Fine-Grained Self-Supervised Learning with Jigsaw puzzles for medical image classification.

Classifying fine-grained lesions is challenging due to minor and subtle differences in medical images. This is because learning features of fine-grained lesions with highly minor differences is very difficult in training deep neural networks. Therefore, in this paper, we introduce Fine-Grained Self-Supervised Learning(FG-SSL) method for classifying subtle lesions in medical images. The proposed method progressively learns the model through hierarchical block such that the cross-correlation between the fine-grained Jigsaw puzzle and regularized original images is close to the identity matrix. We also apply hierarchical block for progressive fine-grained learning, which extracts different information in each step, to supervised learning for discovering subtle differences. Our method does not require an asymmetric model, nor does a negative sampling strategy, and is not sensitive to batch size. We evaluate the proposed fine-grained self-supervised learning method on comprehensive experiments using various medical image recognition datasets. In our experiments, the proposed method performs favorably compared to existing state-of-the-art approaches on the widely-used ISIC2018, APTOS2019, and ISIC2017 datasets.

Towards long-tailed, multi-label disease classification from chest X-ray: Overview of the CXR-LT challenge.

Many real-world image recognition problems, such as diagnostic medical imaging exams, are "long-tailed" - there are a few common findings followed by many more relatively rare conditions. In chest radiography, diagnosis is both a long-tailed and multi-label problem, as patients often present with multiple findings simultaneously. While researchers have begun to study the problem of long-tailed learning in medical image recognition, few have studied the interaction of label imbalance and label co-occurrence posed by long-tailed, multi-label disease classification. To engage with the research community on this emerging topic, we conducted an open challenge, CXR-LT, on long-tailed, multi-label thorax disease classification from chest X-rays (CXRs). We publicly release a large-scale benchmark dataset of over 350,000 CXRs, each labeled with at least one of 26 clinical findings following a long-tailed distribution. We synthesize common themes of top-performing solutions, providing practical recommendations for long-tailed, multi-label medical image classification. Finally, we use these insights to propose a path forward involving vision-language foundation models for few- and zero-shot disease classification.

Light-Driven Fabrication of a Chiral Photonic Lattice of the Helical Nanofilament Liquid Crystal Phase.

A photonic lattice is an efficient platform for optically exploring quantum phenomena. However, its fabrication requires high costs and complex procedures when conventional materials, such as silicon or metals, are used. Here, we demonstrate a simple and cost-effective fabrication method for a reconfigurable chiral photonic lattice of the helical nanofilament (HNF) liquid crystal (LC) phase and diffraction grating showing wavelength-dependent diffraction with a rotated polarization state. Furthermore, the UV-exposed areas of the HNF film having chiral characteristics act as optical building blocks that induce resonant intensity modulation in the reflectance and transmittance modes and the optical rotation of the linear polarization. Our photonic lattice of the HNF can be an efficient platform for a chirality-embedded photonic lattice at a low cost.

Directing Polymorphism in the Helical Nanofilament Phase.

Herein, it is reported that the polymorphism in the helical nanofilament (HNF, B4 ) liquid-crystalline phase depends on the fabrication methods, that is, UV-driven formation and template-assisted self-assembly in the nanoconfined geometry. As a result, uniaxially oriented HNFs with different helical structures were obtained, in which generation of the twisted-ribbon and cylindrical-ribbon polymorphs showed that even the molecular lattice has a different orientation. The detailed structures were directly observed by SEM and grazing-incidence X-ray diffraction with synchrotron radiation. The resultant polymorphs could be used in chiro-optical applications due to the capability for fine control of the helical structures.

Security use of the chiral photonic film made of helical liquid crystal structures.

The color change of photonic crystals (PCs) has been widely studied due to their beauty and anti-counterfeiting applications. Herein, we demonstrated security codes based on chiral PCs that cannot be easily mimicked and are quite different from the conventional technology used currently. The chiral PCs can be made by self-assembly and the structural colors change based on the polarization of the light in the transmission mode. These color changes are easily detected in real-time and are useful in the fabrication of anti-counterfeiting patterns that show beautiful and diverse color changes with rotating polarizers. We believe this can provide a new platform in various security and color-based applications.

Nanoconfinement of the Low-Temperature Dark Conglomerate: Structural Control from Focal Conics to Helical Nanofilaments.

The helical nanofilament (HNF) and low-temperature dark conglomerate (DC) liquid-crystal (LC) phases of bent-core molecules show the same local layer structure but present different bulk morphologies. The DC phase is characterized by the formation of nanoscale toric focal conics, whereas the HNF phase is constructed of bundles of twisted layers. Although the local layer structure is similar in both phases, materials that form these phases tend to form one morphology in preference to the other. Targeted control of the nanostructures would provide pathways to potential applications and insight into how conditions drive a specific phase formation. Here, W624, a compound known to form the DC phase is confined in nanometer scale channels of porous anodized aluminum oxide (AAO) membranes. Within each nanochannel, the DC phase is suppressed forming the HNF structure instead, indicating the nanoscale spatial limitation can control the phase structure of the DC phase.

Manipulation of Supramolecular Columnar Structures of H-Bonded Donor-Acceptor Units through Geometrical Nanoconfinement.

Ambipolar organic semiconductors are considered promising for organic electronics because of their interesting electric properties. Many hurdles remain yet to be overcome before they can be used for practical applications, especially because their orientation is hard to control. We demonstrate a method to control the orientation of columnar structures based on a hydrogen (H)-bonded donor-acceptor complex between a star-shaped tris(triazolyl)triazine and triphenylene-containing benzoic acid, using physicochemical nanoconfinement. The molecular configuration and supramolecular columnar assemblies in a one-dimensional porous anodic aluminium oxide (AAO) film were dramatically modulated by controlling the pore-size and by chemical modification of the inner surface of the porous AAO film. In situ experiments using grazing-incidence X-ray diffraction (GIXRD) were carried out to investigate the structural evolution produced at the nanometer scale by varying physicochemical conditions. The resulting highly ordered nanostructures may open a new pathway to effectively control the alignment of liquid crystal ambipolar semiconductors.

Simple Solvent Engineering for High-Mobility and Thermally Robust Conjugated Polymer Nanowire Field-Effect Transistors.

Electron donor (D)-acceptor (A)-type conjugated polymers (CPs) have emerged as promising semiconductor candidates for organic field-effect transistors. Despite their high charge carrier mobilities, optimization of electrical properties of D-A-type CPs generally suffers from complicated post-deposition treatments such as high-temperature thermal annealing or solvent-vapor annealing. In this work, we report a high-mobility diketopyrrolopyrrole-based D-A-type CP nanowires, self-assembled by a simple but very effective solvent engineering method that requires no additional processes after film deposition. In situ grown uniform nanowires at room temperature were shown to possess distinct edge-on chain orientation that is beneficial for lateral charge transport between source and drain electrodes in FETs. FETs based on the polymer nanowire networks exhibit impressive hole mobility of up to 4.0 cm2 V-1 s-1. Moreover, nanowire FETs showed excellent operational stability in high temperature up to 200 °C because of the strong interchain interaction and alignment.