GSP-AI: An AI-Powered Platform for Identifying Key Growth Stages and the Vegetative-to-Reproductive Transition in Wheat Using Trilateral Drone Imagery and Meteorological Data.

Wheat (Triticum aestivum) is one of the most important staple crops worldwide. To ensure its global supply, the timing and duration of its growth cycle needs to be closely monitored in the field so that necessary crop management activities can be arranged in a timely manner. Also, breeders and plant researchers need to evaluate growth stages (GSs) for tens of thousands of genotypes at the plot level, at different sites and across multiple seasons. These indicate the importance of providing a reliable and scalable toolkit to address the challenge so that the plot-level assessment of GS can be successfully conducted for different objectives in plant research. Here, we present a multimodal deep learning model called GSP-AI, capable of identifying key GSs and predicting the vegetative-to-reproductive transition (i.e., flowering days) in wheat based on drone-collected canopy images and multiseasonal climatic datasets. In the study, we first established an open Wheat Growth Stage Prediction (WGSP) dataset, consisting of 70,410 annotated images collected from 54 varieties cultivated in China, 109 in the United Kingdom, and 100 in the United States together with key climatic factors. Then, we built an effective learning architecture based on Res2Net and long short-term memory (LSTM) to learn canopy-level vision features and patterns of climatic changes between 2018 and 2021 growing seasons. Utilizing the model, we achieved an overall accuracy of 91.2% in identifying key GS and an average root mean square error (RMSE) of 5.6 d for forecasting the flowering days compared with manual scoring. We further tested and improved the GSP-AI model with high-resolution smartphone images collected in the 2021/2022 season in China, through which the accuracy of the model was enhanced to 93.4% for GS and RMSE reduced to 4.7 d for the flowering prediction. As a result, we believe that our work demonstrates a valuable advance to inform breeders and growers regarding the timing and duration of key plant growth and development phases at the plot level, facilitating them to conduct more effective crop selection and make agronomic decisions under complicated field conditions for wheat improvement.

Association between admission inflammatory indicators and 3-year mortality risk in geriatric patients after hip fracture surgery: a retrospective cohort study.

Background: Recent research indicates that the monocyte lymphocyte ratio (MLR), neutrophil lymphocyte ratio (NLR), platelet lymphocyte ratio (PLR), C-reactive protein (CRP), and systemic immune-inflammation index (SII) may serve as valuable predictors of early postoperative mortality in elderly individuals with hip fractures. The primary objective of the study was to examine the association between preoperative MLR, NLR, PLR, CRP, and SII levels and 3-year mortality risk in geriatric patients after hip fracture surgery. Patients and methods: The study included patients aged 65 years or older who underwent hip fracture surgery between November 2018 and November 2019. Admission levels of MLR, NLR, PLR, CRP, and SII were measured. The median follow-up period was 3.1 years. Cox proportional hazards models were used to calculate the hazard ratio (HR) for mortality with adjusting for potential covariates. Time-dependent receiver operating characteristic (ROC) curves were employed to assess the predictive capability of inflammatory indicators for mortality. Results: A total of 760 patients completed the follow-up (79.4 ± 7.8 years, 71.1% female). A higher preoperative MLR was found to be significantly associated with an increased 3-year postoperative mortality risk (HR 1.811, 95% CI 1.047-3.132, P = 0.034). However, no significant correlations were observed between preoperative NLR, PLR, CRP, SII and 3-year mortality. The areas under the ROC curve (AUCs) of MLR for predicting 30-day, 120-day, 1-year, and 3-year mortality were 0.74 (95% CI 0.53-0.95), 0.70 (95% CI 0.57-0.83), 0.67 (95% CI 0.60-0.74), and 0.61 (95% CI 0.56-0.66), respectively. Conclusion: Preoperative MLR is a useful inflammatory marker for predicting 3-year mortality in elderly hip fracture patients, but its predictive ability diminishes over time.

