Schaffter, T. et al, Albiol, F., & Caballero, L. (2020). Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms. JAMA Netw. Open, 3(3), e200265–15pp.
Abstract: Importance Mammography screening currently relies on subjective human interpretation. Artificial intelligence (AI) advances could be used to increase mammography screening accuracy by reducing missed cancers and false positives. Objective To evaluate whether AI can overcome human mammography interpretation limitations with a rigorous, unbiased evaluation of machine learning algorithms. Design, Setting, and Participants In this diagnostic accuracy study conducted between September 2016 and November 2017, an international, crowdsourced challenge was hosted to foster AI algorithm development focused on interpreting screening mammography. More than 1100 participants comprising 126 teams from 44 countries participated. Analysis began November 18, 2016. Main Outcomes and Measurements Algorithms used images alone (challenge 1) or combined images, previous examinations (if available), and clinical and demographic risk factor data (challenge 2) and output a score that translated to cancer yes/no within 12 months. Algorithm accuracy for breast cancer detection was evaluated using area under the curve and algorithm specificity compared with radiologists' specificity with radiologists' sensitivity set at 85.9% (United States) and 83.9% (Sweden). An ensemble method aggregating top-performing AI algorithms and radiologists' recall assessment was developed and evaluated. Results Overall, 144231 screening mammograms from 85580 US women (952 cancer positive <= 12 months from screening) were used for algorithm training and validation. A second independent validation cohort included 166578 examinations from 68008 Swedish women (780 cancer positive). The top-performing algorithm achieved an area under the curve of 0.858 (United States) and 0.903 (Sweden) and 66.2% (United States) and 81.2% (Sweden) specificity at the radiologists' sensitivity, lower than community-practice radiologists' specificity of 90.5% (United States) and 98.5% (Sweden). Combining top-performing algorithms and US radiologist assessments resulted in a higher area under the curve of 0.942 and achieved a significantly improved specificity (92.0%) at the same sensitivity. Conclusions and Relevance While no single AI algorithm outperformed radiologists, an ensemble of AI algorithms combined with radiologist assessment in a single-reader screening environment improved overall accuracy. This study underscores the potential of using machine learning methods for enhancing mammography screening interpretation. Question How do deep learning algorithms perform compared with radiologists in screening mammography interpretation? Findings In this diagnostic accuracy study using 144231 screening mammograms from 85580 women from the United States and 166578 screening mammograms from 68008 women from Sweden, no single artificial intelligence algorithm outperformed US community radiologist benchmarks; including clinical data and prior mammograms did not improve artificial intelligence performance. However, combining best-performing artificial intelligence algorithms with single-radiologist assessment demonstrated increased specificity. Meaning Integrating artificial intelligence to mammography interpretation in single-radiologist settings could yield significant performance improvements, with the potential to reduce health care system expenditures and address resource scarcity experienced in population-based screening programs. This diagnostic accuracy study evaluates whether artificial intelligence can overcome human mammography interpretation limits with a rigorous, unbiased evaluation of machine learning algorithms.
|
LHCb Collaboration(Aaij, R. et al), Henry, L., Jashal, B. K., Martinez-Vidal, F., Oyanguren, A., Remon Alepuz, C., et al. (2021). Observation of the Bs0 -> (DD -/+)-D-*+/- decay. J. High Energy Phys., 03(3), 099–19pp.
Abstract: A search for the B-s(0) -> D*(+/-) D--/+ decay is performed using proton-proton collision data at centre-of-mass energies of 7, 8 and 13TeV collected by the LHCb experiment, corresponding to an integrated luminosity of 9 fb(-1). The decay is observed with a high significance and its branching fraction relative to the B-0 -> D*(+/-) D--/+ decay is measured to be B(B-s(0) -> D*D-+/-(-/+))/B(B-0 -> D*D-+/-(-/+)) = 0.137 +/- 0.017 +/- 0.002 +/- 0.006, where the first uncertainty is statistical, the second systematic and the third is due to the uncertainty on the ratio of the B-s(0) and B-0 hadronisation fractions.
|
LHCb Collaboration(Aaij, R. et al), Henry, L., Jashal, B. K., Martinez-Vidal, F., Oyanguren, A., Remon Alepuz, C., et al. (2021). Observation of CP violation in two-body B-(s)(0)-meson decays to charged pions and kaons. J. High Energy Phys., 03(3), 075–43pp.
