Children's bone age development is delayed with increasing altitude: a multicenter study.

OBJECTIVES: To investigate the bone age (BA) characteristics of children living at high-altitude regions and determine the impact of altitude on the development of BA. METHODS: From June 2014 to July 2022, 1,318 children with left hand-wrist radiographs were retrospectively enrolled from three different geographical altitudes (Beijing 43.5 m above sea level [asl], Lhasa 3650 m asl, and Nagqu 4500 m asl). The predicted age difference (PAD), defined as the difference between BA and chronologic age (CA), was considered the indicator for delayed or advanced growth. The PAD of children from the three regions in total and according to different age groups, genders, and ethnicities were compared. The linear regression model was used to assess the effect of altitude on PAD. RESULTS: A total of 1284 children (CA: 12.00 [6.45, 15.72] years; male: 837/1284, 65.2%) were included in the study with 407 from Beijing, 491 from Lhasa, and 386 from Nagqu. The PAD for Beijing, Lhasa, and Nagqu were 0.1 [-0.30, 0.65], -0.40 [-1.20, 0.27], and -1.42 [-2.32, -0.51] years, respectively. A linear regression analysis showed that altitude significantly contributed to PAD (compared to Beijing, Lhasa coefficient = -0.57, P < 0.001; Nagqu coefficient = -1.55, P < 0.001). CONCLUSIONS: High altitude might be an independent contributor to the delayed BA development of children. ADVANCES IN KNOWLEDGE: The impact of altitude on BA development was revealed for the first time, highlighting the necessity of considering the altitude of the area when evaluating BA development for children residing in high-altitude regions.

Opportunistic use of chest low-dose computed tomography (LDCT) imaging for low bone mineral density and osteoporosis screening: cutoff thresholds for the attenuation values of the lower thoracic and upper lumbar vertebrae.

Background: Osteoporosis remains substantially underdiagnosed and undertreated worldwide. Chest low-dose computed tomography (LDCT) may provide a valuable and popular opportunity for osteoporosis screening. This study sought to evaluate the feasibility of the screening of low bone mineral density (BMD) and osteoporosis with mean attenuation values of the lower thoracic compared to upper lumbar vertebrae. The cutoff thresholds of the mean attenuation values in Hounsfield units (HU) were derived to facilitate implementation of opportunistic screening using chest LDCT. Methods: The participants aged 30 years or older who underwent chest LDCT and quantitative computed tomography (QCT) examinations from August 2018 to October 2020 in our hospital were consecutively included in this retrospective study. A region of interest (ROI) was placed in the trabecular bone of each vertebral body to measure the HU values. The correlations of mean HU values of lower thoracic (T11-T12) and upper lumbar (L1-L2) vertebrae with age and lumbar BMD obtained with QCT were performed using the Pearson correlation coefficient, respectively. The area under the curve (AUC) of the receiver operator characteristic (ROC) curve was generated to determine the cutoff thresholds for distinguishing low BMD from normal and osteoporosis from non-osteoporosis. Results: A total of 1,112 participants were included in the final study cohort (743 men and 369 women, mean age 58.2±8.9 years; range, 32-88 years). The mean HU values of T11-T12 and L1-L2 were significantly different among 3 QCT-defined BMD categories of osteoporosis, osteopenia, and normal (P<0.001). The differences in HU values between T11-T12 and L1-L2 in each category of bone status were statistically significant (P<0.001). The mean HU values of T11-T12 (r=-0.453, P<0.001) and L1-L2 (r=-0.498, P<0.001) had negative correlations with age. Positive correlations were observed between the mean HU values of T11-T12 (r=0.872, P<0.001) and L1-L2 (r=0.899, P<0.001) with BMD. The optimal cutoff thresholds for distinguishing low BMD from normal were average T11-T12 ≤157 HU [AUC =0.941, 95% confidence interval (CI): 0.925-0.954, P<0.001] and L1-L2 ≤138 HU (AUC =0.950, 95% CI: 0.935-0.962, P<0.001), as well as distinguishing osteoporosis from non-osteoporosis were average T11-T12 ≤125 HU (AUC =0.960, 95% CI: 0.947-0.971, P<0.001) and L1-L2 ≤107 HU (AUC =0.961, 95% CI: 0.948-0.972, P<0.001). There was no significant difference between the AUC values of T11-T12 and L1-L2 for low BMD (P=0.07) and osteoporosis (P=0.92) screening. Conclusions: We have conducted a study on low BMD and osteoporosis screening using mean attenuation values of lower thoracic and upper lumbar vertebrae. Assessment of mean attenuation values of T11-T12 and L1-L2 can be used interchangeably for low BMD and osteoporosis screening using chest LDCT, and their cutoff thresholds were established.

