Registries,Databases and Repositories for Developing Artificial Intelligence in Cancer Care

Modern artificial intelligence techniques have solved some previously intractable problems and produced impressive results in selected medical domains. One of their drawbacks is that they often need very large amounts of data. Pre-existing datasets in the form of national cancer registries, image/genetic depositories and clinical datasets already exist and have been used for research. In theory, the combination of healthcare Big Data with modern, data-hungry artificial intelligence techniques should offer significant opportunities for artificial intelligence development, but this has not yet happened. Here we discuss some of the structural reasons for this, barriers preventing artificial intelligence from making full use of existing datasets, and make suggestions as to enable progress. To do this, we use the framework of the 6Vs of Big Data and the FAIR criteria for data sharing and availability (Findability, Accessibility, Interoperability, and Reuse). We share our experience in navigating these barriers through The Brain Tumour Data Accelerator, a Brain Tumour Charity-supported initiative to integrate fragmented patient data into an enriched dataset. We conclude with some comments as to the limits of such approaches.     

Bench to bedside: recent advances in lung cancer pathological research with implications for diagnostic clinical practice

After decades of relative stagnation lung cancer is emerging as a disease type where rapid progress is being made in diagnosis and therapy, as well as in our understanding of disease biology. Much of this progress is of immediate impact to diagnosticians, and more is likely to affect diagnostic practice in the near future. In this review we seek to briefly summarize several key areas of active research of immediate or probable imminent value to trainee and consultant pulmonary pathologists alike. We cover some major changes in tumour classification, grading, and patient stratification, as well as considering the state of the art in machine-assisted interpretation of lung cancer histology, and the use of genetically modified lung cancer models.     

Utilization of and Outcomes Associated with Intravascular Ultrasound during Deep Venous Stent Placement among Medicare Beneficiaries

PurposeTo evaluate temporal trends, practice variation, and associated outcomes with the use of intravascular ultrasound (US) during deep venous stent placement among Medicare beneficiaries.Materials and MethodsAll lower extremity deep venous stent placement procedures performed between January 1, 2017, and December 31, 2019 among Medicare beneficiaries were included. Temporal trends in intravascular US use were stratified by procedural setting and physician specialty. The primary outcome was a composite of 12-month all-cause mortality, all-cause hospitalization, or repeat target vessel intervention. The secondary outcome was a composite of 12-month stent thrombosis, embolization, or restenosis.ResultsAmong the 20,984 deep venous interventions performed during the study period, 15,184 (72.4%) utilized intravascular US. Moderate growth in intravascular US use was observed during the study period in all clinical settings. There was a variation in the use of intravascular US among all operators (median, 77.3% of cases; interquartile range, 20.0%–99.2%). In weighted analyses, intravascular US use during deep venous stent placement was associated with a lower risk of both the primary (adjusted hazard ratio, 0.72; 95% confidence interval [CI], 0.69–0.76; P < .001) and secondary (adjusted hazard ratio, 0.32; 95% CI, 0.27–0.39; P < .001) composite end points.ConclusionsIntravascular US is frequently used during deep venous stent placement among Medicare beneficiaries, with further increase in use from 2017 to 2019. The utilization of intravascular US as part of a procedural strategy was associated with a lower cumulative incidence of adverse outcomes after the procedure, including venous stent thrombosis and embolization.     

DEep learning-based rapid Spiral Image REconstruction (DESIRE) for high-resolution spiral first-pass myocardial perfusion imaging

