Human Activity Prediction Based on Forecasted IMU Activity Signals by Sequence-to-Sequence Deep Neural Networks.

Human Activity Recognition (HAR) has gained significant attention due to its broad range of applications, such as healthcare, industrial work safety, activity assistance, and driver monitoring. Most prior HAR systems are based on recorded sensor data (i.e., past information) recognizing human activities. In fact, HAR works based on future sensor data to predict human activities are rare. Human Activity Prediction (HAP) can benefit in multiple applications, such as fall detection or exercise routines, to prevent injuries. This work presents a novel HAP system based on forecasted activity data of Inertial Measurement Units (IMU). Our HAP system consists of a deep learning forecaster of IMU activity signals and a deep learning classifier to recognize future activities. Our deep learning forecaster model is based on a Sequence-to-Sequence structure with attention and positional encoding layers. Then, a pre-trained deep learning Bi-LSTM classifier is used to classify future activities based on the forecasted IMU data. We have tested our HAP system for five daily activities with two tri-axial IMU sensors. The forecasted signals show an average correlation of 91.6% to the actual measured signals of the five activities. The proposed HAP system achieves an average accuracy of 97.96% in predicting future activities.

Development and validation of a deep learning model to diagnose COVID-19 using time-series heart rate values before the onset of symptoms.

One of the effective ways to minimize the spread of COVID-19 infection is to diagnose it as early as possible before the onset of symptoms. In addition, if the infection can be simply diagnosed using a smartwatch, the effectiveness of preventing the spread will be greatly increased. In this study, we aimed to develop a deep learning model to diagnose COVID-19 before the onset of symptoms using heart rate (HR) data obtained from a smartwatch. In the deep learning model for the diagnosis, we proposed a transformer model that learns HR variability patterns in presymptom by tracking relationships in sequential HR data. In the cross-validation (CV) results from the COVID-19 unvaccinated patients, our proposed deep learning model exhibited high accuracy metrics: sensitivity of 84.38%, specificity of 85.25%, accuracy of 84.85%, balanced accuracy of 84.81%, and area under the receiver operating characteristics (AUROC) of 0.8778. Furthermore, we validated our model using external multiple datasets including healthy subjects, COVID-19 patients, as well as vaccinated patients. In the external healthy subject group, our model also achieved high specificity of 77.80%. In the external COVID-19 unvaccinated patient group, our model also provided similar accuracy metrics to those from the CV: balanced accuracy of 87.23% and AUROC of 0.8897. In the COVID-19 vaccinated patients, the balanced accuracy and AUROC dropped by 66.67% and 0.8072, respectively. The first finding in this study is that our proposed deep learning model can simply and accurately diagnose COVID-19 patients using HRs obtained from a smartwatch before the onset of symptoms. The second finding is that the model trained from unvaccinated patients may provide less accurate diagnosis performance compared with the vaccinated patients. The last finding is that the model trained in a certain period of time may provide degraded diagnosis performances as the virus continues to mutate.

Real-Time Human Activity Recognition with IMU and Encoder Sensors in Wearable Exoskeleton Robot via Deep Learning Networks.

Wearable exoskeleton robots have become a promising technology for supporting human motions in multiple tasks. Activity recognition in real-time provides useful information to enhance the robot's control assistance for daily tasks. This work implements a real-time activity recognition system based on the activity signals of an inertial measurement unit (IMU) and a pair of rotary encoders integrated into the exoskeleton robot. Five deep learning models have been trained and evaluated for activity recognition. As a result, a subset of optimized deep learning models was transferred to an edge device for real-time evaluation in a continuous action environment using eight common human tasks: stand, bend, crouch, walk, sit-down, sit-up, and ascend and descend stairs. These eight robot wearer's activities are recognized with an average accuracy of 97.35% in real-time tests, with an inference time under 10 ms and an overall latency of 0.506 s per recognition using the selected edge device.

BCNet: A Deep Learning Computer-Aided Diagnosis Framework for Human Peripheral Blood Cell Identification.

