks still exists and is incalculable. This paper proposes a rapid prediction model for the overall re-identification risk based on the statistics of data sets (i.e., the number of individuals, number of attributes, distribution of attribute values, and attribute dependency). Our proposed model reveals the impact of statistics on the overall re-identification risk and adopts random sampling and semi-random sampling methods to predict the overall re-identification risk of data sets with and without strong dependency ordered attribute pairs. Experimental results show that for the data sets without strong dependency ordered attribute pairs, the random sampling method has a high prediction accuracy (the prediction error is less than 0.05). For the data sets with strong dependency ordered attribute pairs, the semi-random sampling method has a high prediction accuracy (the prediction error is less than 0.09). Exploiting our model, governments and individuals can quickly assess the privacy leakage risk of their data sets, given only the statistic of the data sets. Besides, this model can also evaluate the privacy risk of data collection schemes in advance according to historical statistics, and identify suspected services.

Image Detection Model for Construction Worker Safety Conditions using Faster R-CNN

Many accidents occur on construction sites leading to injury and death. According to the Occupational Safety Health Administration (OSHA), falls, electrocutions, being struck-by-objects and being caught in or between an object were the four main causes of worker deaths on construction sites. Many factors contribute to the increase in accidents, and personal protective equipment (PPE) is one of the defense mechanisms used to mitigate them. Thus, this paper presents an image detection model about workers’ safety conditions based on PPE compliance by using the Faster Region-based Convolutional Neural Networks (R-CNN) algorithm. This experiment was conducted using Tensorflow involving 1,129 images from the MIT Places Database (from Scene Recognition) as a training dataset, and 333 anonymous dataset images from real construction sites for evaluation purposes. The experimental results showed 276 of the images being detected as safe, and an average accuracy rate of 70%. The strength of this paper is based on the image detection of the three PPE combinations, involving hardhats, vests and boots in the case of construction workers. In future, the threshold and image sharpness (low resolution) will be two main characteristics of further refinement in order to improve the accuracy rate.

Behind the Mask: Understanding the Structural Forces That Make Social Graphs Vulnerable to Deanonymization

The tradeoff between anonymity and utility in the context of the anonymization of graph data sets is well acknowledged; for better privacy, some of the graph structural properties must be lost. What is not well understood, however, is what forces shape this tradeoff. Specifically, for the data practitioner who wants to publish an anonymized graph data set, it is unclear what graph structural properties can be preserved and what are the anonymity costs associated with preserving them. This article proposes a framework that examines the interplay between graph properties and the vulnerability to deanonymization attacks. We demonstrate its applicability via extensive experiments on thousands of graphs with controlled properties generated from real data sets. In addition, we show empirically that there are structural properties that affect graph vulnerability to reidentification attacks independent of degree distribution.

Analysis of disease comorbidity patterns in a large-scale China population

BackgroundDisease comorbidity is popular and has significant indications for disease progress and management. We aim to detect the general disease comorbidity patterns in Chinese populations using a large-scale clinical data set.MethodsWe extracted the diseases from a large-scale anonymized data set derived from 8,572,137 inpatients in 453 hospitals across China. We built a Disease Comorbidity Network (DCN) using correlation analysis and detected the topological patterns of disease comorbidity using both complex network and data mining methods. The comorbidity patterns were further validated by shared molecular mechanisms using disease-gene associations and pathways. To predict the disease occurrence during the whole disease progressions, we applied four machine learning methods to model the disease trajectories of patients.ResultsWe obtained the DCN with 5702 nodes and 258,535 edges, which shows a power law distribution of the degree and weight. It further indicated that there exists high heterogeneity of comorbidities for different diseases and we found that the DCN is a hierarchical modular network with community structures, which have both homogeneous and heterogeneous disease categories. Furthermore, adhering to the previous work from US and Europe populations, we found that the disease comorbidities have their shared underlying molecular mechanisms. Furthermore, take hypertension and psychiatric disease as instance, we used four classification methods to predicte the disease occurrence using the comorbid disease trajectories and obtained acceptable performance, in which in particular, random forest obtained an overall best performance (with F1-score 0.6689 for hypertension and 0.6802 for psychiatric disease).ConclusionsOur study indicates that disease comorbidity is significant and valuable to understand the disease incidences and their interactions in real-world populations, which will provide important insights for detection of the patterns of disease classification, diagnosis and prognosis.

