Deceptive learning in histopathology.

AIMS: Deep learning holds immense potential for histopathology, automating tasks that are simple for expert pathologists and revealing novel biology for tasks that were previously considered difficult or impossible to solve by eye alone. However, the extent to which the visual strategies learned by deep learning models in histopathological analysis are trustworthy or not has yet to be systematically analysed. Here, we systematically evaluate deep neural networks (DNNs) trained for histopathological analysis in order to understand if their learned strategies are trustworthy or deceptive. METHODS AND RESULTS: We trained a variety of DNNs on a novel data set of 221 whole-slide images (WSIs) from lung adenocarcinoma patients, and evaluated their effectiveness at (1) molecular profiling of KRAS versus EGFR mutations, (2) determining the primary tissue of a tumour and (3) tumour detection. While DNNs achieved above-chance performance on molecular profiling, they did so by exploiting correlations between histological subtypes and mutations, and failed to generalise to a challenging test set obtained through laser capture microdissection (LCM). In contrast, DNNs learned robust and trustworthy strategies for determining the primary tissue of a tumour as well as detecting and localising tumours in tissue. CONCLUSIONS: Our work demonstrates that DNNs hold immense promise for aiding pathologists in analysing tissue. However, they are also capable of achieving seemingly strong performance by learning deceptive strategies that leverage spurious correlations, and are ultimately unsuitable for research or clinical work. The framework we propose for model evaluation and interpretation is an important step towards developing reliable automated systems for histopathological analysis.

Design and fabrication of superamphiphobic paper surfaces.

Cellulose-based paper remains a vital component of modern day society; however, its use is severely limited in certain applications because of hydrophilic and oleophilic properties. In this manuscript we present a novel method to create superamphiphobic paper by combining the control of fiber size and structure with plasma etching and fluoropolymer deposition. The heterogeneous nature of the paper structure is drastically different from that of artificially created superamphiphobic surfaces. By refining the wood fibers, smaller diameter fibers (fibrils) are created to support fluid droplets. After oxygen plasma etching and deposition of a fluoropolymer film, paper samples are able to support motor oil contact angles of 149 ± 3°, although these structures readily absorb n-hexadecane. Exchange of water in the pulp solution with sec-butanol provides additional control over fiber spacing to create superamphiphobic substrates with contact angles >150° for water, ethylene glycol, motor oil, and n-hexadecane.

Creation of superhydrophobic stainless steel surfaces by acid treatments and hydrophobic film deposition.

In this work, we present a method to render stainless steel surfaces superhydrophobic while maintaining their corrosion resistance. Creation of surface roughness on 304 and 316 grade stainless steels was performed using a hydrofluoric acid bath. New insight into the etch process is developed through a detailed analysis of the chemical and physical changes that occur on the stainless steel surfaces. As a result of intergranular corrosion, along with metallic oxide and fluoride redeposition, surface roughness was generated on the nano- and microscales. Differences in alloy composition between 304 and 316 grades of stainless steel led to variations in etch rate and different levels of surface roughness for similar etch times. After fluorocarbon film deposition to lower the surface energy, etched samples of 304 and 316 stainless steel displayed maximum static water contact angles of 159.9 and 146.6°, respectively. However, etching in HF also caused both grades of stainless steel to be susceptible to corrosion. By passivating the HF-etched samples in a nitric acid bath, the corrosion resistant properties of stainless steels were recovered. When a three step process was used, consisting of etching, passivation and fluorocarbon deposition, 304 and 316 stainless steel samples exhibited maximum contact angles of 157.3 and 134.9°, respectively, while maintaining corrosion resistance.