Preoperative Thyroid-Stimulating Hormone Levels and Three-Year Mortality in Elderly Hip Fracture Patients: Insights from a Prospective Cohort Study.

BACKGROUND It is unclear whether preoperative thyroid-stimulating hormone (TSH) level is correlated with long-term mortality in the elderly after hip fracture surgery. We aimed to assess the association between TSH levels and 3-year mortality in these patients. MATERIAL AND METHODS We enrolled patients aged 65 and above who had hip fracture surgery and thyroid function tests upon admission from 2018 to 2019. Patients were categorized based on TSH median value, quartiles, or thyroid function status. The median follow-up time was 3.1 years. Cox proportional hazards models were used to examine the correlation between TSH levels and mortality, adjusting for covariates. RESULTS Out of 799 eligible patients, 92.7% (741/799) completed the follow-up, with 20.6% (153/741) of those having died by the end of the follow-up. No statistically significant differences in mortality risks were found when stratified by TSH median value (HR 0.88, 95% CI 0.64-1.22, P=0.448) or quartiles (HR ranging from 0.90 to 1.13, P>0.05). Similarly, when categorized based on admission thyroid function status, patients who presented with hypothyroidism, subclinical hypothyroidism, hyperthyroidism, and subclinical hyperthyroidism upon admission did not demonstrate a statistically significant difference in mortality risk compared to those who were considered euthyroid (HR 1.34, 95% CI 0.72-2.49, P=0.359; HR 0.77, 95% CI 0.38-1.60, P=0.489; HR 1.15, 95% CI 0.16-8.30, P=0.890; HR 1.07, 95% CI 0.34-3.38, P=0.913, respectively). CONCLUSIONS Admission TSH is not significantly associated with 3-year mortality in geriatric patients after hip fracture surgery.

Prognostic Nutritional Index (PNI) as an Independent Predictor of 3-Year Postoperative Mortality in Elderly Patients with Hip Fracture: A Post hoc Analysis of a Prospective Cohort Study.

OBJECTIVE: The prognostic nutritional index (PNI) has been reported as a significant predictor in various diseases. However, the prognostic value of the PNI in geriatric hip fracture patients has not been thoroughly evaluated. This study aimed to investigate the association between admission PNI and 3-year mortality in those patients. METHODS: In this post hoc analysis, we included patients aged ≥65 years who underwent surgery for hip fracture between 2018 and 2019. The admission PNI was calculated as serum albumin (g/L) +5 × total lymphocyte count (×109/L). Patients were categorized into four groups based on PNI quartiles (≤ 43.55, 43.55-46.55, 46.55-49.20, and >49.20, respectively). The median follow-up duration was 3.1 years. Cox proportional hazards models were used to calculate the hazard ratio (HR). Receiver operating characteristic curve (ROC) was conducted for using PNI to predict mortality. RESULTS: Of the 942 eligible patients, 190 (20.2%) patients died during the follow-up. Compared to patients in the first quartile (Q1), those in the second (Q2), third (Q3), and fourth (Q4) quartiles had significantly lower mortality risks (HRs 0.50, 95% CI 0.35-0.74; 0.41, 95% CI 0.26-0.64; and 0.26, 95% CI 0.15-0.45, respectively). The optimal cutoff of PNI for predicting mortality was set as 45.275 (sensitivity, 0.674; specificity, 0.692; area under the curve (AUC), 0.727). Patients with higher PNI (>45.275) had a significant lower mortality risk (HR 0.39, 95% CI 0.28-0.55) compared to those with lower PNI (≤ 45.275). CONCLUSION: PNI is a reliable and independent predictor of 3-year mortality after hip fracture surgery in the elderly.

Bagging Improves the Performance of Deep Learning-Based Semantic Segmentation with Limited Labeled Images: A Case Study of Crop Segmentation for High-Throughput Plant Phenotyping.

