No abstract

Automatic detection and classification of the masses in mammograms are still a big challenge and play a crucial role to assist radiologists for accurate diagnosis. In this paper, we propose a novel computer-aided diagnose (CAD) system based on one of the regional deep learning techniques: a ROI-based Convolutional Neural Network (CNN) which is called You Only Look Once (YOLO). Our proposed YOLO-based CAD system contains four main stages: mammograms preprocessing, feature extraction utilizing multi convolutional deep layers, mass detection with confidence model, and finally mass classification using fully connected neural network (FC-NN). A set of training mammograms with the information of ROI masses and their types are used to train YOLO. The trained YOLO-based CAD system detects the masses and classifies their types into benign or malignant. Our results show that the proposed YOLO-based CAD system detects the mass location with an overall accuracy of 96.33%. The system also distinguishes between benign and malignant lesions with an overall accuracy of 85.52%. Our proposed system seems to be feasible as a CAD system capable of detection and classification at the same time. It also overcomes some challenging breast cancer cases such as the mass existing in the pectoral muscles or dense regions.

Dual high and low energy images of Dual Energy X-ray Absorptiometry (DEXA) suffer from noises due to the use of weak amount of X-rays. Denoising these DEXA images could be a key process to enhance and improve a Bone Mineral Density (BMD) map which is derived from a pair of high and low energy images. This could further improve the accuracy of diagnosis of bone fractures, osteoporosis, and etc. In this paper, we present a denoising technique for dual high and low energy images of DEXA via non-local means filter (NLMF). The noise of dual DEXA images is modeled based on both source and detector noises of a DEXA system. Then, the parameters of the proposed NLMF are optimized for denoising utilizing the experimental data from uniform phantoms. The optimized NLMF is tested and verified with the DEXA images of the uniform phantoms and real human spine. The quantitative evaluation shows the improvement of Signal-to-Noise Ratio (SNR) for the high and low phantom images on the order of 30.36% and 27.02% and for the high and low real spine images on the order of 22.28% and 33.43%, respectively. Our work suggests that denoising via NLMF could be a key preprocessing process for clinical DEXA imaging.

Anthropomorphic robotic hands are designed to attain dexterous movements and flexibility much like human hands. Achieving human-like object manipulation remains a challenge especially due to the control complexity of the anthropomorphic robotic hand with a high degree of freedom. In this work, we propose a deep reinforcement learning (DRL) to train a policy using a synergy space for generating natural grasping and relocation of variously shaped objects using an anthropomorphic robotic hand. A synergy space is created using a continuous normalizing flow network with point clouds of haptic areas, representing natural hand poses obtained from human grasping demonstrations. The DRL policy accesses the synergistic representation and derives natural hand poses through a deep regressor for object grasping and relocation tasks. Our proposed synergy-based DRL achieves an average success rate of 88.38% for the object manipulation tasks, while the standard DRL without synergy space only achieves 50.66%. Qualitative results show the proposed synergy-based DRL policy produces human-like finger placements over the surface of each object including apple, banana, flashlight, camera, lightbulb, and hammer.

In this study, the authors propose a novel three‐dimensional (3D) convolutional neural network for shape reconstruction via a trilateral convolutional neural network (Tri‐CNN) from a single depth view. The proposed approach produces a 3D voxel representation of an object, derived from a partial object surface in a single depth image. The proposed Tri‐CNN combines three dilated convolutions in 3D to expand the convolutional receptive field more efficiently to learn shape reconstructions. To evaluate the proposed Tri‐CNN in terms of reconstruction performance, the publicly available ShapeNet and Big Data for Grasp Planning data sets are utilised. The reconstruction performance was evaluated against four conventional deep learning approaches: namely, fully connected convolutional neural network, baseline CNN, autoencoder CNN, and a generative adversarial reconstruction network. The proposed experimental results show that Tri‐CNN produces superior reconstruction results in terms of intersection over union values and Brier scores with significantly less number of model parameters and memory.