Double-Balanced Loss for Imbalanced Colorectal Lesion Classification.

Colorectal cancer has a high incidence rate in all countries around the world, and the survival rate of patients is improved by early detection. With the development of object detection technology based on deep learning, computer-aided diagnosis of colonoscopy medical images becomes a reality, which can effectively reduce the occurrence of missed diagnosis and misdiagnosis. In medical image recognition, the assumption that training samples follow independent identical distribution (IID) is the key to the high accuracy of deep learning. However, the classification of medical images is unbalanced in most cases. This paper proposes a new loss function named the double-balanced loss function for the deep learning model, to improve the impact of datasets on classification accuracy. It introduces the effects of sample size and sample difficulty to the loss calculation and deals with both sample size imbalance and sample difficulty imbalance. And it combines with deep learning to build the medical diagnosis model for colorectal cancer. Experimentally verified by three colorectal white-light endoscopic image datasets, the double-balanced loss function proposed in this paper has better performance on the imbalance classification problem of colorectal medical images.

White-Light Endoscopic Colorectal Lesion Detection Based on Improved YOLOv5.

As an effective tool for colorectal lesion detection, it is still difficult to avoid the phenomenon of missed and false detection when using white-light endoscopy. In order to improve the lesion detection rate of colorectal cancer patients, this paper proposes a real-time lesion diagnosis model (YOLOv5x-CG) based on YOLOv5 improvement. In this diagnostic model, colorectal lesions were subdivided into three categories: micropolyps, adenomas, and cancer. In the course of convolutional network training, Mosaic data enhancement strategy was used to improve the detection rate of small target polyps. At the same time, coordinate attention (CA) mechanism was introduced to take into account channel and location information in the network, so as to realize the effective extraction of three kinds of pathological features. The Ghost module was also used to generate more feature maps through linear processing, which reduces the stress of learning model parameters and speeds up detection. The experimental results show that the lesion diagnosis model proposed in this paper has a more rapid and accurate lesion detection ability, and the AP value of polyps, adenomas, and cancer is 0.923, 0.955, and 0.87, and mAP@50 is 0.916.

Spontaneous curling of freestanding Janus monolayer transition-metal dichalcogenides.

In this paper, using molecular dynamics simulations we report spontaneous curling behaviors of freestanding Janus monolayer S-Mo-Se (MoSeS) structures. Density functional theory calculations are performed to obtain the phonon dispersion and phonon spectra of the Janus monolayer MoSeS for analyzing its structural stability. The results show that the Janus monolayer MoSeS is structurally stable. Due to the lattice mismatch between MoS and MoSe domains, the Janus monolayer MoSeS at the freestanding state always spontaneously rolls up in a constant temperature and pressure system. The direction of curling is preferred along the armchair orientation. Specifically, as for the Janus monolayer MoSeS whose size is larger than ∼30 nm, it can spontaneously roll up into a nanotube structure. The underlying physical mechanisms of these phenomena are well uncovered by using classical Timoshenko plate theory and the minimum energy principle.

The negative Poisson's ratio in graphene-based carbon foams.

Using molecular dynamics simulations, we find an in-plane negative Poisson's ratio intrinsically existing in the graphene-based three-dimensional (3D) carbon foams (CFs) when they are compressed uniaxially. Our study shows that the negative Poisson's ratio in the present CFs is attributed to their unique molecular structures and triggered by the buckling of the CF structures. This mechanism makes the negative Poisson's ratio of CFs strongly depend on their cell length, which offers us an efficient means to tune the negative Poisson's ratio in nanomaterials. Moreover, as the buckling modes of CFs are topographically different when they are compressed in different directions, their negative Poisson's ratio is found to be strongly anisotropic, which is in contrast to the isotropic positive Poisson's ratio observed in CFs prior to buckling. The discovery of the intrinsic negative Poisson's ratio in 3D CFs will significantly expand the family of auxetic nanomaterials. Meanwhile, the mechanism of nano-auxetics proposed here may open up a door to manufacture new auxetic materials on the nanoscale.

Effect of Crystal Orientation on Femtosecond Laser-Induced Thermomechanical Responses and Spallation Behaviors of Copper Films.

Ultrafast thermomechanical responses and spallation behaviours of monocrystal copper films irradiated by femtosecond laser pulse are investigated using molecular dynamics simulation (MDS). Films with 〈100〉, 〈110〉 and 〈111〉 crystal orientations along the thickness direction were studied. The results show that the crystal orientation has a significant effect on femtosecond laser-induced thermomechanical responses and spallation behaviors of monocrystal copper films. The discrepancy between normal stresses in copper films with different crystal orientation leads to distinct differences in lattice temperature. Moreover, the copper films with different crystal orientations present distinct spallation behaviors, including structural melting (atomic splashing) and fracture. The melting depth of 〈100〉 copper film is lower than that of 〈110〉 and 〈111〉 copper films for the same laser intensity. The dislocations and slip bands are formed and propagate from the solid-liquid interface of 〈110〉 and 〈111〉 copper films, while these phenomena do not appear in 〈100〉 copper film. Additionally, numerous slip bands are generated in the non-irradiated surface region of copper films due to reflection of mechanical stress. These slip bands can finally evolve into cracks (nanovoids) with time, which further result in the fracture of the entire films.

Effects of atomic vacancies and temperature on the tensile properties of single-walled MoS2 nanotubes.

Using molecular dynamics simulations, we study the effects of Mo and S atomic vacancies and different temperatures on the tensile properties of single-walled MoS2 nanotubes through a series of tensile tests. Both armchair and zigzag MoS2 nanotubes under uniaxial tensions show phase transitions. Two types of Mo-S bonds play different roles in this phase transition of MoS2 nanotubes. Moreover, the influences of Mo and S atomic vacancies and temperature on the Young's modulus, ultimate strength and fracture strain of single-walled MoS2 nanotubes are investigated systematically. The results show that Mo and S atomic vacancies have no influence on the Young's modulus of MoS2 nanotubes. However, Mo atomic vacancies result in a significant decrease of ultimate strength and fracture strain of MoS2 nanotubes, while S atomic vacancies have a relatively small influence on the fracture properties of MoS2 nanotubes. With an increase in temperature, the Young's modulus and ultimate strength decrease. When the temperature is higher than 300 K, the fracture is changed from brittle to ductile together with an enhanced fracture strain.

Fracture behaviors of pre-cracked monolayer molybdenum disulfide: A molecular dynamics study.

The fracture strength and crack propagation of monolayer molybdenum disulfide (MoS2) sheets with various pre-existing cracks are investigated using molecular dynamics simulation (MDS). The uniaxial tensions of pre-cracked monolayer MoS2 sheets with different crack tips, different locations of crack, different crack lengths and angled cracks are simulated and studied. The results show that the configuration of crack tip can influence significantly the fracture behaviors of monolayer MoS2 sheets while the location of crack does not influence the fracture strength. With the increase of crack length, the fracture strength of monolayer MoS2 sheets reduces almost linearly, and the fracture of monolayer MoS2 sheets is transformed from almost brittle to ductile. By making comparison between the MDS results and the predictions of continuum fracture mechanics theories, including Inglis' model, Griffith's model with and without finite size effect, it is found that MDS results agree well with the predictions of Griffith's model with finite size effect, differ from the predictions of Inglis' model and Griffith's model without finite size effect. Finally, the MDS results of monolayer MoS2 sheets with different angled crack are also analyzed based on the continuum fracture mechanics model.