Advancements in imaging, computer vision, and automation have revolutionized various fields, including field-based high-throughput plant phenotyping (FHTPP). This integration allows for the rapid and accurate measurement of plant traits. Deep Convolutional Neural Networks (DCNNs) have emerged as a powerful tool in FHTPP, particularly in crop segmentation-identifying crops from the background-crucial for trait analysis. However, the effectiveness of DCNNs often hinges on the availability of large, labeled datasets, which poses a challenge due to the high cost of labeling. In this study, a deep learning with bagging approach is introduced to enhance crop segmentation using high-resolution RGB images, tested on the NU-Spidercam dataset from maize plots. The proposed method outperforms traditional machine learning and deep learning models in prediction accuracy and speed. Remarkably, it achieves up to 40% higher Intersection-over-Union (IoU) than the threshold method and 11% over conventional machine learning, with significantly faster prediction times and manageable training duration. Crucially, it demonstrates that even small labeled datasets can yield high accuracy in semantic segmentation. This approach not only proves effective for FHTPP but also suggests potential for broader application in remote sensing, offering a scalable solution to semantic segmentation challenges. This paper is accompanied by publicly available source code.

FNDC5 prevents oxidative stress and neuronal apoptosis after traumatic brain injury through SIRT3-dependent regulation of mitochondrial quality control.

Mitochondrial dysfunction and oxidative stress are important mechanisms for secondary injury after traumatic brain injury (TBI), which result in progressive pathophysiological exacerbation. Although the Fibronectin type III domain-containing 5 (FNDC5) was reported to repress oxidative stress by retaining mitochondrial biogenesis and dynamics, its possible role in the secondary injury after TBI remain obscure. In present study, we observed that the level of serum irisin (the cleavage product of FNDC5) significantly correlated with the neurological outcomes of TBI patients. Knockout of FNDC5 increased the lesion volume and exacerbated apoptosis and neurological deficits after TBI in mice, while FNDC5 overexpression yielded a neuroprotective effect. Moreover, FNDC5 deficiency disrupted mitochondrial dynamics and function. Activation of Sirtuin 3 (SIRT3) alleviated FNDC5 deficiency-induced disruption of mitochondrial dynamics and bioenergetics. In neuron-specific SIRT3 knockout mice, FNDC5 failed to attenuate TBI-induced mitochondrial damage and brain injuries. Mechanically, FNDC5 deficiency led to reduced SIRT3 expression via enhanced ubiquitin degradation of transcription factor Nuclear factor erythroid 2-related factor 2 (NRF2), which contributed to the hyperacetylation and inactivation of key regulatory proteins of mitochondrial dynamics and function, including OPA1 and SOD2. Finally, engineered RVG29-conjugated nanoparticles were generated to selectively and efficiently deliver irisin to the brain of mice, which yielded a satisfactory curative effect against TBI. In conclusion, FNDC5/irisin exerts a protective role against acute brain injury by promoting SIRT3-dependent mitochondrial quality control and thus represents a potential target for neuroprotection after TBI.

Inter-rater variability and repeatability in the assessment of the Tanner-Whitehouse classification of hand radiographs for the estimation of bone age.

OBJECTIVE: To determine which bones and which grades had the highest inter-rater variability when employing the Tanner-Whitehouse (T-W) method. MATERIALS AND METHODS: Twenty-four radiologists were recruited and trained in the T-W classification of skeletal development. The consistency and skill of the radiologists in determining bone development status were assessed using 20 pediatric hand radiographs of children aged 1 to 18 years old. Four radiologists had a poor concordance rate and were excluded. The remaining 20 radiologists undertook a repeat reading of the radiographs, and their results were analyzed by comparing them with the mean assessment of two senior experts as the reference standard. Concordance rate, scoring, and Kendall's W were calculated to evaluate accuracy and consistency. RESULTS: Both the radius, ulna, and short finger (RUS) system (Kendall's W = 0.833) and the carpal (C) system (Kendall's W = 0.944) had excellent consistency, with the RUS system outperforming the C system in terms of scores. The repeatability analysis showed that the second rating test, performed after 2 months of further bone age assessment (BAA) practice, was more consistent and accurate than the first. The capitate had the lowest average concordance rate and scoring, as well as the lowest overall concordance rate for its D classification. Moreover, the G classifications of the seven carpal bones all had a concordance rate less than 0.6. The bones with lower Kendall's W were likewise those with lower scores and concordance rates. CONCLUSION: The D grade of the capitate showed the highest variation, and the use of the Tanner-Whitehouse 3rd edition (T-W3) to determine bone age (BA) was frequently inconsistent. A more comprehensive description with a focus on inaccuracy bones or ratings and a modification to the T-W3 approach would significantly advance BAA.