Blood cells carry important information that can be used to represent a person's current state of health. The identification of different types of blood cells in a timely and precise manner is essential to cutting the infection risks that people face on a daily basis. The BCNet is an artificial intelligence (AI)-based deep learning (DL) framework that was proposed based on the capability of transfer learning with a convolutional neural network to rapidly and automatically identify the blood cells in an eight-class identification scenario: Basophil, Eosinophil, Erythroblast, Immature Granulocytes, Lymphocyte, Monocyte, Neutrophil, and Platelet. For the purpose of establishing the dependability and viability of BCNet, exhaustive experiments consisting of five-fold cross-validation tests are carried out. Using the transfer learning strategy, we conducted in-depth comprehensive experiments on the proposed BCNet's architecture and test it with three optimizers of ADAM, RMSprop (RMSP), and stochastic gradient descent (SGD). Meanwhile, the performance of the proposed BCNet is directly compared using the same dataset with the state-of-the-art deep learning models of DensNet, ResNet, Inception, and MobileNet. When employing the different optimizers, the BCNet framework demonstrated better classification performance with ADAM and RMSP optimizers. The best evaluation performance was achieved using the RMSP optimizer in terms of 98.51% accuracy and 96.24% F1-score. Compared with the baseline model, the BCNet clearly improved the prediction accuracy performance 1.94%, 3.33%, and 1.65% using the optimizers of ADAM, RMSP, and SGD, respectively. The proposed BCNet model outperformed the AI models of DenseNet, ResNet, Inception, and MobileNet in terms of the testing time of a single blood cell image by 10.98, 4.26, 2.03, and 0.21 msec. In comparison to the most recent deep learning models, the BCNet model could be able to generate encouraging outcomes. It is essential for the advancement of healthcare facilities to have such a recognition rate improving the detection performance of the blood cells.

Accelerating 3D Convolutional Neural Network with Channel Bottleneck Module for EEG-Based Emotion Recognition.

Deep learning-based emotion recognition using EEG has received increasing attention in recent years. The existing studies on emotion recognition show great variability in their employed methods including the choice of deep learning approaches and the type of input features. Although deep learning models for EEG-based emotion recognition can deliver superior accuracy, it comes at the cost of high computational complexity. Here, we propose a novel 3D convolutional neural network with a channel bottleneck module (CNN-BN) model for EEG-based emotion recognition, with the aim of accelerating the CNN computation without a significant loss in classification accuracy. To this end, we constructed a 3D spatiotemporal representation of EEG signals as the input of our proposed model. Our CNN-BN model extracts spatiotemporal EEG features, which effectively utilize the spatial and temporal information in EEG. We evaluated the performance of the CNN-BN model in the valence and arousal classification tasks. Our proposed CNN-BN model achieved an average accuracy of 99.1% and 99.5% for valence and arousal, respectively, on the DEAP dataset, while significantly reducing the number of parameters by 93.08% and FLOPs by 94.94%. The CNN-BN model with fewer parameters based on 3D EEG spatiotemporal representation outperforms the state-of-the-art models. Our proposed CNN-BN model with a better parameter efficiency has excellent potential for accelerating CNN-based emotion recognition without losing classification performance.

Development of a Cs-free negative hydrogen ion source system using multi-pulsed plasma sources.

The Korea Atomic Energy Research Institute has recently proposed and developed a novel cesium-free negative hydrogen/deuterium ion source system based on two pulsed plasma sources for fusion and particle accelerator applications. The main feature of this ion source system is the use of both magnetic filters and plasma pulsing (also called the temporal filter). The system operates with two alternate pulsing sequences related to the respective plasma sources, thereby switching the plasmas in the after-glow state in an alternating manner. This study investigates the temporal behavior of deuterium negative ions in the system in a qualitative way by conducting a time-resolved measurement of laser photodetachment current commensurate with the negative ion density. In preliminary experiments, the current in the initial after-glow state remains higher than in the active-glow state identical to a steady-state continuous wave plasma, and the ratio reaches a maximum of about three times. This indicates that the pulsing gives highly efficient negative ion volume formation. Furthermore, it is observed that the time duration when the current is maintained at high values can be prolonged (or modulated) with the alternate dual pulsing, which is not possible with conventional single pulsing. These results provide a clue that the multi-pulsed ion source system may offer a continuous supply of negative ions at high densities and consequently become an alternative to cesium seeded ion sources.

