In a world focused on buzzword-driven models and algorithms, you’d be forgiven for forgetting about the unreasonable importance of data preparation and quality: your models are only as good as the data you feed them. This is the garbage in, garbage out principle: flawed data going in leads to flawed results, algorithms, and business decisions. […]

Three months ago, Apple released a new credit card in partnership with Goldman Sachs that aimed to disrupt the highly regulated world of consumer finance. However, a well-known software developer tweeted that he was given 20x the credit line offered to his wife, despite the fact that they have been filing joint tax returns and […]

Corporate Members Event

Event participants

Dr Christopher Smart, Whitehead Senior Fellow, Chatham House; Senior Fellow, Mossavar-Rahmani Center for Business and Government, Harvard Kennedy School; Special Assistant to President Obama, International Economics, Trade and Investment (2013-15)

Chair: Kenneth Cukier, Senior Editor of Digital Products, The Economist

28 June 2017

This paper examines how governments on both sides of the Atlantic are establishing frameworks that attempt to govern the commercial uses of data. It covers areas such as data analytics driving productivity and growth, the 'industrial internet of things', and the policy context and political forces shaping data rules in the US and Europe.

28 June 2017

Members Event Webinar

Event participants

David Aanensen, Director, The Centre for Genomic Pathogen Surveillance
Marietje Schaake, International Policy Director, Stanford University Cyber Policy Center
Stefaan Verhulst, Co-Founder and Chief of Research and Development, NYU Govlab
Chair: Marjorie Buchser, Executive Director, Digital Society Initiative, Chatham House

Roland Bruderer
May 1, 2015; 14:1400-1410
Research

M. Wang
Aug 1, 2012; 11:492-500
Technological Innovation and Resources

Alexey I. Nesvizhskii
Oct 1, 2005; 4:1419-1440
Tutorial

Ludovic C. Gillet
Jun 1, 2012; 11:O111.016717-O111.016717
Research

Hazy CEO Harry Keen explains why artificially produced data is gaining favour over information generated by real-world events

Esther McVey said the government will release local data on properties and land to help the proptech sector thrive

What happens when a service you shared your personal data with is acquired by a giant corporation?

Erik Burman, Ali Feizmohammadi and Lauri Oksanen
Math. Comp. 89 (2020), 1681-1709.
Abstract, references and article information

The ability to connect to a Google BigQuery database via OAuth is now available with this hot fix. Three new options have been added, REFRESH_TOKEN=, CLIENT_ID=, and CLIENT_SECRET=. You can use these options with 

The ability to connect to a Google BigQuery database via a proxy is available with this hot fix. You can use the newly added option, PROXY=, with the following methods of connection to the Google BigQuery database:

(Moscow Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies) Researchers of the Moscow Diagnostics and Telemedicine Center collected a dataset that includes more than a thousand sets of chest CT scans of patients with imaging finding of COVID-19. As of today, it is the largest completely anonymized database of CT studies, which has no analogues in Russia or in the world. It is available for download and can be used for developing services based on artificial intelligence technologies.

(DOE/Lawrence Berkeley National Laboratory) A new study, led by researchers at Berkeley Lab and UC Berkeley, suggests new paths for catching the signals of dark matter particles that have their energy absorbed by atomic nuclei.

(NASA/Goddard Space Flight Center) NASA has selected a new pathfinding CubeSat mission to gather data not collected since the agency flew the Dynamics Explorer in the early 1980s.

(American Association for the Advancement of Science) Ice sheet losses from Greenland and Antarctica have outpaced snow accumulation and contributed approximately 14 millimeters to sea level rise over 16 years (2003 to 2019), a new analysis of data from NASA's laser-shooting satellites has revealed.

