Universal lesion detection (ULD) on computed tomography (CT) images is an important but underdeveloped problem. Recently, deep learning-based approaches have been proposed for ULD, aiming to learn representative features from annotated CT data. However, the hunger for data of deep learning models and the scarcity of medical annotation hinders these approaches to advance further. In this paper, we propose to incorporate domain knowledge in clinical practice into the model design of universal lesion detectors. Specifically, as radiologists tend to inspect multiple windows for an accurate diagnosis, we explicitly model this process and propose a multi-view feature pyramid network (FPN), where multi-view features are extracted from images rendered with varied window widths and window levels; to effectively combine this multi-view information, we further propose a position-aware attention module. With the proposed model design, the data-hunger problem is relieved as the learning task is made easier with the correctly induced clinical practice prior. We show promising results with the proposed model, achieving an absolute gain of 5.65% (in the sensitivity of FPs@4.0) over the previous state-of-the-art on the NIH DeepLesion dataset.

No abstract

No abstract

Aluminum (Al) is one of the most widely used metals for industry and household applications, but its longevity is limited by its tendency for corrosion. In this work, we report a facile method to fabricate superhydrophobic Al surfaces that have excellent anti-corrosion effect. The surface is obtained by etching Al in CuCl2 solution to form the micro–nano-pit surface texture followed by lowering its surface energy in an aqueous ethanol solution of stearic acid. The superhydrophobic Al surfaces show water contact angles as high as 165°. Electrochemical tests demonstrate that the corrosion rate of the Al surface drops by 94.5% after the superhydrophobic modification (corrosion current density lowers from 1.11 × 10–4 to 6.10 × 10–6 A cm–2). We also show that the superhydrophobic surface will protect the Al from corrosion even under a very harsh environment. In addition, our method is scalable and the superhydrophobic surfaces exhibit excellent flexible and reparable properties. This anti-corrosive superhydrophobic Al surface will prolong Al in its broad usage.