Geometry Regularized Autoencoders Duque, Andres F.; Morin, Sacha; Wolf, Guy ...
IEEE transactions on pattern analysis and machine intelligence,
2023-June-1, 2023-Jun, 2023-6-1, 20230601, Letnik:
45, Številka:
6
Journal Article
Recenzirano
A fundamental task in data exploration is to extract low dimensional representations that capture intrinsic geometry in data, especially for faithfully visualizing data in two or three dimensions. ...Common approaches use kernel methods for manifold learning. However, these methods typically only provide an embedding of the input data and cannot extend naturally to new data points. Autoencoders have also become popular for representation learning. While they naturally compute feature extractors that are extendable to new data and invertible (i.e., reconstructing original features from latent representation), they often fail at representing the intrinsic data geometry compared to kernel-based manifold learning. We present a new method for integrating both approaches by incorporating a geometric regularization term in the bottleneck of the autoencoder. This regularization encourages the learned latent representation to follow the intrinsic data geometry, similar to manifold learning algorithms, while still enabling faithful extension to new data and preserving invertibility. We compare our approach to autoencoder models for manifold learning to provide qualitative and quantitative evidence of our advantages in preserving intrinsic structure, out of sample extension, and reconstruction. Our method is easily implemented for big-data applications, whereas other methods are limited in this regard.
The use of Unmanned Aerial Vehicle (UAV) images for biomass and nitrogen estimation offers multiple opportunities for improving rice yields. UAV images provide detailed, high-resolution visual ...information about vegetation properties, enabling the identification of phenotypic characteristics for selecting the best varieties, improving yield predictions, and supporting ecosystem monitoring and conservation efforts. In this study, an analysis of biomass and nitrogen is conducted on 59 rice plots selected at random from a more extensive trial comprising 400 rice genotypes. A UAV acquires multispectral reflectance channels across a rice field of subplots containing different genotypes. Based on the ground-truth data, yields are characterized for the 59 plots and correlated with the Vegetation Indices (VIs) calculated from the photogrammetric mapping. The VIs are weighted by the segmentation of the plants from the soil and used as a feature matrix to estimate, via machine learning models, the biomass and nitrogen of the selected rice genotypes. The genotype IR 93346 presented the highest yield with a biomass gain of 10,252.78 kg/ha and an average daily biomass gain above 49.92 g/day. The VIs with the highest correlations with the ground-truth variables were NDVI and SAVI for wet biomass, GNDVI and NDVI for dry biomass, GNDVI and SAVI for height, and NDVI and ARVI for nitrogen. The machine learning model that performed best in estimating the variables of the 59 plots was the Gaussian Process Regression (GPR) model with a correlation factor of 0.98 for wet biomass, 0.99 for dry biomass, and 1 for nitrogen. The results presented demonstrate that it is possible to characterize the yields of rice plots containing different genotypes through ground-truth data and VIs.
Pressure is one of the essential variables to give information about engine condition and monitoring. Direct recording of this signal is complex and invasive, while angular velocity can be measured. ...Nonetheless, the challenge is to predict the cylinder pressure using the shaft kinematics accurately. In this paper, a time-delay neural network (TDNN), interpreted as a finite pulse response (FIR) filter, is proposed to estimate the in-cylinder pressure of a single-cylinder internal combustion engine (ICE) from fluctuations in shaft angular velocity. The experiments are conducted over data obtained from an ICE operating in 12 different states by changing the angular velocity and load. The TDNN's delay is adjusted to get the highest possible correlation-based score. Our methodology can predict pressure with an R2 >0.9, avoiding complicated pre-processing steps.