Intramedullary nailing of subtrochanteric fractures in elderly patients: Comparative study of helical blade cephalomedullary nail versus reconstruction nail.

INTRODUCTION: Intramedullary nailing (IMN), which is a common method for treating subtrochanteric fractures, is conducted as cephalomedullary (CMN) or reconstruction (RCN) nailing. Numerous studies have reported the effectiveness of CMN, which requires a shorter surgery time and provides stronger fixation strength with blade-type devices. However, the radiographic and clinical outcomes of the use of CMN and RCN in elderly patients aged ≥65 years have not been compared yet. This study aimed to investigate whether CMN offers superior outcomes over RCN in the treatment of subtrochanteric fractures in elderly patients. MATERIALS AND METHODS: This retrospective study included 60 elderly patients (17 men and 43 women; mean age: 74.9 years) diagnosed with subtrochanteric fractures and treated with IMN with helical blade CMN (CMN group: 30 patients) or RCN (RCN group: 30 patients) between January 2013 and December 2018 with at least 1 year of follow-up period. Radiologic outcomes were evaluated based on the postoperative state of alignment and the achievement and timing of bony union at the final follow-up. Clinical outcomes were evaluated using the Merle d'Aubigné-Postel score. Radiologic and clinical outcomes in the two groups were compared and analyzed, and the occurrence of complications was examined. RESULTS: The difference in malalignment between the two groups was not significant; however, the RCN group achieved more effective reduction. At the final follow-up, bony union was achieved within 18.9 weeks, on average, in 28 patients in the CMN group and within 21.6 weeks, on average, in 27 patients in the RCN group. Twenty patients in the CMN group and 26 in the RCN group showed good or better results according to the Merle d'Aubigné-Postel score. No significant differences were found for any of the parameters. CONCLUSIONS: In the treatment of difficult subtrochanteric fractures in elderly patients, RCN can provide excellent reduction and strong fixation similar to CMN and can result in outstanding clinical and radiologic outcomes.

Object Manipulation with an Anthropomorphic Robotic Hand via Deep Reinforcement Learning with a Synergy Space of Natural Hand Poses.

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.

Staged Management of Severely Displaced Calcaneal Fractures With Transarticular Pinning: A Damage Control Strategy.

BACKGROUND: Severely displaced calcaneal fractures can result in considerable morphology derangement and may be accompanied by soft tissue compromise. Delayed operative restoration of the calcaneal morphology may result in acute retensioning of the damaged soft tissue with associated wound-related complications. In this study, we describe a staged treatment of displaced intra-articular calcaneal fractures that uses temporary transarticular Kirschner wire (K-wire) fixation and staged conversion to definite fixation. METHODS: We identified all of the patients who were treated at our institution for calcaneal fractures between 2015 and 2019. A total of 17 patients with 20 calcaneal fractures were selectively treated with 2-stage management. Temporary transarticular K-wire fixation was performed 24 hours after the injury to restore calcaneal morphology and the surrounding soft tissue. After the soft tissue was considered safe, delayed open reduction and internal fixation was performed. The time to definite surgery, radiographic alignment, wound complications, time to radiographic union, and hindfoot American Orthopaedic Foot & Ankle Society (AOFAS) scores were recorded. RESULTS: The average follow-up period was 17 months (range, 12-43). The average Böhler angle increased from a mean of -22 degrees (range, -109 to 25) to 25 degrees (range, 0 to 47) after temporary transarticular K-wire fixation. The mean time from temporary pinning to conversion to definite internal fixation was 20 (range, 10-32) days. There were no immediate postoperative complications. The average time to radiographic union was 13.7 (range, 10-16) weeks. The mean AOFAS score was 87 (range, 55-100). No infections or wound complications were reported during the follow-up period. CONCLUSION: Temporary transarticular pinning for staged calcaneal fracture treatment is safe and effective in restoring the calcaneal morphology. This novel and relatively simple method may facilitate delayed operation and decrease wound-related complications. LEVEL OF EVIDENCE: Level IV, retrospective case series.

Hand Gesture Recognition Using Single Patchable Six-Axis Inertial Measurement Unit via Recurrent Neural Networks.