Image Filtering to Improve Maize Tassel Detection Accuracy Using Machine Learning Algorithms.

Unmanned aerial vehicle (UAV)-based imagery has become widely used to collect time-series agronomic data, which are then incorporated into plant breeding programs to enhance crop improvements. To make efficient analysis possible, in this study, by leveraging an aerial photography dataset for a field trial of 233 different inbred lines from the maize diversity panel, we developed machine learning methods for obtaining automated tassel counts at the plot level. We employed both an object-based counting-by-detection (CBD) approach and a density-based counting-by-regression (CBR) approach. Using an image segmentation method that removes most of the pixels not associated with the plant tassels, the results showed a dramatic improvement in the accuracy of object-based (CBD) detection, with the cross-validation prediction accuracy (r2) peaking at 0.7033 on a detector trained with images with a filter threshold of 90. The CBR approach showed the greatest accuracy when using unfiltered images, with a mean absolute error (MAE) of 7.99. However, when using bootstrapping, images filtered at a threshold of 90 showed a slightly better MAE (8.65) than the unfiltered images (8.90). These methods will allow for accurate estimates of flowering-related traits and help to make breeding decisions for crop improvement.

Sagittal support rather than medial cortical support matters in geriatric intertrochanteric fracture: A finite element analysis study.

Hip fracture, increasing exponentially with age, is osteoporosis's most severe clinical consequence. Intertrochanteric fracture, one of the main types of hip fracture, is associated with higher mortality and morbidity. The current research hotspots lay in improving the treatment effect and optimizing the secondary stability after intertrochanteric fracture surgery. Cortex buttress reduction is a widely accepted method for treating intertrochanteric fracture by allowing the head-neck fragment to slide and rigidly contact the femoral shaft's cortex. Medial cortical support is considered a more effective option in treating young patients. However, osteo-degenerations features, including bone weakness and cortical thickness thinning, affect the performance of cortex support in geriatric intertrochanteric fracture treatment. Literature focusing on the age-specific difference in cortex performance in the fractured hip is scarce. We hypothesized that this osteo-19 degenerative feature affects the performance of cortex support in treating intertrochanteric fractures between the young and the elderly. We established twenty models for the old and the young with intertrochanteric fractures and performed static and dynamic simulations under one-legged stance and walking cycle conditions. The von Mises stress and displacement on the femur, proximal femoral nail anti-rotation (PFNA) implant, fracture plane, and the cutting volume of cancellous bone of the femur were compared. It was observed that defects in the anterior and posterior cortical bone walls significantly increase the stress on the PFNA implant, the displacement of the fracture surface, and cause a greater volume of cancellous bone to be resected. We concluded that ensuring the integrity and alignment of the anterior and posterior cortical bones is essential for elderly patients, and sagittal support is recommended. This finding suggests that the treatment method for intertrochanteric fracture may differ, considering the patient's age difference.

Muscle density, but not size, is independently associated with cognitive health in older adults with hip fractures.