Automated detection of small bowel lesions based on capsule endoscopy using deep learning algorithm.

BACKGROUND: In order to overcome the challenges of lesion detection in capsule endoscopy (CE), we improved the YOLOv5-based deep learning algorithm and established the CE-YOLOv5 algorithm to identify small bowel lesions captured by CE. METHODS: A total of 124,678 typical abnormal images from 1,452 patients were enrolled to train the CE-YOLOv5 model. Then 298 patients with suspected small bowel lesions detected by CE were prospectively enrolled in the testing phase of the study. Small bowel images and videos from the above 298 patients were interpreted by the experts, non-experts and CE-YOLOv5, respectively. RESULTS: The sensitivity of CE-YOLOv5 in diagnosing vascular lesions, ulcerated/erosive lesions, protruding lesions, parasite, diverticulum, active bleeding and villous lesions based on CE videos was 91.9 %, 92.2 %, 91.4 %, 93.1 %, 93.3 %, 95.1 %, and 100 % respectively. Furthermore, CE-YOLOv5 achieved specificity and accuracy of more than 90 % for all lesions. Compared with experts, the CE-YOLOv5 showed comparable overall sensitivity, specificity and accuracy (all P > 0.05). Compared with non-experts, the CE-YOLOv5 showed significantly higher overall sensitivity (P < 0.0001) and overall accuracy (P < 0.0001), and a moderately higher overall specificity (P = 0.0351). Furthermore, the time for AI-reading (5.62 ± 2.81 min) was significantly shorter than that for the other two groups (both P < 0.0001). CONCLUSIONS: CE-YOLOv5 diagnosed small bowel lesions in CE videos with high sensitivity, specificity and accuracy, providing a reliable approach for automated lesion detection in real-world clinical practice.

An artificial intelligence-based bone age assessment model for Han and Tibetan children.

Background: Manual bone age assessment (BAA) is associated with longer interpretation time and higher cost and variability, thus posing challenges in areas with restricted medical facilities, such as the high-altitude Tibetan Plateau. The application of artificial intelligence (AI) for automating BAA could facilitate resolving this issue. This study aimed to develop an AI-based BAA model for Han and Tibetan children. Methods: A model named "EVG-BANet" was trained using three datasets, including the Radiology Society of North America (RSNA) dataset (training set n = 12611, validation set n = 1425, and test set n = 200), the Radiological Hand Pose Estimation (RHPE) dataset (training set n = 5491, validation set n = 713, and test set n = 79), and a self-established local dataset [training set n = 825 and test set n = 351 (Han n = 216 and Tibetan n = 135)]. An open-access state-of-the-art model BoNet was used for comparison. The accuracy and generalizability of the two models were evaluated using the abovementioned three test sets and an external test set (n = 256, all were Tibetan). Mean absolute difference (MAD) and accuracy within 1 year were used as indicators. Bias was evaluated by comparing the MAD between the demographic groups. Results: EVG-BANet outperformed BoNet in the MAD on the RHPE test set (0.52 vs. 0.63 years, p < 0.001), the local test set (0.47 vs. 0.62 years, p < 0.001), and the external test set (0.53 vs. 0.66 years, p < 0.001) and exhibited a comparable MAD on the RSNA test set (0.34 vs. 0.35 years, p = 0.934). EVG-BANet achieved accuracy within 1 year of 97.7% on the local test set (BoNet 90%, p < 0.001) and 89.5% on the external test set (BoNet 85.5%, p = 0.066). EVG-BANet showed no bias in the local test set but exhibited a bias related to chronological age in the external test set. Conclusion: EVG-BANet can accurately predict the bone age (BA) for both Han children and Tibetan children living in the Tibetan Plateau with limited healthcare facilities.

Automated vertebral bone mineral density measurement with phantomless internal calibration in chest LDCT scans using deep learning.

