Peter Harris

Peter Harris

Auto-generated profile, based on publicly available information

Peter Harris

From the Department of Scientific Computing (S.E., S.G., M.T., M.D.H.B., N.S.C., P.H., L.M.W.) and National Co-ordinating Centre for the Physics of Mammography (K.C.Y.), Royal Surrey NHS Foundation Trust, Egerton Road, Guildford GU2 7XX, England; and Centre for Vision, Speech and Signal Processing (M.D.H.B.) and Department of Physics (K.C.Y.), University of Surrey, Guildford, England

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 July 2024 in Radiology: Artificial Intelligence
Verified
Journal Article
Closed access
SESam Ellis
Sam Ellis
SGSandra Gomes
Sandra Gomes
MTMatthew Trumble
Matthew Trumble
MHMark D. Halling-Brown
Mark D. Halling-Brown
KYKenneth C. Young
Kenneth C. Young
LWLucy M. Warren
Lucy M. Warren
"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop an artificial intelligence (AI) deep learning tool capable of predicting future breast cancer risk from a current negative screening mammographic examination and to evaluate the model on data from the UK National Health Service Breast Screening Program. Materials and Methods The OPTIMAM Mammography Imaging Database contains screening data, including mammograms and information on interval cancers, for > 300,000 women who attended screening at three different sites in the UK from 2012 onward. Cancer-free screening examinations from women aged 50-70 years were obtained and classified as risk-positive or risk-negative based on the occurrence of cancer within 3 years of the original examination. Examinations with confirmed cancer and images containing implants were excluded. From the resulting 5264 risk-positive and 191488 risk-negative examinations, training (n = 89285) validation (n = 2106) and test (n = 39351) datasets were produced for model development and evaluation. The AI model was trained to predict future cancer occurrence based on screening mammograms and patient age. Performance was evaluated on the test dataset using the area under the receiver operating characteristic curve (AUC) and compared across subpopulations to assess potential biases. Interpretability of the model was explored, including with saliency maps. Results On the hold-out test set, the AI model achieved an overall AUC of 0.70 (95% CI: 0.69, 0.72). There was no evidence of a difference in performance across the three sites, between patient ethnicities or across age-groups Visualization of saliency maps and sample images provided insights into the mammographic features associated with AI-predicted cancer risk. Conclusion The developed AI tool showed good performance on a multisite, UK-specific dataset. ©RSNA, 2024.

3 March 2010 in Measurement Science and Technology
Verified
Journal Article
Closed access
PCP Clarkson
P Clarkson
TET J Esward
T J Esward
PHP M Harris
P M Harris
ASA A Smith
A A Smith
ISIan Martin Smith
Ian Martin Smith

1 July 2011
Verified
Conference Paper
Open access
YBYann G BOUCHER
Yann G BOUCHER
FVF. Viallefont
F. Viallefont
ADA. Deadman
A. Deadman
NFNigel Fox
Nigel Fox
IBI. Behnert
I. Behnert
Sindy SterckxSindy Sterckx
Sindy Sterckx
A field campaign had been organized in August 2010 on Tuz Golu salt lake, Turkey, with the aim of characterizing the site for satellite optical sensor vicarious calibration, and of comparing different methodologies of surface reflectance factor characterization. Several teams have made ground based reflectance measurements with a field spectrometer on different areas of the salt lake of 100 m × 300 m and 1km × 1 km size. Different types of sampling strategies and measurements methods have been used by the participants, and are described in this paper. Preliminary results on one area are presented, that show a good agreement between the different measurements.

1 June 2020
Verified
Conference Paper
Closed access
YLYuhui Luo
Yuhui Luo
PHPeter Harris
Peter Harris
In this paper, we study the problem of process optimisation, exemplified by a testbed concerned with radial forging of parts, henceforth called the STRATH testbed. A total of 81 parts have been forged for the test. The forging process has been recorded with 99 sensors. Given these time series observations, the aim of the study is to predict or classify the dimensional variations of manufactured metal parts. Classification for STRATH testbed is challenging. This is because not only there are very limited number of completed parts in current dataset, but also the dimensions of parts are all within the tolerance range. Meanwhile, the dimensional data does not have clear cluster structure. However, despite the disadvantage stated above, STRATH testbed still serves as a good example for demonstrating the treatment of uncertainty as part of the data analysis for such problems. It shows that uncertainty is essential and valuable in all stages of signal processing and uncertainty can be propagated down the signal processing chain.

23 March 2020
Verified
Other
Open access
SCSei-Him Cheong
Sei-Him Cheong
SRStephen P Robinson
Stephen P Robinson
PHPeter M Harris
Peter M Harris
LWLian S Wang
Lian S Wang
VLValerie Livina
Valerie Livina
Underwater noise is recognised as a form of marine pollutant and there is evidence that over exposure to excessive levels of noise can have effects on the wellbeing of the marine ecosystem. Consequently, the variation in the ambient sound levels in the deep ocean has been the subject of a number of recent studies, with particular interest in the identification of long-term trends. We describe a statistical method for performing long-term trend analysis and uncertainty evaluation of the estimated trends from deep-ocean noise data. This study has been extended to include  measured data  from four monitoring stations located in the Indian (Cape Leeuwin & Diego Garcia), Pacific (Wake Island) and Southern Atlantic (Ascension Islands) Oceans over periods spanning between 8 to 15 years. The data were obtained from the hydro-acoustic monitoring stations of the Preparatory Commission for the Comprehensive Nuclear Test Ban Treaty Organization (CTBTO). The monitoring stations provide information at a sampling frequency of 250 Hz, leading to very large datasets, and at acoustic frequencies up to 105 Hz.The analysis method uses a flexible discrete model that incorporates terms that capture seasonal variations in the data together with a moving-average statistical model to describe the serial correlation of residual deviations. The trend analysis is applied to time series representing daily aggregated statistical levels for four frequency bands to obtain estimates for the change in sound pressure level (SPL) over the examined period with associated coverage intervals. The analysis demonstrates that there are statistically significant changes in the levels of deep-ocean noise over periods exceeding a decade. The main features of the approach include (a) using a functional model  with terms  that represent both long-term and seasonal behaviour of deep-ocean noise, (b) using a statistical model to capture the serial correlation of the residual deviations that are not explained by the functional model, (c) using daily aggregation intervals derived from 1-minute  sound pressure level averages, and (d) applying a non-parametric approach to validate the uncertainties of the trend estimates that avoids the need to make an assumption about the distribution of the residual deviations.The obtained results show the long term trends vary differently at the four stations. It was observed that low frequency noise generally dominated the significant trends in these oceans. The relative differences between the various statistical levels are remarkably similar for all the frequency bands. Given the complexity of the acoustic environment, it is difficult to identify the main causes of these trends. Some possible explanations for the observed trends are discussed. It was however observed some stations are subjected to strong seasonal variation with a high degree of correlation with climatic factors such as sea surface temperature, Antarctic ice coverage and wind speed. The same seasonal effects is less pronounced in station located closer to the equator.