Emerging evidence indicates a complex interplay between skeletal muscle and cognitive function. Despite the known differences between muscle quantity and quality, which can be measured via computed tomography (CT), the precise nature of their associations with cognitive performance remain underexplored. To investigate the links between muscle size and density and cognitive impairment (CI) in the older adults with hip fractures, we conducted a post hoc, cross-sectional analysis within a prospective cohort study on 679 patients with hip fractures over 65. Mini-Mental State Examination (MMSE) and routine hip CT imaging were utilized to assess cognition function and muscle characteristics in older adults with hip fractures. The CT scans provided data on cross-sectional area and attenuation for the gluteus maximus (G.MaxM) and the combined gluteus medius and minimus (G.Med/MinM). Participants were categorized into CI and non-CI groups based on education levels and MMSE scores. Multivariate logistic regressions, propensity score (PS) methods, and subgroup analysis were employed to analyze associations and validate findings. This study included 123 participants (81.6 ± 6.8 years, 74% female) with CI and 556 participants (78.5 ± 7.7 years, 72% female) without. Compared to the non-CI group, muscle parameters, especially density, were significantly lower in the CI group. Specifically, G.Med/Min muscle density, but not size was robustly associated with CI (odds ratio (OR) = 0.77, 95% confidence interval = 0.62-0.96, P = 0.02), independent of other medical situations. Sensitivity analysis corroborated that G.Med/Min muscle density was consistently lower in the CI group than the non-CI group, as evidenced in the PS matched (P = 0.024) and weighted cohort (P = 0.033). Enhanced muscle parameters, particularly muscle density in the G.Med/MinM muscle, correlate with a lower risk of CI. Muscle density demonstrates a stronger association with cognitive performance than muscle size, highlighting its potential as a key focus in future cognitive health research.

Risk prediction of second hip fracture by bone and muscle density of the hip varies with time after first hip fracture: A prospective cohort study.

Purpose: Predictors of 'imminent' risk of second hip fracture are unknown. The aims of the study were to explore strength of hip areal bone mineral density (aBMD), and muscle area and density for predicting second hip fracture at different time intervals. Methods: Data of the Chinese Second Hip Fracture Evaluation were analyzed, a longitudinal study to evaluate the risk of second hip fracture (of the contralateral hip) by using CT images obtained immediately after first hip fracture. Muscle cross-sectional area and density were measured of the gluteus maximus (G.MaxM) and gluteus medius and minimus (G.Med/MinM) and aBMD of the proximal femur at the contralateral unfractured side. Patients were followed up for a median time of 4.5 years. Separate Cox models were used to predict second hip fracture risk at different time intervals after first event adjusted for age, sex, BMI and diabetes. Results: The mean age of subjects with imminent (within 1st or 2nd year) second hip fracture was 79.80 ± 5.16 and 81.56 ± 3.64 years. In the 1st year after the first hip fracture, femoral neck (FN) aBMD predicted second hip fracture (HR 5.88; 95 % CI, 1.32-26.09). In the remaining years of follow-up after 2nd year, muscle density predicted second hip fracture (G.MaxM HR 2.13; 95 % CI, 1.25-3.65,G.Med/MinM HR 2.10; 95 % CI, 1.32-3.34). Conclusions: Our results show that femoral neck aBMD is an important predictor for second hip fracture within the first year and therefore suggest supports the importance concept of early and rapid-acting bone-active drugs to increase hip BMD. In addition, the importance of muscle density predicting second hip fracture after the second year suggest post hip fracture rehabilitation and exercise programs could also be important to reduce muscle fatty infiltration.

Associations of Quantitative and Qualitative Muscle Parameters With Second Hip Fracture Risk in Older Women: A Prospective Cohort Study.