OBJECTIVE: To develop and evaluate a fully automated method based on deep learning and phantomless internal calibration for bone mineral density (BMD) measurement and opportunistic low BMD (osteopenia and osteoporosis) screening using chest low-dose CT (LDCT) scans. METHODS: A total of 1175 individuals were enrolled in this study, who underwent both chest LDCT and BMD examinations with quantitative computed tomography (QCT), by two different CT scanners (Siemens and GE). Two convolutional neural network (CNN) models were employed for vertebral body segmentation and labeling, respectively. A histogram technique was applied for vertebral BMD calculation using paraspinal muscle and surrounding fat as references. 195 cases (by Siemens scanner) as fitting cohort were used to build the calibration function. 698 cases as validation cohort I (VCI, by Siemens scanner) and 282 cases as validation cohort II (VCII, by GE scanner) were performed to evaluate the performance of the proposed method, with QCT as the standard for analysis. RESULTS: The average BMDs from the proposed method were strongly correlated with QCT (in VCI: r = 0.896, in VCII: r = 0.956, p < 0.001). Bland-Altman analysis showed a small mean difference of 1.1 mg/cm3, and large interindividual differences as seen by wide 95% limits of agreement (-29.9 to +32.0 mg/cm3) in VCI. The proposed method measured BMDs were higher than QCT measured BMDs in VCII (mean difference = 15.3 mg/cm3, p < 0.001). Osteoporosis and low BMD were diagnosed by proposed method with AUCs of 0.876 and 0.903 in VCI, 0.731 and 0.794 in VCII, respectively. The AUCs of the proposed method were increased to over 0.920 in both VCI and VCII after adjusting the cut-off. CONCLUSION: Without manual selection of the region of interest of body tissues, the proposed method based on deep learning and phantomless internal calibration has the potential for preliminary screening of patients with low BMD using chest LDCT scans. However, the agreement between the proposed method and QCT is insufficient to allow them to be used interchangeably in BMD measurement. ADVANCES IN KNOWLEDGE: This study proposed an automated vertebral BMD measurement method based on deep learning and phantomless internal calibration with paraspinal muscle and fat as reference.

Assessment of Helicobacter pylori infection by deep learning based on endoscopic videos in real time.

BACKGROUND AND AIMS: Endoscopic assessment of Helicobacter pylori infection is a simple and effective method. Here, we aimed to develop a deep learning-based system named Intelligent Detection Endoscopic Assistant-Helicobacter pylori (IDEA-HP) to assess H. pylori infection by using endoscopic videos in real time. METHODS: Endoscopic data were retrospectively obtained from Zhejiang Cancer Hospital (ZJCH) for the development, validation, and testing of the system. Stored videos from ZJCH were used for assessing and comparing the performance of IDEA-HP with that of endoscopists. Prospective consecutive patients undergoing esophagogastroduodenoscopy were enrolled to assess the applicability of clinical practice. The urea breath test was used as the gold standard for diagnosing H. pylori infection. RESULTS: In 100 videos, IDEA-HP achieved a similar overall accuracy of assessing H. pylori infection to that of experts (84.0% vs. 83.6% [P = 0.729]). Nevertheless, the diagnostic accuracy (84.0% vs. 74.0% [P<0.001]) and sensitivity (82.0% vs. 67.2% [P<0.001]) of IDEA-HP were significantly higher than those of the beginners. In 191 prospective consecutive patients, IDEA-HP achieved accuracy, sensitivity, and specificity of 85.3% (95% CI: 79.0%-89.3%), 83.3% (95% CI: 72.8%-90.5%), and 85.8% (95% CI: 77.7%-91.4%), respectively. CONCLUSIONS: Our results show that IDEA-HP has great potential for assisting endoscopists in assessing H. pylori infection status during actual clinical work.

A mucosal recovery software tool for endoscopic submucosal dissection in early gastric cancer.

Background: Due to the limited diagnostic ability, the low detection rate of early gastric cancer (EGC) is a serious health threat. The establishment of the mapping between endoscopic images and pathological images can rapidly improve the diagnostic ability to detect EGC. To expedite the learning process of EGC diagnosis, a mucosal recovery map for the mapping between ESD mucosa specimen and pathological images should be performed in collaboration with endoscopists and pathologists, which is a time-consuming and laborious work. Methods: 20 patients at the Zhejiang Provincial People's Hospital, Affiliated People's Hospital of Hangzhou Medical College from March 2020 to July 2020 were enrolled in this study. We proposed the improved U-Net to obtain WSI-level segmentation results, and the WSI-level results can be mapped to the macroscopic image of the specimen. For the convenient use, a software pipeline named as "Pathology Helper" for integration the workflow of the construction of mucosal recovery maps was developed. Results: The MIoU and Dice of our model can achieve 0.955 ± 0.0936 and 0.961 ± 0.0874 for WSI-level segmentation, respectively. With the help of "Pathology Helper", we can construct the high-quality mucosal recovery maps to reduce the workload of endoscopists and pathologists. Conclusion: "Pathology Helper" will accelerate the learning of endoscopists and pathologists, and rapidly improve their abilities to detect EGC. Our work can also improve the detection rate of early gastric cancer, so that more patients with gastric cancer will be treated in a timely manner.