The objective of the current study was to develop and evaluate a DEep learning-based rapid Spiral Image REconstruction (DESIRE) technique for high-resolution spiral first-pass myocardial perfusion imaging with whole-heart coverage, to provide fast and accurate image reconstruction for both single-slice (SS) and simultaneous multislice (SMS) acquisitions. Three-dimensional U-Net–based image enhancement architectures were evaluated for high-resolution spiral perfusion imaging at 3 T. The SS and SMS MB = 2 networks were trained on SS perfusion images from 156 slices from 20 subjects. Structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and normalized root mean square error (NRMSE) were assessed, and prospective images were blindly graded by two experienced cardiologists (5: excellent; 1: poor). Excellent performance was demonstrated for the proposed technique. For SS, SSIM, PSNR, and NRMSE were 0.977 [0.972, 0.982], 42.113 [40.174, 43.493] dB, and 0.102 [0.080, 0.125], respectively, for the best network. For SMS MB = 2 retrospective data, SSIM, PSNR, and NRMSE were 0.961 [0.950, 0.969], 40.834 [39.619, 42.004] dB, and 0.107 [0.086, 0.133], respectively, for the best network. The image quality scores were 4.5 [4.1, 4.8], 4.5 [4.3, 4.6], 3.5 [3.3, 4], and 3.5 [3.3, 3.8] for SS DESIRE, SS L1-SPIRiT, MB = 2 DESIRE, and MB = 2 SMS-slice-L1-SPIRiT, respectively, showing no statistically significant difference (p = 1 and p = 1 for SS and SMS, respectively) between L1-SPIRiT and the proposed DESIRE technique. The network inference time was ~100 ms per dynamic perfusion series with DESIRE, while the reconstruction time of L1-SPIRiT with GPU acceleration was ~ 30 min. It was concluded that DESIRE enabled fast and high-quality image reconstruction for both SS and SMS MB = 2 whole-heart high-resolution spiral perfusion imaging.     

Frame Invariance and Scalability of Neural Operators for Partial Differential Equations

Partial differential equations (PDEs) play a dominant role in the mathematical modeling of many complex dynamical processes. Solving these PDEs often requires prohibitively high computational costs, especially when multiple evaluations must be made for different parameters or conditions. After training, neural operators can provide PDEs solutions significantly faster than traditional PDE solvers. In this work, invariance properties and computational complexity of two neural operators are examined for transport PDE of a scalar quantity. Neural operator based on graph kernel network (GKN) operates on graph-structured data to incorporate nonlocal dependencies. Here we propose a modified formulation of GKN to achieve frame invariance. Vector cloud neural network (VCNN) is an alternate neural operator with embedded frame invariance which operates on point cloud data. GKN-based neural operator demonstrates slightly better predictive performance compared to VCNN. However, GKN requires an excessively high computational cost that increases quadratically with the increasing number of discretized objects as compared to a linear increase for VCNN.     

Deepm5C: A deep-learning-based hybrid framework for identifying human RNA N5-methylcytosine sites using a stacking strategy

1. Download : Download high-res image (173KB)
2. Download : Download full-size image

     

Ovine Iliac Vein Model for Endovascular Thrombectomy of Acute Deep Venous Thrombosis

An ovine iliac vein thrombosis model was devised to test a wall-contacting rotational thrombectomy device. Thrombosis was successfully induced in 9 sheep with an average clot length of 31 mm ± 12 and >60% vessel occlusion on angiography. The thrombus was subsequently removed, maintaining normal intraoperative pulmonary arterial pressure (5.9 mm Hg ± 3.6) and complete distal reperfusion after thrombectomy. Additionally, the sheep were without signs of vascular trauma or embolic complications on gross necropsy and histopathologic analysis. The findings from this study support the use of an ovine iliac deep vein thrombosis model for testing of a lower extremity thrombectomy device.     

不同介入方案治疗急性下肢静脉血栓的临床效果与安全性比较

目的 探讨不同介入方案治疗急性下肢静脉血栓(DVT)的临床效果与安全性。方法 选取我院2015年1月～2019年12月介入科收治的急性DVT患者54例作为研究对象,按照治疗方式的不同分为研究组27例和对照组27例。研究组采用血管腔内综合治疗,对照组采用导管介导溶栓治疗,比较两组患者疗效评价、治疗前后膝上、下15 cm与患肢周径差、治疗前后皮肤肿胀张力、静脉通畅率、住院时间和并发症发生情况。结果 经治疗后研究组患者膝上、下15 cm与患肢周径差均短于对照组(P0.05);经治疗后研究组患者皮肤肿胀张力恢复程度显著优于对照组(P0.05);经治疗后研究组患者静脉通畅率显著高于对照组(P0.05);研究组治疗有效率显著高于对照组(P0.05);研究组患者住院时间明显少于对照组,且其并发症发生率显著低于对照组(P0.05)。结论 血管腔内综合介入治疗能够有效改善患者下肢症状,提高DVT恢复情况,降低并发症的发生,从而有效提高患者生活质量,为临床治疗DVT提供重要参考。