Older women with a first hip fracture exhibit heightened susceptibility and incidence of second fracture and potentially severe consequences. This prospective study was to compare the predictive power of qualitative and quantitative muscle parameters for a second hip fracture in older women with a first hip fracture. A total of 206 subjects were recruited from the longitudinal Chinese Second Hip Fracture Evaluation study. Hip computed tomography (CT) scans were obtained immediately after the first fracture. Muscle fat infiltration was assessed according to the Goutallier classification qualitatively. Quantitative parameters included cross-sectional area and density of gluteus maximus (G.MaxM) and gluteus medius and minimus (G.Med/MinM) muscles. CT X-ray absorptiometry was used to measure the areal bone mineral density (aBMD) of the contralateral femur. Cox proportional hazards models were used to compute hazard ratios (HR) of second hip fracture risk. The mean age of subjects was 74.9 (±9.5) years at baseline. After 4.5 years, 35 had a second hip fracture, 153 without a second hip fracture, and 18 died. Except for the combined G.MinM Goutallier grade 3 and 4 groups before adjustment for covariates (HR = 5.83; 95% confidence interval [CI] 1.49-22.83), there were no significant HRs for qualitative classification to predict a second hip fracture. Among quantitative metrics, after adjustment for covariates, G.Med/MinM density was significant in the original (HR = 1.44; CI 1.02-2.04) and competing risk analyses (HR = 1.46; CI 1.02-2.07). After additional adjustment for femoral neck (FN) aBMD, G.Med/MinM density remained borderline significant for predicting a second hip fracture in competing risk analysis (HR = 1.43; CI 0.99-2.06; p = 0.057). Our study revealed that Goutallier classification was less effective than quantitative muscle metrics for predicting hip second fracture in this elderly female cohort. After adjustment for FN aBMD, G.Med/MinM density is a borderline independent predictor of second hip fracture risk. © 2023 The Authors. JBMR Plus published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research.

Rapid online plant leaf area change detection with high-throughput plant image data.

High-throughput plant phenotyping (HTPP) has become an emerging technique to study plant traits due to its fast, labor-saving, accurate and non-destructive nature. It has wide applications in plant breeding and crop management. However, the resulting massive image data has raised a challenge associated with efficient plant traits prediction and anomaly detection. In this paper, we propose a two-step image-based online detection framework for monitoring and quick change detection of the individual plant leaf area via real-time imaging data. Our proposed method is able to achieve a smaller detection delay compared with some baseline methods under some predefined false alarm rate constraint. Moreover, it does not need to store all past image information and can be implemented in real time. The efficiency of the proposed framework is validated by a real data analysis.

Orthogeriatric co-managements lower early mortality in long-lived elderly hip fracture: a post-hoc analysis of a prospective study.

OBJECTIVE: To evaluate the clinical effectiveness of orthogeriatric co-management care in long-lived elderly hip fracture patients (age ≥ 90). METHODS: Secondary analysis was conducted in long-lived hip fracture patients between 2018 to 2019 in 6 hospitals in Beijing, China. Patients were divided into the orthogeriatric co-management group (CM group) and traditional consultation mode group (TC group) depending on the management mode. With 30-day mortality as the primary outcome, multivariate regression analyses were performed after adjusting for potential covariates. 30-day mobility and quality of life were compared between groups. RESULTS: A total of 233 patients were included, 223 of whom completed follow-up (125 in CM group, 98 in TC group). The average age was 92.4 ± 2.5 years old (range 90-102). The 30-day mortality in CM group was significantly lower than that in TC group after adjustments for (2.4% vs. 10.2%; OR = 0.231; 95% CI 0.059 ~ 0.896; P = 0.034). The proportion of patients undergoing surgery and surgery performed within 48 h also favored the CM group (97.6% vs. 85.7%, P = 0.002; 74.4% vs. 24.5%, P < 0.001; respectively). In addition, much more patients in CM group could walk with or without aids in postoperative 30 days than in the TC group (87.7% vs. 60.2%, P < 0.05), although differences were not found after 1-year follow-up. And there was no significant difference in total cost between the two groups (P > 0.05). CONCLUSIONS: For long-lived elderly hip fracture patients, orthogeriatric co-management care lowered early mortality, improved early mobility and compared with the traditional consultation mode.

Association Between Hip Bone Mineral Density and Mortality Risk After Hip Fracture: A Prospective Cohort Study.