Steered Microaggregation as a Unified Primitive to Anonymize Data Sets and Data Streams

As data grow in quantity and complexity, data anonymization is becoming increasingly challenging. On one side, a great diversity of masking methods, synthetic data generation methods, and privacy models exists, and this diversity is often perceived as unsettling by practitioners. On the other side, most of the anonymization methodology was designed for static, structured, and small data, whereas the current landscape includes big data and, in particular, data streams. We explore here a unified and conceptually simple anonymization approach, by presenting a primitive called steered microaggregation that can be tailored to enforce various privacy models on static data sets and also on data streams. Steered microaggregation is based on adding artificial attributes that are properly initialized and weighted in order to guide the microaggregation process into meeting certain desired constraints. To demonstrate the potential of this type of microaggregation, we show how it can be used to achieve $k$ -anonymity, $t$ -closeness, $l$ -diversity, and $\epsilon $ -differential privacy in the context of static data sets; furthermore, we discuss how it can be used to achieve $k$ -anonymity of data streams while controlling tuple reordering. Beyond its flexibility and theoretical appeal, steered microaggregation can drastically reduce information loss, as shown by our experimental evaluation.

Moral bureaucracies and social network research

In the wake of Europe’s General Data Protection Regulation (GDPR), research ethics governance does not just affect the ethical dimensions of social research but also the range of scientific decisions available to researchers. Because of the sensitive status of personal data and the aversion to even minimal risk by what we call “moral bureaucracies”, we are concerned that social network researchers will increasingly limit their research decisions to “safe” options, like reusing anonymized datasets, choosing target populations based on convenience rather than theoretical relevance, and routinely subcontracting fieldwork to professional data collection companies, among others. We also suggest that scientific associations and social scientists in general need to adopt a proactive role in preserving both scientific freedom and genuine ethics advice within this new regulatory framework.

Real-World Treatment Patterns, Health-Care Costs and Predictors for TKI Changes in CML: Results from a Population Representative German Claims Data Analysis

