Application of multiple deep learning models for automatic burn wound assessment.

PURPOSE: Accurate assessment of the percentage of total body surface area (%TBSA) burned is crucial in managing burn injuries. It is difficult to estimate the size of an irregular shape by inspection. Many articles reported the discrepancy of estimating %TBSA burned by different doctors. We set up a system with multiple deep learning (DL) models for %TBSA estimation, as well as the segmentation of possibly poor-perfused deep burn regions from the entire wound. METHODS: We proposed boundary-based labeling for datasets of total burn wound and palm, whereas region-based labeling for the dataset of deep burn wound. Several powerful DL models (U-Net, PSPNet, DeeplabV3+, Mask R-CNN) with encoders ResNet101 had been trained and tested from the above datasets. With the subject distances, the %TBSA burned could be calculated by the segmentation of total burn wound area with respect to the palm size. The percentage of deep burn area could be obtained from the segmentation of deep burn area from the entire wound. RESULTS: A total of 4991 images of early burn wounds and 1050 images of palms were boundary-based labeled. 1565 out of 4994 images with deep burn were preprocessed with superpixel segmentation into small regions before labeling. DeeplabV3+ had slightly better performance in three tasks with precision: 0.90767, recall: 0.90065 for total burn wound segmentation; precision: 0.98987, recall: 0.99036 for palm segmentation; and precision: 0.90152, recall: 0.90219 for deep burn segmentation. CONCLUSION: Combining the segmentation results and clinical data, %TBSA burned, the volume of fluid for resuscitation, and the percentage of deep burn area can be automatically diagnosed by DL models with a pixel-to-pixel method. Artificial intelligence provides consistent, accurate and rapid assessments of burn wounds.

A pressure ulcers assessment system for diagnosis and decision making using convolutional neural networks.

BACKGROUND/PURPOSE: Pressure ulcers are a common problem in hospital care and long-term care. Pressure ulcers are caused by prolonged compression of soft tissues, which can cause local tissue damage and even lead to serious infections. This study uses a deep learning algorithm to construct a system that diagnoses pressure ulcers and assists in making treatment decisions, thus providing additional reference for first-line caregivers. METHODS: We performed a retrospective research of medical records to find photos of patients with pressure ulcers at National Taiwan University Hospital from 2016 to 2020. We used photos from 2016 to 2019 for training and after removing the photos which were vague, underexposed, or overexposed, 327 photos were obtained. The photos were then labeled as "erythema" or "non-erythema" for the first classification task and "extensive necrosis", "moderate necrosis" or "limited necrosis" for the second, by consensus of three recruited physicians. An Inception-ResNet-v2 model, a kind of Convolutional Neural Network (CNN), was applied for training these two classification tasks to construct an assessment system. Finally, we tested the model with the photos of pressure ulcers taken from 2019 to 2020 to verify its accuracy. RESULTS: For the task of classification of erythema and non-erythema wounds, our CNN model achieved an accuracy of about 98.5%. For the task of classification of necrotic tissue, our model achieved accuracy of about 97%. CONCLUSION: Our CNN model, which was based on Inception-ResNet-v2, achieved high accuracy when classifying different types of pressure ulcers, making it applicable in clinical circumstances.

Human induced pluripotent stem cell line with genetically encoded fluorescent voltage indicator generated via CRISPR for action potential assessment post-cardiogenesis.

Genetically encoded fluorescent voltage indicators, such as ArcLight, have been used to report action potentials (APs) in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). However, the ArcLight expression, in all cases, relied on a high number of lentiviral vector-mediated random genome integrations (8-12 copy/cell), raising concerns such as gene disruption and alteration of global and local gene expression, as well as loss or silencing of reporter genes after differentiation. Here, we report the use of clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 nuclease technique to develop a hiPSC line stably expressing ArcLight from the AAVS1 safe harbor locus. The hiPSC line retained proliferative ability with a growth rate similar to its parental strain. Optical recording with conventional epifluorescence microscopy allowed the detection of APs as early as 21 days postdifferentiation, and could be repeatedly monitored for at least 5 months. Moreover, quantification and analysis of the APs of ArcLight-CMs identified two distinctive subtypes: a group with high frequency of spontaneous APs of small amplitudes that were pacemaker-like CMs and a group with low frequency of automaticity and large amplitudes that resembled the working CMs. Compared with FluoVolt voltage-sensitive dye, although dimmer, the ArcLight reporter exhibited better optical performance in terms of phototoxicity and photostability with comparable sensitivities and signal-to-noise ratios. The hiPSC line with targeted ArcLight engineering design represents a useful tool for studying cardiac development or hiPSC-derived cardiac disease models and drug testing.