Factors related to mortality after osteoporotic hip fracture (HF) have been investigated intensively, except for proximal femoral bone mineral density (BMD), which is also the primary cause of osteoporosis. In this study, we aimed to investigate the association of hip BMD with mortality risk after HF. Four hundred and eleven elderly patients with HF in Beijing, China, were included and prospectively followed up with a median time of 3 years. At baseline, quantitative CT technique (QCT) was used to measure areal BMD (aBMD) of the unaffected hip. Areal BMDs of the total hip (TH), femoral neck (FN), trochanter (TR), and intertrochanter were analyzed with postoperative mortality as the primary outcome. A total of 394 patients (78.59 ± 7.59 years, 75.4% female) were included in our final analysis, with 86 (82.23 ± 7.00 years, 81.4% female) dead. All hip bone densities demonstrated a significant association with mortality risks in the unadjusted model, but only TR aBMD remained significantly correlated after adjusting for all covariates. Compared to the lower TR aBMD group, the higher TR aBMD group yielded significantly lower death risks (HR 0.21 95% CI 0.05-0.9, P = 0.036). Higher survival probabilities were observed for higher TH and TR aBMD in survival analysis (P < 0.001). Hip BMD, especially TR BMD assessed by QCT, is an independent risk factor for postoperative mortality following HF. QCT may present a promising avenue for opportunistic analysis in immobilized patients, providing valuable information for early detection and personalized interventions to enhance patient outcomes.

Muscle size and density are independently associated with death after hip fracture: A prospective cohort study.

BACKGROUND: Mortality following hip fracture is high and incompletely understood. We hypothesize that hip musculature size and quality are related to mortality following hip fracture. This study aims to investigate the associations of hip muscle area and density from hip CT with death following hip fracture as well as assess the dependence of this association on time after hip fracture. METHODS: In this secondary analysis of the prospectively collected CT images and data from the Chinese Second Hip Fracture Evaluation, 459 patients were enrolled between May 2015 and June 2016 and followed up for a median of 4.5 years. Muscle cross-sectional area and density were measured of the gluteus maximus (G.MaxM) and gluteus medius and minimus (G.Med/MinM) and aBMD of the proximal femur. The Goutallier classification (GC) was used for qualitatively assessing muscle fat infiltration. Separate Cox models were used to predict mortality risk adjusted for covariates. RESULTS: At the end of the follow-up, 85 patients were lost, 81 patients (64% women) had died, and 293 (71% women) survived. The mean age of non-surviving patients at death (82.0 ± 8.1 years) was higher than that of the surviving patients (74.4 ± 9.9 years). The Parker Mobility Score and the American Society of Anesthesiologists scores of the patients that died were respectively lower and higher compared to the surviving patients. Hip fracture patients received different surgical procedures, and no significant difference in the percentage of hip arthroplasty was observed between the dead and the surviving patients (P = 0.11). The cumulative survival was significantly lower for patients with low G.MaxM area and density and low G.Med/MinM density, independent of age and clinical risk scores. The GC grades were not associated with the mortality after hip fracture. Muscle density of both G.MaxM (adj. HR 1.83; 95% CI, 1.06-3.17) and G.Med/MinM (adj. HR 1.98; 95% CI, 1.14-3.46) was associated with mortality in the 1st year after hip fracture. G.MaxM area (adj. HR 2.11; 95% CI, 1.08-4.14) was associated with mortality in the 2nd and later years after hip fracture. CONCLUSION: Our results for the first time show that hip muscle size and density are associated with mortality in older hip fracture patients, independent of age and clinical risk scores. This is an important finding to better understand the factors contributing to the high mortality in older hip fracture patients and to develop better future risk prediction scores that include muscle parameters.

Self-Supervised Plant Phenotyping by Combining Domain Adaptation with 3D Plant Model Simulations: Application to Wheat Leaf Counting at Seedling Stage.