Abstract: The time-dependent CP asymmetries of B-0 -> pi(+)pi(-) and B-s(0) -> K+K- decays are measured using a data sample of pp collisions corresponding to an integrated luminosity of 1.9 fb(-1), collected with the LHCb detector at a centre-of-mass energy of 13TeV. The results are C-pi pi = 0.311 +/- 0.045 +/- 0.015; S-pi pi = 0.706 +/- 0.042 +/- 0.013; C-KK = 0.164 +/- 0.034 +/- 0.014; S-KK = 0.123 +/- 0.034 +/- 0.015; A(KK)(Delta Gamma) = -0.83 +/- 0.05 +/- 0.09; where the first uncertainties are statistical and the second systematic. The same data sample is used to measure the time-integrated CP asymmetries of B-0 -> K + pi(-) and B-s(0) -> K-pi(+) decays and the results are AB(CP)(B0) = -0.0824 +/- 0.0033 +/- 0.0033; A(CP)(Bs0) = 0.236 +/- 0.013 +/- 0.011. All results are consistent with earlier measurements. A combination of LHCb measurements provides the first observation of time-dependent CP violation in B-s(0) decays.
|
LHCb Collaboration(Aaij, R. et al), Henry, L., Jashal, B. K., Martinez-Vidal, F., Oyanguren, A., Remon Alepuz, C., et al. (2021). Measurement of the CKM angle gamma and Bs0-Bs0bar mixing frequency with Bs0 -> Ds-/+ h +/ pi+/- pi-/+ decays. J. High Energy Phys., 03(3), 137–46pp.
Abstract: The CKM angle gamma is measured for the first time from mixing-induced CP violation between Bs0 -> Ds -/+ K pi +/- pi -/+ and Bs0bar -> Ds +/- K -/+ pi -/+ pi +/- decays reconstructed in proton-proton collision data corresponding to an integrated luminosity of 9 fb(-1) recorded with the LHCb detector. A time-dependent amplitude analysis is performed to extract the CP-violating weak phase gamma – 2 beta (s) and, subsequently, gamma by taking the Bs0-Bs0bar mixing phase beta (s) as an external input. The measurement yields gamma = (44 +/- 12) degrees modulo 180 degrees, where statistical and systematic uncertainties are combined. An alternative model-independent measurement, integrating over the five-dimensional phase space of the decay, yields gamma = (44 -13+20) degrees modulo 180 degrees. Moreover, the Bs0-Bs0bar oscillation frequency is measured from the flavour-specific control channel Bs0 -> Ds- pi+ pi+ pi- to be m(s) = (17.757 +/- 0.007(stat) +/- 0.008(syst)) ps(-1), consistent with and more precise than the current world-average value.
|
Qin, W., Dai, L. Y., & Portoles, J. (2021). Two and three pseudoscalar production in e(+)e(-) annihilation and their contributions to (g-2)(mu). J. High Energy Phys., 03(3), 092–38pp.
Abstract: A coherent study of e(+)e(-) annihilation into two (pi(+)pi(-), K+K-) and three (pi(+)pi(-)pi(0), pi(+)pi(-)eta) pseudoscalar meson production is carried out within the framework of resonance chiral theory in energy region E less than or similar to 2 GeV. The work of [L.Y. Dai, J. Portoles, and O. Shekhovtsova, Phys. Rev. D88 (2013) 056001] is revisited with the latest experimental data and a joint analysis of two pseudoscalar meson production. Hence, we evaluate the lowest order hadronic vacuum polarization contributions of those two and three pseudoscalar processes to the anomalous magnetic moment of the muon. We also estimate some higher-order additions led by the same hadronic vacuum polarization. Combined with the other contributions from the standard model, the theoretical prediction differs still by (21.6 +/- 7.4) x 10(-10) (2.9 sigma) from the experimental value.
|