Lila M. Gierasch
Apr 1, 2020; 19:569-570
Editorial

Lindsay K Pino
Apr 20, 2020; 0:P119.001913v1-mcp.P119.001913
Perspective

Head motion degrades image quality and causes erroneous parameter estimates in tracer kinetic modeling in brain PET studies. Existing motion correction methods include frame-based image-registration (FIR) and correction using real-time hardware-based motion tracking (HMT) information. However, FIR cannot correct for motion within one predefined scan period while HMT is not readily available in the clinic since it typically requires attaching a tracking device to the patient. In this study, we propose a motion correction framework with a data-driven algorithm, i.e., using the PET raw data itself, to address these limitations. Methods: We propose a data-driven algorithm, Centroid of Distribution (COD), to detect head motion. In COD, the central coordinates of the line of response (LOR) of all events are averaged over 1-sec intervals to generate a COD trace. A point-to-point change in the COD trace in one direction that exceeded a user-defined threshold was defined as a time point of head motion, which was followed by manually adding additional motion time points. All the frames defined by such time points were reconstructed without attenuation correction and rigidly registered to a reference frame. The resulting transformation matrices were then used to perform the final motion compensated reconstruction. We applied the new COD framework to 23 human dynamic datasets, all containing large head motions, with ¹⁸F-FDG (N = 13) and ¹¹C-UCB-J (N = 10), and compared its performance with FIR and with HMT using the Vicra, which can be considered as the "gold standard". Results: The COD method yielded 1.0±3.2% (mean ± standard deviation across all subjects and 12 grey matter regions) SUV difference for ¹⁸F-FDG (3.7±5.4% for ¹¹C-UCB-J) compared to HMT while no motion correction (NMC) and FIR yielded -15.7±12.2% (-20.5±15.8%) and -4.7±6.9% (-6.2±11.0%), respectively. For ¹⁸F-FDG dynamic studies, COD yielded differences of 3.6±10.9% in Ki value as compared to HMT, while NMC and FIR yielded -18.0±39.2% and -2.6±19.8%, respectively. For ¹¹C-UCB-J, COD yielded 3.7±5.2% differences in VT compared to HMT, while NMC and FIR yielded -20.0±12.5% and -5.3±9.4%, respectively. Conclusion: The proposed COD-based data-driven motion correction method outperformed FIR and achieved comparable or even better performance as compared to the Vicra HMT method in both static and dynamic studies.

Respiratory gating is the standard to overcome respiration effects degrading image quality in positron emission tomography (PET). Data-driven gating (DDG) using signals derived from PET raw data are promising alternatives to gating approaches requiring additional hardware. However, continuous bed motion (CBM) scans require dedicated DDG approaches for axially-extended PET, compared to DDG for conventional step-and-shoot scans. In this study, a CBM-capable DDG algorithm was investigated in a clinical cohort, comparing it to hardware-based gating using gated and fully motion-corrected reconstructions. Methods: 56 patients with suspected malignancies in thorax or abdomen underwent whole-body ¹⁸F-FDG CBM-PET/CT imaging using DDG and hardware-based respiratory gating (pressure-sensitive belt gating, BG). Correlation analyses were performed on both gating signals. Besides static reconstructions, BG and DDG were used for optimally-gated PET (BG-OG, DDG-OG) and fully motion-corrected PET (elastic motion correction; BG-EMOCO, DDG-EMOCO). Metabolic volumes, SUV_max and SUVmean of lesions were compared amongst the reconstructions. Additionally, the quality of lesion delineation in different PET reconstructions was independently evaluated by three experts. Results: Global correlation coefficients between BG and DDG signals amounted to 0.48±0.11, peaking at 0.89±0.07 when scanning the kidney and liver region. In total, 196 lesions were analyzed. SUV measurements were significantly higher in BG-OG, DDG-OG, BG-EMOCO and DDG-EMOCO compared to static images (P<0.001; median SUV_max: static, 14.3±13.4; BG-EMOCO, 19.8±15.7; DDG-EMOCO, 20.5±15.6; BG-OG, 19.6±17.1; DDG-OG, 18.9±16.6). No significant differences between BG-OG and DDG-OG, and BG-EMOCO and DDG-EMOCO, respectively, were found. Visual lesion delineation was significantly better in BG-EMOCO and DDG-EMOCO than in static reconstructions (P<0.001); no significant difference was found comparing BG and DDG (EMOCO, OG, respectively). Conclusion: DDG-based motion-compensation of CBM-PET acquisitions outperforms static reconstructions, delivering qualities comparable to hardware-based approaches. The new algorithm may be a valuable alternative for CBM-PET systems.