Recording human gestures from a wearable sensor produces valuable information to implement control gestures or in healthcare services. The wearable sensor is required to be small and easily worn. Advances in miniaturized sensor and materials research produces patchable inertial measurement units (IMUs). In this paper, a hand gesture recognition system using a single patchable six-axis IMU attached at the wrist via recurrent neural networks (RNN) is presented. The IMU comprises IC-based electronic components on a stretchable, adhesive substrate with serpentine-structured interconnections. The proposed patchable IMU with soft form-factors can be worn in close contact with the human body, comfortably adapting to skin deformations. Thus, signal distortion (i.e., motion artifacts) produced for vibration during the motion is minimized. Also, our patchable IMU has a wireless communication (i.e., Bluetooth) module to continuously send the sensed signals to any processing device. Our hand gesture recognition system was evaluated, attaching the proposed patchable six-axis IMU on the right wrist of five people to recognize three hand gestures using two models based on recurrent neural nets. The RNN-based models are trained and validated using a public database. The preliminary results show that our proposed patchable IMU have potential to continuously monitor people's motions in remote settings for applications in mobile health, human-computer interaction, and control gestures recognition.

Evaluation of deep learning detection and classification towards computer-aided diagnosis of breast lesions in digital X-ray mammograms.

BACKGROUND AND OBJECTIVE: Deep learning detection and classification from medical imagery are key components for computer-aided diagnosis (CAD) systems to efficiently support physicians leading to an accurate diagnosis of breast lesions. METHODS: In this study, an integrated CAD system of deep learning detection and classification is proposed aiming to improve the diagnostic performance of breast lesions. First, a deep learning YOLO detector is adopted and evaluated for breast lesion detection from entire mammograms. Then, three deep learning classifiers, namely regular feedforward CNN, ResNet-50, and InceptionResNet-V2, are modified and evaluated for breast lesion classification. The proposed deep learning system is evaluated over 5-fold cross-validation tests using two different and widely used databases of digital X-ray mammograms: DDSM and INbreast. RESULTS: The evaluation results of breast lesion detection show the capability of the YOLO detector to achieve overall detection accuracies of 99.17% and 97.27% and F1-scores of 99.28% and 98.02% for DDSM and INbreast datasets, respectively. Meanwhile, the YOLO detector could predict 71 frames per second (FPS) at the testing time for both DDSM and INbreast datasets. Using detected breast lesions, the classification models of CNN, ResNet-50, and InceptionResNet-V2 achieve promising average overall accuracies of 94.50%, 95.83%, and 97.50%, respectively, for the DDSM dataset and 88.74%, 92.55%, and 95.32%, respectively, for the INbreast dataset. CONCLUSION: The capability of the YOLO detector boosted the classification models to achieve a promising breast lesion diagnostic performance. Such prediction results should help to develop a feasible CAD system for practical breast cancer diagnosis.

Multiple skin lesions diagnostics via integrated deep convolutional networks for segmentation and classification.

BACKGROUND AND OBJECTIVE: Computer automated diagnosis of various skin lesions through medical dermoscopy images remains a challenging task. METHODS: In this work, we propose an integrated diagnostic framework that combines a skin lesion boundary segmentation stage and a multiple skin lesions classification stage. Firstly, we segment the skin lesion boundaries from the entire dermoscopy images using deep learning full resolution convolutional network (FrCN). Then, a convolutional neural network classifier (i.e., Inception-v3, ResNet-50, Inception-ResNet-v2, and DenseNet-201) is applied on the segmented skin lesions for classification. The former stage is a critical prerequisite step for skin lesion diagnosis since it extracts prominent features of various types of skin lesions. A promising classifier is selected by testing well-established classification convolutional neural networks. The proposed integrated deep learning model has been evaluated using three independent datasets (i.e., International Skin Imaging Collaboration (ISIC) 2016, 2017, and 2018, which contain two, three, and seven types of skin lesions, respectively) with proper balancing, segmentation, and augmentation. RESULTS: In the integrated diagnostic system, segmented lesions improve the classification performance of Inception-ResNet-v2 by 2.72% and 4.71% in terms of the F1-score for benign and malignant cases of the ISIC 2016 test dataset, respectively. The classifiers of Inception-v3, ResNet-50, Inception-ResNet-v2, and DenseNet-201 exhibit their capability with overall weighted prediction accuracies of 77.04%, 79.95%, 81.79%, and 81.27% for two classes of ISIC 2016, 81.29%, 81.57%, 81.34%, and 73.44% for three classes of ISIC 2017, and 88.05%, 89.28%, 87.74%, and 88.70% for seven classes of ISIC 2018, respectively, demonstrating the superior performance of ResNet-50. CONCLUSIONS: The proposed integrated diagnostic networks could be used to support and aid dermatologists for further improvement in skin cancer diagnosis.

