The nutritional composition of three recently domesticated culinary-medicinal mushroom species (Oudemansiella sudmusida, Lentinus squarrosulus, and Tremella aurantialba) was evaluated for contents of protein, fiber, fat, total sugar content, amino acid, carbohydrate, and nucleotide components. The data indicated that fruiting bodies of these three mushroom species contained abundant nutritional substances. The protein contents of L. squarrosulus and O. submucida were 26.32% and 14.70%, which could be comparable to other commercially cultivated species. T. aurantialba contained 74.11% of carbohydrate, of which soluble polysaccharide was 40.55%. Oudemansiella sudmusida contained 15.95% of arabitol as the highest sugar alcohol in three mushrooms. These mushrooms also possessed distinct taste by their flavor component composition. Among them, L. squarrosulus contained 10.68% and 9.25% of monosodium glutamate-like and sweet amino acids, which were higher than the other two mushrooms. However, the nucleotide amounts of the three mushrooms were all lower than those of other commercially cultivated mushrooms. Among them, L. squarrosulus contained the highest amount of flavor nucleotides, which was 1.01%o. Results revealed that these three mushroom species are potentially suitable resources for commercial cultivation and healthy food.

Deep analysis of radiographic images can quantify the extent of intra-tumoral heterogeneity for personalized medicine. In this paper, we propose a novel content-based multi-feature image retrieval (CBMFIR) scheme to discriminate pulmonary nodules benign or malignant. Two types of features are applied to represent the pulmonary nodules. With each type of features, a single-feature distance metric model is proposed to measure the similarity of pulmonary nodules. And then, multiple single-feature distance metric models learned from different types of features are combined to a multi-feature distance metric model. Finally, the learned multi-feature distance metric is used to construct a content-based image retrieval (CBIR) scheme to assist the doctors in diagnosis of pulmonary nodules. The classification accuracy and retrieval accuracy are used to evaluate the performance of the scheme. The classification accuracy is 0.955 ± 0.010, and the retrieval accuracies outperform the comparison methods. The proposed CBMFIR scheme is effective in diagnosis of pulmonary nodules. Our method can better integrate multiple types of features from pulmonary nodules.

Rib fracture is the most common thoracic clinical trauma. Most patients have multiple different types of rib fracture regions, so accurate and rapid identification of all trauma regions is crucial for the treatment of rib fracture patients. In this study, a two-stage rib fracture recognition model based on nnU-Net is proposed. First, a deep learning segmentation model is trained to generate candidate rib fracture regions, and then, a deep learning classification model is trained in the second stage to classify the segmented local fracture regions according to the candidate fracture regions generated in the first stage to determine whether they are fractures or not. The results show that the two-stage deep learning model proposed in this study improves the accuracy of rib fracture recognition and reduces the false-positive and false-negative rates of rib fracture detection, which can better assist doctors in fracture region recognition.

Nowadays, computer aided diagnosis (CAD) system become popular because it improves the diagnostic efficiency of the disease compared to the early diagnosis of the various diseases for the doctors and the medical expert specialists. Similarly, bone fractures is a common problem due to pressure, accident and osteoporosis. Bone fractures detection using computer vision is getting more and more important in CAD system because it can help to reduce workload of the doctor by screening out the easy case. The purpose of this work was to detect whether fractures occurred in seven different parts of the human body in the MURA database. The process of a bone fractures detection is mainly divided into three steps, which includes preprocessing, feature extraction and classification. In preprocessing, the 40561 sizes of images in the MURA data set were reshaped to same size and the grayscale images were changed to the RGB images. The convolutional neural network (CNN) was then used to extract the depth features of the bone images. Finally, three classic classifiers were selected, such as support vector machine (SVM), extreme learning machine (ELM) and random forest (RF) to detect bone cracks or non-cracks. Then compare it with the Alexnet model and the classification results of the Wu Enda team. The result shows that the classification of SVM based depth features is the best and the accuracy can reach 78.63%. The experimental result indicates that the depth features of the extracted bone images provide a new feature representation method for the prediction of bone fracturess, which greatly improves the accuracy of clinical diagnosis of bone fracturess.