Rationale: Neuroinflammation has been implicated in Amyotrophic Lateral Sclerosis (ALS) and can be visualized using translocator protein (TSPO) radioligands. To become a reliable pharmacodynamic biomarker for ALS multicenter trials, some challenges have to be overcome. We aimed to investigate whether multicenter data pooling of different TSPO tracers (¹¹C-PBR28 and ¹⁸F-DPA714) is feasible, after validation of an established ¹¹C-PBR28 PET pseudoreference analysis technique for ¹⁸F-DPA714. Methods: 7 ALS-Belgium (58.9±6.7 years,5M) and 8 HV-Belgium (52.1±15.2 years,3M); and 7 ALS-US (53.4±9.8 years,5M) and 7 HV-US (54.6±9.6 years,4M) from a previously published study (1) underwent dynamic ¹⁸F-DPA714 (Leuven, Belgium) or ¹¹C-PBR28 (Boston, US) PET-MR scans. For ¹⁸F-DPA714, volume of distribution (VT) maps were compared to standardized uptake value ratios (SUVR)40-60 calculated using the pseudoreference regions (1)cerebellum, (2)occipital cortex, and (3)whole brain without ventricles (WB-ventricles). Also for ¹¹C-PBR28, SUVR60-90 using WB-ventricles were calculated. Results: In line with previous studies, increased ¹⁸F-DPA714 uptake (17.0±5.6%) in primary motor cortices was observed in ALS, as measured by both VT and SUVR40-60 approaches. Highest sensitivity was found for SUVRWB-ventricles (average cluster 21.6±0.1%). ¹⁸F-DPA714 VT ratio and SUVR40-60 results were highly correlated (r>0.8, p<0.001). A similar pattern of increased uptake (average cluster 20.5±0.5%) in primary motor cortices was observed in ALS with ¹¹C-PBR28 using the SUVRWB-ventricles. Analysis of the ¹⁸F-DPA714 and ¹¹C-PBR28 data together, resulted in a more extensive pattern of significant increased glial activation in the bilateral primary motor cortices. Conclusion: The same pseudoreference region analysis technique for ¹¹C-PBR28 PET imaging can be extended towards ¹⁸F-DPA714 PET. Therefore, in ALS, standardized analysis across these two tracers enables pooling of TSPO PET data across multiple centers and increase power of TSPO as biomarker for future therapeutic trials.

A data-driven method for respiratory gating in PET has recently been commercially developed. We sought to compare the performance of the algorithm to an external, device-based system for oncological [¹⁸F]-FDG PET/CT imaging. Methods: 144 whole-body [¹⁸F]-FDG PET/CT examinations were acquired using a Discovery D690 or D710 PET/CT scanner (GE Healthcare), with a respiratory gating waveform recorded by an external, device based respiratory gating system. In each examination, two of the bed positions covering the liver and lung bases were acquired with duration of 6 minutes. Quiescent period gating retaining ~50% of coincidences was then able to produce images with an effective duration of 3 minutes for these two bed positions, matching the other bed positions. For each exam, 4 reconstructions were performed and compared: data driven gating (DDG-retro), external device-based gating (RPM Gated), no gating but using only the first 3 minutes of data (Ungated Matched), and no gating retaining all coincidences (Ungated Full). Lesions in the images were quantified and image quality was scored by a radiologist, blinded to the method of data processing. Results: The use of DDG-retro was found to increase SUVmax and to decrease the threshold-defined lesion volume in comparison to each of the other reconstruction options. Compared to RPM-gated, DDG-retro gave an average increase in SUV_max of 0.66 ± 0.1 g/mL (n=87, p<0.0005). Although results from the blinded image evaluation were most commonly equivalent, DDG-retro was preferred over RPM gated in 13% of exams while the opposite occurred in just 2% of exams. This was a significant preference for DDG-retro (p=0.008, n=121). Liver lesions were identified in 23 exams. Considering this subset of data, DDG-retro was ranked superior to Ungated Full in 6/23 (26%) of cases. Gated reconstruction using the external device failed in 16% of exams, while DDG-retro always provided a clinically acceptable image. Conclusion: In this clinical evaluation, the data driven respiratory gating technique provided superior performance as compared to the external device-based system. For the majority of exams the performance was equivalent, but data driven respiratory gating had superior performance in 13% of exams, leading to a significant preference overall.