Deep Learning Computer-Aided Diagnosis for Breast Lesion in Digital Mammogram.

For computer-aided diagnosis (CAD), detection, segmentation, and classification from medical imagery are three key components to efficiently assist physicians for accurate diagnosis. In this chapter, a completely integrated CAD system based on deep learning is presented to diagnose breast lesions from digital X-ray mammograms involving detection, segmentation, and classification. To automatically detect breast lesions from mammograms, a regional deep learning approach called You-Only-Look-Once (YOLO) is used. To segment breast lesions, full resolution convolutional network (FrCN), a novel segmentation model of deep network, is implemented and used. Finally, three conventional deep learning models including regular feedforward CNN, ResNet-50, and InceptionResNet-V2 are separately adopted and used to classify or recognize the detected and segmented breast lesion as either benign or malignant. To evaluate the integrated CAD system for detection, segmentation, and classification, the publicly available and annotated INbreast database is used over fivefold cross-validation tests. The evaluation results of the YOLO-based detection achieved detection accuracy of 97.27%, Matthews's correlation coefficient (MCC) of 93.93%, and F1-score of 98.02%. Moreover, the results of the breast lesion segmentation via FrCN achieved an overall accuracy of 92.97%, MCC of 85.93%, Dice (F1-score) of 92.69%, and Jaccard similarity coefficient of 86.37%. The detected and segmented breast lesions are classified via CNN, ResNet-50, and InceptionResNet-V2 achieving an average overall accuracies of 88.74%, 92.56%, and 95.32%, respectively. The performance evaluation results through all stages of detection, segmentation, and classification show that the integrated CAD system outperforms the latest conventional deep learning methodologies. We conclude that our CAD system could be used to assist radiologists over all stages of detection, segmentation, and classification for diagnosis of breast lesions.

Automatic hip geometric feature extraction in DXA imaging using regional random forest.

BACKGROUND: Hip fracture is considered one of the salient disability factors across the global population. People with hip fractures are prone to become permanently disabled or die from complications. Although currently the premier determiner, bone mineral density has some notable limitations in terms of hip fracture risk assessment. OBJECTIVES: To learn more about bone strength, hip geometric features (HGFs) can be collected. However, organizing a hip fracture risk study for a large population using a manual HGFs collection technique would be too arduous to be practical. Thus, an automatic HGFs extraction technique is needed. METHOD: This paper presents an automated HGFs extraction technique using regional random forest. Regional random forest localizes landmark points from femur DXA images using local constraints of hip anatomy. The local region constraints make random forest robust to noise and increase its performance because it processes the least number of points and patches. RESULTS: The proposed system achieved an overall accuracy of 96.22% and 95.87% on phantom data and real human scanned data respectively. CONCLUSION: The proposed technique's ability to measure HGFs could be useful in research on the cause and facts of hip fracture and could help in the development of new guidelines for hip fracture risk assessment in the future. The technique will reduce workload and improve the use of X-ray devices.

A fully integrated computer-aided diagnosis system for digital X-ray mammograms via deep learning detection, segmentation, and classification.