Study on Molecular Information Intelligent Diagnosis and Treatment of Bladder Cancer on Pathological Tissue Image.

Background: Molecular information about bladder cancer is significant for treatment and prognosis. The immunohistochemistry (IHC) method is widely used to analyze the specific biomarkers to determine molecular subtypes. However, procedures in IHC and plenty of reagents are time and labor-consuming and expensive. This study established a computer-aid diagnosis system for predicting molecular subtypes, p53 status, and programmed death-ligand 1 (PD-L1) status of bladder cancer with pathological images. Materials and Methods: We collected 119 muscle-invasive bladder cancer (MIBC) patients who underwent radical cystectomy from January 2016 to September 2018. All the pathological sections are scanned into digital whole slide images (WSIs), and the IHC results of adjacent sections were recorded as the label of the corresponding slide. The tumor areas are first segmented, then molecular subtypes, p53 status, and PD-L1 status of those tumor-positive areas would be identified by three independent convolutional neural networks (CNNs). We measured the performance of this system for predicting molecular subtypes, p53 status, and PD-L1 status of bladder cancer with accuracy, sensitivity, and specificity. Results: For the recognition of molecular subtypes, the accuracy is 0.94, the sensitivity is 1.00, and the specificity is 0.909. For PD-L1 status recognition, the accuracy is 0.897, the sensitivity is 0.875, and the specificity is 0.913. For p53 status recognition, the accuracy is 0.846, the sensitivity is 0.857, and the specificity is 0.750. Conclusion: Our computer-aided diagnosis system can provide a novel and simple assistant tool to obtain the molecular subtype, PD-L1 status, and p53 status. It can reduce the workload of pathologists and the medical cost.

"Pink Pattern" Visualized in Magnifying Endoscopy With Narrow-Band Imaging Is a Novel Feature of Early Differentiated Gastric Cancer: A Bridge Between Endoscopic Images and Histopathological Changes.

Background: A pink color change occasionally found by us under magnifying endoscopy with narrow-band imaging (ME-NBI) may be a special feature of early gastric cancer (EGC), and was designated the "pink pattern". The purposes of this study were to determine the relationship between the pink pattern and the cytopathological changes in gastric cancer cells and whether the pink pattern is useful for the diagnosis of EGC. Methods: The color features of ME-NBI images and pathological images of cancerous gastric mucosal surfaces were extracted and quantified. The cosine similarity was calculated to evaluate the correlation between the pink pattern and the nucleus-to-cytoplasm ratio of cancerous epithelial cells. Two diagnostic tests were performed by 12 endoscopists using stored ME-NBI images of 185 gastric lesions to investigate the diagnostic efficacy of the pink pattern for EGC. The diagnostic values, such as the area under the curve (AUC), the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), of test 1 and test 2 were compared. Results: The cosine similarity between the color values of ME-NBI images and pathological images of 20 lesions was at least 0.744. The median AUC, accuracy, sensitivity, specificity, PPV, and NPV of test 2 were significantly better than those of test 1 for all endoscopists and for the junior and experienced groups. Conclusions: The pink pattern observed in ME-NBI images correlated strongly with the change in the nucleus-to-cytoplasm ratio of gastric epithelial cells, and could be considered a useful marker for the diagnosis of differentiated EGC.

Rib fracture detection system based on deep learning.