It was hypothesized that the brain β-amyloid buildup curve plateaus at an early symptomatic Alzheimer's disease (AD) stage. Atrophy-related partial volume effects (PVEs) degrade signal in hot-spot imaging techniques, such as amyloid positron emission tomography (PET). This longitudinal analysis of amyloid-sensitive PET data investigated the shape of the β-amyloid curve in AD applying PVE correction (PVEC). We analyzed baseline and 2-year follow-up data of 216 symptomatic individuals on the AD continuum (positive amyloid status) enrolled in Alzheimer's Disease Neuroimaging Initiative (17 AD dementia, 199 mild cognitive impairment), including 18F-florbetapir PET, magnetic resonance imaging and mini mental state examination (MMSE) scores. For PVEC, the modified Müller-Gärtner method was performed. Compared to non-PVE-corrected data, PVE-corrected data yielded significantly higher regional and composite standardized uptake value ratio (SUVR) changes over time (P=0.0002 for composite SUVRs). Longitudinal SUVR changes in relation to MMSE decreases showed a significantly higher slope of the regression line in the PVE-corrected as compared to the non-PVE-corrected PET data (F=7.1, P=0.008). These PVEC results indicate that the β-amyloid buildup curve does not plateau at an early symptomatic disease stage. A further evaluation of the impact of PVEC on the in-vivo characterization of time-dependent AD pathology, including the reliable assessment and comparison of other amyloid tracers, is warranted.

In bottom-up, label-free discovery proteomics, biological samples are acquired in a data-dependent (DDA) or data-independent (DIA) manner, with peptide signals recorded in an intact (MS1) and fragmented (MS2) form. While DDA has only the MS1 space for quantification, DIA contains both MS1 and MS2 at high quantitative quality. DIA profiles of complex biological matrices such as tissues or cells can contain quantitative interferences, and the interferences at the MS1 and the MS2 signals are often independent. When comparing biological conditions, the interferences can compromise the detection of differential peptide or protein abundance and lead to false positive or false negative conclusions.

We hypothesized that the combined use of MS1 and MS2 quantitative signals could improve our ability to detect differentially abundant proteins. Therefore, we developed a statistical procedure incorporating both MS1 and MS2 quantitative information of DIA. We benchmarked the performance of the MS1-MS2-combined method to the individual use of MS1 or MS2 in DIA using four previously published controlled mixtures, as well as in two previously unpublished controlled mixtures. In the majority of the comparisons, the combined method outperformed the individual use of MS1 or MS2. This was particularly true for comparisons with low fold changes, few replicates, and situations where MS1 and MS2 were of similar quality. When applied to a previously unpublished investigation of lung cancer, the MS1-MS2-combined method increased the coverage of known activated pathways.

Since recent technological developments continue to increase the quality of MS1 signals (e.g. using the BoxCar scan mode for Orbitrap instruments), the combination of the MS1 and MS2 information has a high potential for future statistical analysis of DIA data.