A computer-aided diagnosis (CAD) system requires detection, segmentation, and classification in one framework to assist radiologists efficiently in an accurate diagnosis. In this paper, a completely integrated CAD system is proposed to screen digital X-ray mammograms involving detection, segmentation, and classification of breast masses via deep learning methodologies. In this work, to detect breast mass from entire mammograms, You-Only-Look-Once (YOLO), a regional deep learning approach, is used. To segment the mass, full resolution convolutional network (FrCN), a new deep network model, is proposed and utilized. Finally, a deep convolutional neural network (CNN) is used to recognize the mass and classify it as either benign or malignant. To evaluate the proposed integrated CAD system in terms of the accuracies of detection, segmentation, and classification, the publicly available and annotated INbreast database was utilized. The evaluation results of the proposed CAD system via four-fold cross-validation tests show that a mass detection accuracy of 98.96%, Matthews correlation coefficient (MCC) of 97.62%, and F1-score of 99.24% are achieved with the INbreast dataset. Moreover, the mass segmentation results via FrCN produced an overall accuracy of 92.97%, MCC of 85.93%, and Dice (F1-score) of 92.69% and Jaccard similarity coefficient metrics of 86.37%, respectively. The detected and segmented masses were classified via CNN and achieved an overall accuracy of 95.64%, AUC of 94.78%, MCC of 89.91%, and F1-score of 96.84%, respectively. Our results demonstrate that the proposed CAD system, through all stages of detection, segmentation, and classification, outperforms the latest conventional deep learning methodologies. Our proposed CAD system could be used to assist radiologists in all stages of detection, segmentation, and classification of breast masses.

Femur segmentation in DXA imaging using a machine learning decision tree.

BACKGROUND: Accurate measurement of bone mineral density (BMD) in dual-energy X-ray absorptiometry (DXA) is essential for proper diagnosis of osteoporosis. Calculation of BMD requires precise bone segmentation and subtraction of soft tissue absorption. Femur segmentation remains a challenge as many existing methods fail to correctly distinguish femur from soft tissue. Reasons for this failure include low contrast and noise in DXA images, bone shape variability, and inconsistent X-ray beam penetration and attenuation, which cause shadowing effects and person-to-person variation. OBJECTIVE: To present a new method namely, a Pixel Label Decision Tree (PLDT), and test whether it can achieve higher accurate performance in femur segmentation in DXA imaging. METHODS: PLDT involves mainly feature extraction and selection. Unlike photographic images, X-ray images include features on the surface and inside an object. In order to reveal hidden patterns in DXA images, PLDT generates seven new feature maps from existing high energy (HE) and low energy (LE) X-ray features and determines the best feature set for the model. The performance of PLDT in femur segmentation is compared with that of three widely used medical image segmentation algorithms, the Global Threshold (GT), Region Growing Threshold (RGT), and artificial neural networks (ANN). RESULTS: PLDT achieved a higher accuracy of femur segmentation in DXA imaging (91.4%) than either GT (68.4%), RGT (76%) or ANN (84.4%). CONCLUSIONS: The study demonstrated that PLDT outperformed other conventional segmentation techniques in segmenting DXA images. Improved segmentation should help accurate computation of BMD which later improves clinical diagnosis of osteoporosis.

Skin lesion segmentation in dermoscopy images via deep full resolution convolutional networks.

BACKGROUND AND OBJECTIVE: Automatic segmentation of skin lesions in dermoscopy images is still a challenging task due to the large shape variations and indistinct boundaries of the lesions. Accurate segmentation of skin lesions is a key prerequisite step for any computer-aided diagnostic system to recognize skin melanoma. METHODS: In this paper, we propose a novel segmentation methodology via full resolution convolutional networks (FrCN). The proposed FrCN method directly learns the full resolution features of each individual pixel of the input data without the need for pre- or post-processing operations such as artifact removal, low contrast adjustment, or further enhancement of the segmented skin lesion boundaries. We evaluated the proposed method using two publicly available databases, the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Challenge and PH2 datasets. To evaluate the proposed method, we compared the segmentation performance with the latest deep learning segmentation approaches such as the fully convolutional network (FCN), U-Net, and SegNet. RESULTS: Our results showed that the proposed FrCN method segmented the skin lesions with an average Jaccard index of 77.11% and an overall segmentation accuracy of 94.03% for the ISBI 2017 test dataset and 84.79% and 95.08%, respectively, for the PH2 dataset. In comparison to FCN, U-Net, and SegNet, the proposed FrCN outperformed them by 4.94%, 15.47%, and 7.48% for the Jaccard index and 1.31%, 3.89%, and 2.27% for the segmentation accuracy, respectively. Furthermore, the proposed FrCN achieved a segmentation accuracy of 95.62% for some representative clinical benign cases, 90.78% for the melanoma cases, and 91.29% for the seborrheic keratosis cases in the ISBI 2017 test dataset, exhibiting better performance than those of FCN, U-Net, and SegNet. CONCLUSIONS: We conclude that using the full spatial resolutions of the input image could enable to learn better specific and prominent features, leading to an improvement in the segmentation performance.