BACKGROUND AND AIMS Real-world data on usage of 1st, 2nd and 3rd generation TKIs in CML are limited. Furthermore, current information on cost of CML treatment from a European perspective is scarce. Thus, this study followed four objectives: First, to get an overview over incidence and prevalence data. Second, to assess TKIs used in 1st and 2nd-line treatment, frequency of TKI switches and differences in cardiovascular comorbidities in different TKI patient groups. Third, to find predictors for a change of TKI treatment. Finally, to assess the costs associated with CML from a payer's perspective and to stratify these to the different health-care sectors. METHODS This observational study was based on the InGef research database, an anonymized representative claims dataset containing 4 million persons insured through the German statutory health insurances. Patients with an ICD-10 code C92.1, C92.2, C92.7 or C92.9 and with at least one prescription of a TKI, combined with a reimbursement code for diagnostic testing were included. Analyses were divided in an incidence cohort (ICD-10 code in at least two quarters between Jan 01, 2013 and Dec 31, 2014 and no corresponding diagnosis in 2012) and a prevalence cohort (diagnoses were assessed in 2012 (index period)). Patients were followed-up for three years in the incidence and up to five years in the prevalence cohort. In the prevalence cohort, patients could be counted multiple times when TKI was switched within the study period. Predictors for a TKI change were assessed using a multivariate logit model with backwards selection in the prevalence cohort for each follow-up year. Age, gender, comorbidities and medication were used as predictors. Annual health-care costs from a payer's perspective were included in the database and assessed in the prevalence cohort and stratified in inpatient, outpatient, pharmaceutical and other costs. RESULTS 151 patients were included in the incidence cohort. This resulted in an extrapolated incidence and prevalence of 1,434 incident and 12,720 prevalent patients in Germany. Median age was 66 years and 46% of patients were female. Data on 1st line therapy was available for 124 patients and distributed across imatinib (N=52), nilotinib (N=44), dasatinib (N=12), hydroxycarbamide (N=11) and ponatinib/bosutinib (N=5). 22% of patients (N=32) switched TKI therapy at least once. Thereof, every second patient (N=18) switched to a 3rd line therapy. 43% of patients who started therapy on imatinib changed their treatment, compared to 34% in nilotinib and 14% in dasatinib. Median time to TKI switch was between 142 days (from imatinib to nilotinib) and 480 days (from nilotinib to dasatinib). In the prevalence cohort, 636 patients were included with a median age of 65, 42% of patients were female. 423 patients (66.5%) had claims on existing or newly emerged cardiovascular diseases. No significant differences between TKIs were found, with 63.9% of imatinib (N=212), 69.6% of nilotinib (N=151) and 63.3% of dasatinib treated patients (N=81) had claims for existing or newly developed cardiovascular diseases (p=0.32). The most important predictors for therapy line change are displayed in table 1. Mean annual costs for CML treatment were 58,230 €. Thereof, 56 % / 32,780 € were pharmaceutical (60-70% accounting on TKI), 28% / 16,383 € inpatient, 13% / 7,610 € outpatient costs. Median costs across different TKI therapies were 63,598 € for imatinib, 65,520 € for nilotinib and 88,342 € for dasatinib treated patients. CONCLUSION This study was the first German CML study based on a representative claims dataset. A difference in cardiovascular diseases across imatinib, nilotinib and dasatinib based on reported claims was not observed. Every fifth patient switched TKI therapy within the first three years after treatment initiation. Switches are more likely when signs of disease progression (e.g. comorbid ICD-10 code C91) or intolerability (e.g. prescription of anti-infective agents) are observable. To our knowledge, this is also the first study to report detailed cost data of CML in Germany. Costs between the TKIs differed only slightly with non-generic imatinib. With 12,720 prevalent CML patients, annual treatment costs amount to 740 million Euro in Germany. Additional information: This study was supported by Incyte. Table 1. Disclosures Saussele: Novartis: Honoraria, Research Funding; Pfizer: Honoraria; Incyte: Honoraria, Research Funding; BMS: Honoraria, Research Funding. Kaworski:Incyte: Employment. Vogelmann:Incyte: Consultancy, Honoraria. Schubert:Incyte: Consultancy, Honoraria.

Employment status and impact on mental wellbeing in the UK working age population: a cross-sectional analysis