Rib fracture detection is time-consuming and demanding work for radiologists. This study aimed to introduce a novel rib fracture detection system based on deep learning which can help radiologists to diagnose rib fractures in chest computer tomography (CT) images conveniently and accurately. A total of 1707 patients were included in this study from a single center. We developed a novel rib fracture detection system on chest CT using a three-step algorithm. According to the examination time, 1507, 100 and 100 patients were allocated to the training set, the validation set and the testing set, respectively. Free Response ROC analysis was performed to evaluate the sensitivity and false positivity of the deep learning algorithm. Precision, recall, F1-score, negative predictive value (NPV) and detection and diagnosis were selected as evaluation metrics to compare the diagnostic efficiency of this system with radiologists. The radiologist-only study was used as a benchmark and the radiologist-model collaboration study was evaluated to assess the model's clinical applicability. A total of 50,170,399 blocks (fracture blocks, 91,574; normal blocks, 50,078,825) were labelled for training. The F1-score of the Rib Fracture Detection System was 0.890 and the precision, recall and NPV values were 0.869, 0.913 and 0.969, respectively. By interacting with this detection system, the F1-score of the junior and the experienced radiologists had improved from 0.796 to 0.925 and 0.889 to 0.970, respectively; the recall scores had increased from 0.693 to 0.920 and 0.853 to 0.972, respectively. On average, the diagnosis time of radiologist assisted with this detection system was reduced by 65.3 s. The constructed Rib Fracture Detection System has a comparable performance with the experienced radiologist and is readily available to automatically detect rib fracture in the clinical setting with high efficacy, which could reduce diagnosis time and radiologists' workload in the clinical practice.

Intelligent detection endoscopic assistant: An artificial intelligence-based system for monitoring blind spots during esophagogastroduodenoscopy in real-time.

BACKGROUND: Observation of the entire stomach during esophagogastroduodenoscopy (EGD) is important; however, there is a lack of effective evaluation tools. AIMS: To develop an artificial intelligence (AI)-assisted EGD system able to automatically monitor blind spots in real-time. METHODS: An AI-based system, called the Intelligent Detection Endoscopic Assistant (IDEA), was developed using a deep convolutional neural network (DCNN) and long short-term memory (LSTM). The performance of IDEA for recognition of gastric sites in images and videos was evaluated. Primary outcomes included diagnostic accuracy, sensitivity, and specificity. RESULTS: A total of 170,297 images and 5779 endoscopic videos were collected to develop the system. As the test group, 3100 EGD images were acquired to evaluate the performance of DCNN in recognition of gastric sites in images. The sensitivity, specificity, and accuracy of DCNN were determined as 97.18%,99.91%, and 99.83%, respectively. To assess the performance of IDEA in recognition of gastric sites in EGD videos, 129 videos were used as the test group. The sensitivity, specificity, and accuracy of IDEA were 96.29%,93.32%, and 95.30%, respectively. CONCLUSIONS: IDEA achieved high accuracy for recognition of gastric sites in real-time. The system can be applied as a powerful assistant tool for monitoring blind spots during EGD.

A Hybrid Network for Large-Scale Action Recognition from RGB and Depth Modalities.

The paper presents a novel hybrid network for large-scale action recognition from multiple modalities. The network is built upon the proposed weighted dynamic images. It effectively leverages the strengths of the emerging Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) based approaches to specifically address the challenges that occur in large-scale action recognition and are not fully dealt with by the state-of-the-art methods. Specifically, the proposed hybrid network consists of a CNN based component and an RNN based component. Features extracted by the two components are fused through canonical correlation analysis and then fed to a linear Support Vector Machine (SVM) for classification. The proposed network achieved state-of-the-art results on the ChaLearn LAP IsoGD, NTU RGB+D and Multi-modal & Multi-view & Interactive ( M 2 I ) datasets and outperformed existing methods by a large margin (over 10 percentage points in some cases).

A novel multivariate performance optimization method based on sparse coding and hyper-predictor learning.

In this paper, we investigate the problem of optimization of multivariate performance measures, and propose a novel algorithm for it. Different from traditional machine learning methods which optimize simple loss functions to learn prediction function, the problem studied in this paper is how to learn effective hyper-predictor for a tuple of data points, so that a complex loss function corresponding to a multivariate performance measure can be minimized. We propose to present the tuple of data points to a tuple of sparse codes via a dictionary, and then apply a linear function to compare a sparse code against a given candidate class label. To learn the dictionary, sparse codes, and parameter of the linear function, we propose a joint optimization problem. In this problem, the both the reconstruction error and sparsity of sparse code, and the upper bound of the complex loss function are minimized. Moreover, the upper bound of the loss function is approximated by the sparse codes and the linear function parameter. To optimize this problem, we develop an iterative algorithm based on descent gradient methods to learn the sparse codes and hyper-predictor parameter alternately. Experiment results on some benchmark data sets show the advantage of the proposed methods over other state-of-the-art algorithms.