Treatment of infra-isthmal femoral fracture with an intramedullary nail: Is retrograde nailing a better option than antegrade nailing?

INTRODUCTION: Antegrade intramedullary (IM) nailing is ideal for femoral shaft fractures, but fixing the fracture distal to the isthmal level may be difficult because of medullary canal widening and the proximity of fracture location from the distal femoral joint line. This study aimed to compare treatment results between antegrade and retrograde nailing for infra-isthmal femoral shaft fracture, and to identify influencing factors of nonunion and malalignment. MATERIALS AND METHODS: Sixty patients with infra-isthmal femoral shaft fractures treated with IM nailing and followed-up for > 1 year were enrolled in this retrospective study, 38 in the antegrade nailing group, and 22 in the retrograde nailing group. The two groups had no significant differences in age, sex, and fracture location (p = 0.297, Mann-Whitney test). Radiological evaluation was performed, and functional result was assessed using the Knee Society scoring system. Complications were analyzed in accordance with fracture location, fracture type, and operative method. RESULTS: According to the AO/OTA classification, 35, 16, and 9 cases were type A (A1: 1, A2: 11, A3: 23), B (B1: 2, B2: 7, B3: 7), and C fractures (C2: 4, C3: 5), respectively. The mean follow-up duration was 29.5 months. In the antegrade and retrograde nailing groups, the primary bony union rates were 73.7% in 20.7 weeks (range 12-41) and 86.4% in 17.4 weeks (range 12-30), respectively. The two groups showed no significant differences in union rate (p = 0.251, Pearson's Chi-square test) and union time (p = 0.897, Mann-Whitney test). No cases of malalignment of > 10° in any plane were found in both groups. The mean Knee Society scores were 92 (range 84-100) and 91 (range 83-95) in the antegrade and retrograde nailing groups, respectively, showing no significant difference (p = 0.297, Pearson's Chi-square test). Although fracture location was not significantly related to union rate (p = 0.584, Mann-Whitney test), patients with an effective working length of the distal segment of < 0.75 were prone to nonunion (p = 0.003, Pearson's Chi-square test). CONCLUSIONS: Although no significant difference was found in IM nail type, the IM nail with a shorter working length distal to the fracture showed a strong relationship with nonunion.

Denoising images of dual energy X-ray absorptiometry using non-local means filters.

BACKGROUND: In general, the image quality of high and low energy images of dual energy X-ray absorptiometry (DXA) suffers from noise due to the use of a small amount of X-rays. Denoising of DXA images could be a key process to improve a bone mineral density map, which is derived from a pair of high and low energy images. This could further improve the accuracy of diagnosis of bone fractures and osteoporosis. OBJECTIVE: This study aims to develop and test a new technology to improve the quality, remove the noise, and preserve the edges and fine details of real DXA images. METHODS: In this study, a denoising technique for high and low energy DXA images using a non-local mean filter (NLM) was presented. The source and detector noises of a DXA system were modeled for both high and low DXA images. Then, the optimized parameters of the NLM filter were derived utilizing the experimental data from CIRS-BFP phantoms. After that, the optimized NLM was tested and verified using the DXA images of the phantoms and real human spine and femur. RESULTS: Quantitative evaluation of the results showed average 24.22% and 34.43% improvement of the signal-to-noise ratio for real high and low spine images, respectively, while the improvements were about 15.26% and 13.55% for the high and low images of the femur. The qualitative visual observations of both phantom and real structures also showed significantly improved quality and reduced noise while preserving the edges in both high and low energy images. Our results demonstrate that the proposed NLM outperforms the conventional method using an anisotropic diffusion filter (ADF) and median techniques for all phantom and real human DXA images. CONCLUSIONS: Our work suggests that denoising via NLM could be a key preprocessing method for clinical DXA imaging.