BackgroundGlobal evidence suggests that exposure to non-permanent or precarious employment can have negative effects on mental wellbeing, but research from the UK is sparse. This study explored whether these findings were applicable to the UK working population. As the employment market is evolving with more precarious working opportunities, we also ascertained whether any type of employment contract arrangement was more beneficial to mental wellbeing than being unemployed. MethodsWe retrospectively analysed the Understanding Society Wave 7 dataset (anonymous pan-UK dataset from 2015). In this study, we only included respondents aged 18–64 years, in paid employment or unemployed (n=18 525, 8473 men, 10 052 women). We used the seven components of, and the overall, Short Warwick Edinburgh Mental Wellbeing Score. A score lower than 18·82 determined low mental wellbeing. Part-time contracts were defined as working 30 h or less per week. To examine the associations between the respective employment statuses (contract types, in or out of employment) and mental wellbeing, we did a χ2 analysis and calculated odds ratios (adjusted for age group and gender). FindingsThe prevalence of low mental wellbeing was highest in those unemployed (30·1%, n=438/1457), followed by those in non-permanent (14·3%, n=135/945) and permanent (12·1%, n=1943/16 123) employment arrangements. Compared with individuals employed on permanent contracts, those in non-permanent arrangements were more likely to report struggling to deal with problems (odds ratio 1·26, 95% CI 1·01–1·58, p=0·042), being unable to make up their mind (1·57, 1·20–2·06, p=0·001) or think clearly (1·35, 1·06–1·72, p=0·014), and report not feeling close to others (1·48, 1·21–1·81, p<0·0001). The association between contract types (permanent vs non-permanent) and low mental wellbeing was not significant (1·16, 0·96–1·40, p=0·13) as were the associations between contract types and future optimism, feeling useful, or feeling relaxed. Unemployed individuals were more likely to report low mental wellbeing than permanent (3·07, 2·72–3·47, p<0·0001), non-permanent (2·59, 2·09–3·21, p<0·0001), full-time (3·16, 2·79–3·58, p<0·0001), or part-time (2·60, 2·25–3·01, p<0·0001) workers. InterpretationPrecarious employment is associated with components of poorer mental wellbeing. However, the strongest associations for low mental wellbeing are observed in currently unemployed people. To understand and explore these complex relationships in greater detail, longitudinal research is required. FundingNone.

The effectiveness of forensic genealogy techniques in The united kingdom – an experimental assessment

The use of forensic genealogy techniques to identify Joseph James DeAngelo as the prime suspect in the Golden State Killer case in 2018 has opened up a new approach to investigation of cold cases. Since that breakthrough, these methods have been reportedly applied to more than 50 investigations. To date, all of these relate to investigations in the US, where the high uptake of “direct-to-consumer” (DTC) genetic testing by individuals conducting private ancestral research has provided the necessary publicly available data for successful forensic investigations. We have conducted a study to assess the likely effectiveness of forensic genealogy techniques if applied to investigations in the UK. Volunteers provided their own SNP array data, downloaded from a DTC provider of their choice. These data sets were anonymised and uploaded to the GEDmatch Genesis genealogy website, mimicking data sets from unsourced crime samples. A team of experienced genealogists then attempted to identify the donors of the anonymised data sets by identifying relatives on the database and triangulating to determine their shared family lineages which were further investigated using traditional resources (such as birth, marriage, death and census records). Four individuals out of a convenience sample of ten we correctly identified, at least to the level of one of a set of siblings.

DUEF-GA: data utility and privacy evaluation framework for graph anonymization

Anonymization of graph-based data is a problem which has been widely studied over the last years, and several anonymization methods have been developed. Information loss measures have been used to evaluate data utility and information loss in the anonymized graphs. However, there is no consensus about how to evaluate data utility and information loss in privacy-preserving and anonymization scenarios, where the anonymous datasets were perturbed to hinder re-identification processes. Authors use diverse metrics to evaluate data utility and, consequently, it is complex to compare different methods or algorithms in the literature. In this paper, we propose a framework to evaluate and compare anonymous datasets in a common way, providing an objective score to clearly compare methods and algorithms. Our framework includes metrics based on generic information loss measures, such as average distance or betweenness centrality and also task-specific information loss measures, such as community detection or information flow. Additionally, we provide some metrics to examine re-identification and risk assessment. We demonstrate that our framework could help researchers and practitioners to select the best parametrization and/or algorithm to reduce information loss and maximize data utility.

A Comparison of User Behaviors for a Fertility-Tracking App: Does Training in an NFP Method Improve Persistence and Use?

"CycleProGo™ users with NFP training were more persistent and diligent about daily data input than those without training."