The number of leaves at a given time is important to characterize plant growth and development. In this work, we developed a high-throughput method to count the number of leaves by detecting leaf tips in RGB images. The digital plant phenotyping platform was used to simulate a large and diverse dataset of RGB images and corresponding leaf tip labels of wheat plants at seedling stages (150,000 images with over 2 million labels). The realism of the images was then improved using domain adaptation methods before training deep learning models. The results demonstrate the efficiency of the proposed method evaluated on a diverse test dataset, collecting measurements from 5 countries obtained under different environments, growth stages, and lighting conditions with different cameras (450 images with over 2,162 labels). Among the 6 combinations of deep learning models and domain adaptation techniques, the Faster-RCNN model with cycle-consistent generative adversarial network adaptation technique provided the best performance (R2 = 0.94, root mean square error = 8.7). Complementary studies show that it is essential to simulate images with sufficient realism (background, leaf texture, and lighting conditions) before applying domain adaptation techniques. Furthermore, the spatial resolution should be better than 0.6 mm per pixel to identify leaf tips. The method is claimed to be self-supervised since no manual labeling is required for model training. The self-supervised phenotyping approach developed here offers great potential for addressing a wide range of plant phenotyping problems. The trained networks are available at https://github.com/YinglunLi/Wheat-leaf-tip-detection.

A leaf-level spectral library to support high-throughput plant phenotyping: predictive accuracy and model transfer.

Leaf-level hyperspectral reflectance has become an effective tool for high-throughput phenotyping of plant leaf traits due to its rapid, low-cost, multi-sensing, and non-destructive nature. However, collecting samples for model calibration can still be expensive, and models show poor transferability among different datasets. This study had three specific objectives: first, to assemble a large library of leaf hyperspectral data (n=2460) from maize and sorghum; second, to evaluate two machine-learning approaches to estimate nine leaf properties (chlorophyll, thickness, water content, nitrogen, phosphorus, potassium, calcium, magnesium, and sulfur); and third, to investigate the usefulness of this spectral library for predicting external datasets (n=445) including soybean and camelina using extra-weighted spiking. Internal cross-validation showed satisfactory performance of the spectral library to estimate all nine traits (mean R2=0.688), with partial least-squares regression outperforming deep neural network models. Models calibrated solely using the spectral library showed degraded performance on external datasets (mean R2=0.159 for camelina, 0.337 for soybean). Models improved significantly when a small portion of external samples (n=20) was added to the library via extra-weighted spiking (mean R2=0.574 for camelina, 0.536 for soybean). The leaf-level spectral library greatly benefits plant physiological and biochemical phenotyping, whilst extra-weight spiking improves model transferability and extends its utility.

Leaf-Counting in Monocot Plants Using Deep Regression Models.

Leaf numbers are vital in estimating the yield of crops. Traditional manual leaf-counting is tedious, costly, and an enormous job. Recent convolutional neural network-based approaches achieve promising results for rosette plants. However, there is a lack of effective solutions to tackle leaf counting for monocot plants, such as sorghum and maize. The existing approaches often require substantial training datasets and annotations, thus incurring significant overheads for labeling. Moreover, these approaches can easily fail when leaf structures are occluded in images. To address these issues, we present a new deep neural network-based method that does not require any effort to label leaf structures explicitly and achieves superior performance even with severe leaf occlusions in images. Our method extracts leaf skeletons to gain more topological information and applies augmentation to enhance structural variety in the original images. Then, we feed the combination of original images, derived skeletons, and augmentations into a regression model, transferred from Inception-Resnet-V2, for leaf-counting. We find that leaf tips are important in our regression model through an input modification method and a Grad-CAM method. The superiority of the proposed method is validated via comparison with the existing approaches conducted on a similar dataset. The results show that our method does not only improve the accuracy of leaf-counting, with overlaps and occlusions, but also lower the training cost, with fewer annotations compared to the previous state-of-the-art approaches.The robustness of the proposed method against the noise effect is also verified by removing the environmental noises during the image preprocessing and reducing the effect of the noises introduced by skeletonization, with satisfactory outcomes.