Abstract Background Periprosthetic joint infection (PJI) is an important cause of failure in total knee arthroplasty (TKA). Irrigation and debridement with liner exchange (I&D/L) success rates have ...varied for acute PJI. The purpose of this study is to present results of a specific protocol for I&D/L with retention of TKA components. Methods Sixty-seven consecutive I&D/L patients were retrospectively evaluated. Inclusion criteria for I&D/L were: fewer than 3 weeks of symptoms, no immunologic compromise, intact soft tissue sleeve, and well-fixed components. I&D/L consisted of extensive synovectomy, irrigation with 3 liters each of Betadine, Dakin’s, bacitracin, and normal saline solutions, and exchange of the polyethylene component. Postoperatively, all patients were treated with intravenous antibiotics. Infection was considered eradicated if the wound healed without persistent drainage, there was no residual pain or evidence of infection. Results Forty-six patients (68.66%) had successful infection eradication regardless of bacterial strain. Those with MRSA had an 80% failure rate and those with Pseudomonas aeruginosa had a 66.67% failure rate. The success rate for bacteria other than MRSA and Pseudomonas was 85.25%1. Conclusion Our protocol for I&D/L was successful in the majority of patients who met strict criteria. We recommend that PJI patients with MRSA or Pseudomonas aeruginosa not undergo I&D/L and be treated with two-stage revision. For nearly all other patients our protocol avoids the cost and patient morbidity of a two-stage revision.
Pulmonary complications after total joint arthroplasty are a burden to patients and the healthcare system. The aim of this study is to demonstrate the effectiveness of a pulmonary screening ...questionnaire and intervention protocol developed at our institution to prevent pulmonary complications.
Between 2010 and 2015, 7658 consecutive total joint arthroplasty patients at our institution were reviewed. Based on our pre-operative pulmonary risk assessment tool, 1625 patients were flagged as high pulmonary risk. Patients were determined to be high risk if they were a current or former heavy smoker with an abnormal spirometry, had a positive obstructive sleep apnea screening, required continuous positive airway pressure/bi-level positive airway pressure use, had a history of significant pulmonary disease, had an oxygen saturation <90%, or had body mass index >40. A standardized monitoring protocol and interventions including smoking cessation, treatment and optimization of primary pulmonary conditions, peri-operative inhaler use, spinal anesthesia, aspiration precautions, elevated head of bed >20° resting and >45° while eating, maintaining oxygen saturation ≥92%, early use of incentive spirometer, avoidance of narcotics and early respiratory therapy consult were initiated for all high risk patients.
Only 7 of 7658 (0.091%) patients suffered pulmonary complications after initiating our intervention protocol. These included 3 aspiration pneumonias, 1 asthma exacerbation, 1 chronic obstructive pulmonary disease exacerbation, 1 continuous positive airway pressure intolerance in a patient with obstructive sleep apnea, and 1 requirement of bi-level positive airway pressure. The pulmonary risk questionnaire accurately identified all patients who had pulmonary complications. The overall pulmonary complication rate at our institution decreased from 5.7% to 0.09% after implementing our screening questionnaire and intervention protocol (P < .0001).
Our results demonstrate a more than 63-fold reduction in pulmonary complications at our institution. Our screening questionnaire and intervention protocol is an effective way of identifying and preventing pulmonary complications.
As a contribution to the development of new dual/multifunctional drugs, a novel therapeutical scaffold merging key structural features from memantine and M30D was designed, synthesized, and explored ...for its AChE/BuChE inhibitory activity and neuroprotective effects. All synthetized hybrids were not able to inhibit AChE, but most of them exhibit inhibition with high selectivity toward butyrylcholinesterase (BuChE). Notably, among the tested compounds, amantadine/M30D hybrids with six, seven, nine, and twelve methylene groups in the spacer (5d, 5e, 5f, and 5g) not only highlighted having the best potency and selective butyrylcholinesterase inhibition greater than 83% but also, particularly 5e and 5d, elicited considerable neuroprotection when evaluated in pretreatment conditions, by reducing injury effects caused by glutamate with maximum protection reached about 47.82 ± 0.81% (5e) and 42 ± 2.20% (5d) in comparison with memantine (37.27 ± 2.69%). Likewise, we chose 5e as the hit compound, which in a glutamate excitotoxity coculture model prevented astroglia reactivity and neuronal death, as well as a 91% restoration of calcium levels and an increasing ATP level in both pre-/post-treatments of 61.48 ± 4.60 and 45.16 ± 10.55%, respectively. Regarding docking studies, a blockade of the NMDA channel pore by 5e would explain its neuroprotective response. Finally, the hit compound 5e exhibited in vitro blood–brain barrier (BBB) permeability and human plasma stability, as well as an optimal in silico neuropharmacokinetic profile. From a therapeutic perspective, merging key pharmacophoric features from memantine and M30D provides a new medicinal scaffold with dual-/multifunctional properties and human plasma stability for the future development of potential drugs for treating AD.