Beyond the Click: A First Look at the Role of a Microblogging Platform in the Web Ecosystem

As the largest microblogging service in China, Sina Weibo, is associated with a huge amount of traffic to and from other major websites. Thanks to a massive set of anonymized data from a major operator covering a large-scale metropolitan area, we extract the click-based transitions between Weibo and other websites. Such transitions have not been looked into in most studies that rely on the visible hyper-links among websites. Based on this unique dataset, we show that 38% of the outgoing transitions (leaving Weibo) go to video clip services. Comparatively, links related to news and long videos are clicked less but posted more. This indicates that photograph and text-sharing in microblog have been gradually replaced by video clip sharing. Also, ignoring video clip sharing, 53% of the transitions go to five of the largest Chinese websites, showing how closely connected Weibo has become in the Chinese Web ecosystem. We further look at users’ click behavior in transitions, and find that through co-clustering there do exist a number of prevalent and distinct behavior patterns, suggesting that only seven co-clusters can capture the click behavior of 192 000 studied users. Overall, our work is the first to unveil the significant role of microblogging platforms in the Web ecosystem. Considering the massive user base of Weibo, our findings can aid the design of more personalized link recommendation in Weibo, thus generating more referral traffic, i.e., user visits, from Weibo to other major websites.

An Improved Classification Analysis on Utility Aware K-Anonymized Dataset

K anonymization is one of the worldwide used approaches to protect the individual records from the privacy leakage attack of Privacy Preserving Data Mining (PPDM) arena. Typically anonymized dataset will impact the effectiveness of data mining results. Anyhow, currently researchers of PPDM progress in driving their efforts in finding out the optimum trade-off between privacy and utility. This work tends in bringing out the optimum classifier from a set of best classifiers of data mining approaches that are capable enough in generating value-added classifying results on utility aware k-anonymized data set. We performed the analytical approach on the data set that are anonymized in sense of accompanying the anonymity utility factors like null values count and transformation pattern loss. The experimentation is done with three widely used classifiers HNB, PART and J48 and these classifiers are analysed with Accuracy, F-measure, and ROC-AUC which are literately proved to be the perfect measures of classification. Our experimental analysis reveals the best classifiers on the utility aware anonymized data sets of Cell oriented Anonymization (CoA), Attribute oriented Anonymization (AoA) and Record oriented Anonymization (RoA).

A New Method of Privacy Protection: Random k-Anonymous

A new k-anonymous method which is different from traditional k-anonymous was proposed to solve the problem of privacy protection. Specifically, numerical data achieves k-anonymous by adding noises, and categorical data achieves k-anonymous by using randomization. Using the above two methods, the drawback that at least k elements must have the same quasi identifier in the k-anonymous data set has been solved. Since the process of finding anonymous equivalence is very time consuming, a two-step clustering method is used to divide the original data set into equivalence classes. First, the original data set is divided into several different sub-datasets, and then the equivalence classes are formed in the sub-datasets, thus greatly reducing the computational cost of finding anonymous equivalence classes. The experiments are conducted on three different data sets, and the results show that the proposed method is more efficient and the information loss of anonymous dataset is much smaller.

Better Patient Outcomes Through Mining of Biomedical Big Data

Digitalization is changing healthcare today. Big data analytics of medical information allows diagnostics, therapy and development of personalized medicines, to provide unprecedented treatment. This leads to better patient outcomes, while containing costs. In this review, opportunities, challenges and solutions for this health-data revolution are discussed. Integration and near-instant-response analytics across large datasets ¬can support care-givers and researchers to test and discard hypotheses more quickly. Physicians want to compare a patient to other similar patients, to learn and communicate about treatment best-practices with peers, across large cohorts and sets of parameters. Real-time interactions between physician and patient are becoming more important, allowing ‘live’ support of patients instead of single interactions once every few weeks. Researchers from many disciplines (biomedical, payers, governments) want to interpret large anonymized datasets, to uncover trends in drug-candidate behavior, treatment regimens, clinical trials or reimbursements, and to act on those insights. These opportunities are however met by daunting challenges. Biomedical information is available in data silos of structured and unstructured formats (doctor letters, patient records, omics data, device data). Efficient usage of biomedical information is also hampered by data privacy concerns. This has led to a highly-regulated industry, as a result of which digitalization in healthcare has progressed slower than in other industries. This review concludes with examples of how integration and interpretation of big data can be used to break down data silos and pave the way to better patient outcomes, value-based care, and the creation of an intelligent enterprise for healthcare.