Response of a local hospital to a burn disaster: Contributory factors leading to zero mortality outcomes.

OBJECTIVE: To investigate the outcomes of a local healthcare system in managing a burn mass casualty incident (BMCI). METHODS: Thirty-three victims admitted to the National Taiwan University Hospital within 96h of the explosion were included in the study. Data were recorded on: patient demographics, Baux score, laboratory data, management response, treatment strategies, and outcomes. Case notes from June 27, 2015 to November 2015 were reviewed with a focus on fluid resuscitation, ventilation support, nutrition, infection control, sepsis treatment, and wound closure plan. RESULTS: Female predominance (mean age: 21.7 years) and lower extremity circumferential flame burns were the characteristics of the burn injury. The mean Baux score was 70±18. The mean burn area was 42% of the total body surface area (TBSA). A total of 79% patients arrived at the hospital within 24h of sustaining injuries. Intensive care unit (ICU) admission criteria were modified to accommodate patients with 40% TBSA of burns, facilities were expanded from 4 ICU beds to 18 beds, and new staff was recruited. A total of 36% patients (n=12/33, 62±13 TBSA of burns) required fluid resuscitation. The mean volume of Lactate Ringer administered in the first 24h of burns was 3.34±2.18ml/kg/%TBSA, while the mean volume of fresh frozen plasma administered was 0.60±0.63ml/kg/h. Forty-two percent patients were intubated on the day of admission, and 71% of the intubated patients had inhalation injuries that were confirmed by diagnostic bronchoscopy. The mean intubation period was 17±9 days. The incidence of pulmonary edema was 58% (n=7/12), possibly due to sub-optimal monitoring. Of these, 57% (n=4/7) patients progressed to adult respiratory distress syndrome, but were successfully treated with early strict fluid restriction, systemic antibiotics, ventilation support, and bronchial lavage. A total of 94% patients received grafting. The mean grafted area was 4432.3±3891cm2. Tube feeding was provided to patients with burns >40% TBSA. All patients tolerated gastric tube feeding without conversion to duodenal switch. On admission, all patients received prophylactic antibiotics. Septic shock was noted in 12 patients, but no mortality occurred. The mean hospital stay was 1.5 days per percent burn. CONCLUSIONS: This article highlights the value of precise triage, traffic control, and effective resource allocation in treating a BMCI. Effective supporting systems for facility expansion, staff recruitment, medical supplies and clear-cut treatment strategies for severely burned patients are contributory factors leading to zero mortalities in our series, in addition to young age and minimal inhalation injuries. The need for reevaluation of the safety of cornstarch powder in festival activities is clear.

Risk factors for complications following immediate tissue expander based breast reconstruction in Taiwanese population.

BACKGROUND/PURPOSE: Breast cancer patients in Asia show considerable disparities from Caucasian patients, such as younger age of onset and lower rates of smoking, obesity, and diabetes. Findings of prior studies regarding risk factors associated with complications in tissue expander may not hold for Asian populations, since most of these studies involved Caucasian patients. In this study, we surveyed risk factors in the Taiwanese population, providing additional evidence about the important differences and discuss the implications for clinical practice. METHODS: Patients who underwent immediate, two-stage, tissue expander breast reconstruction from December 2008 to August 2014 in the National Taiwan University Hospital, Taipei, Taiwan were included. Follow-up observations of all patients were conducted until December 2014. Complications occurring during the tissue expander stage were evaluated. Multivariate regression modeling was used to identify risk factors for complications. RESULTS: A total of 246 consecutive, immediate, smooth round tissue expander placements were performed for breast reconstruction. The most common complication was skin necrosis (4.9%), followed by wound dehiscence (4.1%). In the multivariate model, body mass index (BMI) ≥ 24 kg/m2 was the only risk factor that reached statistical significance (odds ratio: 2.41, 95% confidence interval: 1.17-4.96). CONCLUSION: We provided evidence that racial disparities have an impact on the risk factors for complications associated with tissue expander breast reconstruction. BMI≥24 kg/m2 was the only risk factor significantly associated with complications. Clinically, BMI≥24 kg/m2, rather than the standard definition of obesity (BMI > 30 kg/m2), may be a more suitable cutoff point for risk in patients of Asian ethnicity.

The virtual cooperation platform in enterprise and supplier cooperation models.

Abstract This study examines the use of the virtual enterprise network supplier supply-chain model of business behavior in creating synergies of cooperation. To explore virtual network behavior, it evaluates 60 samples, taken from of a few supply chains, and 17 items meeting certain behavioral criteria. Such an analysis may help to reduce costs and processing time effectively, as well as promote effective communication. Furthermore, the study of behavior in this electronic setting is a reliable and useful assessment method.