The etiology of patellar component loosening can be multifactorial, including component malposition, trauma, infection, and poor implant design. These cases may be managed with isolated patellar ...component revision or simultaneous patellar component with femoral and/or tibial component revision. Isolated patellar revision in the setting of aseptic loosening historically has had limited success with high rates of repeat revision.
We performed a retrospective cohort study of 75 cases diagnosed with patellar component loosening that underwent revision. Patients were followed for a minimum of 2 years. Cases were categorized as either isolated patellar (IP) revision or patellar with femoral and/or tibial component (P + O) revisions. Survivorship and re-revision causes were compared between groups. Secondary outcomes included surgical time, estimated blood loss, range of motion, and length of stay.
Fifty patients underwent IP revision, and 25 patients had P + O revision. Overall survivorship at the 2-year follow-up interval was 94.6%. Survivorship of IP revision undertaken for aseptic loosening was 94%. Survivorship of P + O revision was 96%. Eight percent of patients required reoperation from the P + O revision group, while 12% of patients in the IP revision group underwent a reoperation. Patients undergoing IP revision had better postoperative range of motion, lower surgical times, lower estimated blood loss, and decreased length of stay.
IP revision demonstrates excellent survivorship and clinical outcomes comparable to P + O revision. When appropriate, IP revision should be considered as a potential treatment option.
Level III, retrospective cohort study.
OBJECTIVE:To identify whether residents are at greater risk of radiation exposure from intraoperative fluoroscopy while earlier in their training and during more complex procedures.
METHODS:We ...analyzed 852 extracapsular proximal femur fracture fixation cases. We compared fluoroscopy times by various levels of resident training, fracture type, and implant used. Attending-only cases were used as a control group.
RESULTS:Fluoroscopy times during subtrochanteric fracture fixation (176.1 ± 11.27 seconds) were longer than intertrochanteric (111.4 ± 2.44 seconds) and basicervical fractures (91.49 ± 5.77 seconds). Long nail (150.2 ± 3.75 seconds) times were longer than short (92.3 ± 3.15 seconds) and intermediate (76.45 ± 3.01 seconds) nails. Significantly, more fluoroscopy was used in junior (115.9 ± 4.24 seconds), senior (123.0 ± 6.08 seconds), junior combo (130.6 ± 7.74 seconds), and senior combo cases (131.8 ± 6.11 seconds) compared with the control (94.91 ± 3.91).
CONCLUSION:Orthopaedic surgery residents and attendings must remain aware of radiation exposure secondary to intraoperative fluoroscopy. Appropriate personal protective equipment should be worn, and more experienced surgeons should take a more active role in the complex cases to decrease exposure risk.
LEVEL OF EVIDENCE:Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
Delirium is a common complication among elderly patients undergoing total joint arthroplasty (TJA). Its incidence has been reported from 4% to 53%. The Centers for Medicare and Medicaid Services ...consider delirium following TJA a “never-event.” The purpose of this study is to evaluate a simple perioperative protocol used to identify delirium risk patients and prevent its incidence following TJA.
Our group developed a protocol to identify and prevent delirium in patients undergoing TJA. All patients were screened and scored in the preoperative assessment, on criteria such as age, history of forgetfulness, history of agitation or visual hallucinations, history of falls, history of postoperative confusion, and inability to perform higher brain functions. Patients were scored on performance in a simple mental examination. The patients were classified as low, medium, or high risk. Patients who were identified as high risk were enrolled in a delirium avoidance protocol that minimized narcotics and emphasized nursing involvement and fluids administration.
Five of 7659 (0.065%) consecutive TJA patients from 2010 to 2015 developed delirium. A total of 422 patients were identified as high risk. All 5 patients who suffered delirium were within the high risk group. No low or medium risk patients suffered a delirium complication. Three (0.039%) patients suffered drug-induced delirium, 1 (0.013%) had delirium related to alcohol withdrawal, and 1 (0.013%) had delirium after a systemic infection.
This protocol is effective in identifying patients at high delirium risk and diminishing the incidence of this complication by utilizing a simple screening tool and perioperative protocol.