Revealing Privacy Vulnerabilities of Anonymous Trajectories

The proliferation of various mobile devices equipped with GPS positioning modules makes the collection of trajectories more easier than ever before, and more and more trajectory datasets have been available for business applications or academic researches. Normally, published trajectories are often anonymized by replacing real identities of mobile objects with pseudonyms (e.g., random identifiers); however, privacy leaks can hardly be prevented. In this paper, we introduce a novel paradigm of de-anonymization attack re-identifying trajectories of victims from anonymous trajectory datasets. Different from existing attacks, no background knowledge or side channel information about the target dataset is required. Instead, we claim that, for each moving object, there exist some mobility patterns that reflect the preference or usual behavior of the object, and will not change dramatically over a period of time. As long as those relatively stable patterns can be extracted from trajectories and be utilized as quasi-identifiers, trajectories can be linked to anonymous historical ones. To implement such kind of de-anonymization attacks, an adversary only needs to collect a few trajectory segments of a victim, the durations of which do not necessarily overlap with that of trajectories in the target dataset (in simple terms, those trajectory segments are not necessary sub-trajectories included in the target dataset). Since the movements of victims in public areas could be observed openly, an adversary can obtain traces or locations about the victims either by direct monitoring them (e.g., tracking) or from third parties (e.g., social-networks). Then, the adversary extracts useful patterns from both the historical trajectories in the accessible dataset and newly obtained trajectory segments of victims, the historical trajectory with most similar patterns to that of a victim is considered as belonging to the victim. In order to demonstrate the feasibility of such attacks, we conduct extensive trace-driven simulations. We extract road segment preferences and stop of interests from trajectories of vehicles, and construct feature vectors (mobility patterns) of vehicles according to them, used for trajectory comparisons. Simulation results show that the adversary could re-identify anonymous trajectories effectively.

Multiple Aneurysms AnaTomy CHallenge 2018 (MATCH): Phase I: Segmentation.

Advanced morphology analysis and image-based hemodynamic simulations are increasingly used to assess the rupture risk of intracranial aneurysms (IAs). However, the accuracy of those results strongly depends on the quality of the vessel wall segmentation. To evaluate state-of-the-art segmentation approaches, the Multiple Aneurysms AnaTomy CHallenge (MATCH) was announced. Participants carried out segmentation in three anonymized 3D DSA datasets (left and right anterior, posterior circulation) of a patient harboring five IAs. Qualitative and quantitative inter-group comparisons were carried out with respect to aneurysm volumes and ostia. Further, over- and undersegmentation were evaluated based on highly resolved 2D images. Finally, clinically relevant morphological parameters were calculated. Based on the contributions of 26 participating groups, the findings reveal that no consensus regarding segmentation software or underlying algorithms exists. Qualitative similarity of the aneurysm representations was obtained. However, inter-group differences occurred regarding the luminal surface quality, number of vessel branches considered, aneurysm volumes (up to 20%) and ostium surface areas (up to 30%). Further, a systematic oversegmentation of the 3D surfaces was observed with a difference of approximately 10% to the highly resolved 2D reference image. Particularly, the neck of the ruptured aneurysm was overrepresented by all groups except for one. Finally, morphology parameters (e.g., undulation and non-sphericity) varied up to 25%. MATCH provides an overview of segmentation methodologies for IAs and highlights the variability of surface reconstruction. Further, the study emphasizes the need for careful processing of initial segmentation results for a realistic assessment of clinically relevant morphological parameters.