OBJECTIVES Our goal was to develop high throughput computer vision (CV) algorithms to detect blood stains in thoracoscopic surgery and to determine how the detected blood stains are associated with postoperative outcomes. METHODS Blood pixels in surgical videos were identified by CV algorithms trained with thousands of blood and non-blood pixels randomly selected and manually labelled. The proportion of blood pixels (PBP) was computed for key video frames to summarize the blood stain information during surgery. Statistical regression analyses were utilized to investigate the potential association between PBP and postoperative outcomes, including drainage volume, prolonged tube indwelling duration (≥5 days) and bleeding volume. RESULTS A total of 275 patients undergoing thoracoscopic lobectomy were enrolled. The sum of PBP after flushing (P < 0.022), age (P = 0.005), immediate postoperative air leakage (P < 0.001), surgical duration (P = 0.001) and intraoperative bleeding volume (P = 0.033) were significantly associated with drainage volume in multivariable linear regression analysis. After adjustment using binary logistic regression analysis, the sum of the PBP after flushing [P = 0.017, odds ratio 1.003, 95% confidence interval (CI) 1.000–1.005] and immediate postoperative air leakage (P < 0.001, odds ratio 4.616, 95% CI 1.964–10.847) were independent predictors of prolonged tube indwelling duration. In the multivariable linear regression analysis, surgical duration (P < 0.001) and the sum of the PBP of the surgery (P = 0.005) were significantly correlated with intraoperative bleeding volume. CONCLUSIONS This is the first study on the correlation between CV and postoperative outcomes in thoracoscopic surgery. CV algorithms can effectively detect from surgical videos information that has good prediction power for postoperative outcomes.

BackgroundNeuronal intranuclear inclusion disease (NIID) is a slowly progressive neurodegenerative disease characterized by eosinophilic hyaline intranuclear inclusions and the GGC repeats in the 5'-untranslated region of NOTCH2NLC. The prevalent presence of high-intensity signal along the corticomedullary junction on diffusion-weighted imaging (DWI) helps to recognize this heterogeneous disease despite of highly variable clinical manifestations. However, patients without the typical sign on DWI are often misdiagnosed. Besides, there are no reports of NIID patients presenting with paroxysmal peripheral neuropathy-like onset to date.Case presentationWe present a patient with NIID who suffered recurrent transient numbness in arms for 17 months. Magnetic resonance imaging (MRI) showed diffuse, bilateral white matter lesions without typical subcortical DWI signals. Electrophysiological studies revealed mixed demyelinating and axonal sensorimotor polyneuropathies involving four extremities. After excluding differential diagnosis of peripheral neuropathy through body fluid tests and a sural nerve biopsy, NIID was confirmed by a skin biopsy and the genetic analysis of NOTCH2NLC.ConclusionThis case innovatively demonstrates that NIID could manifest as paroxysmal peripheral neuropathy-like onset, and addresses the electrophysiological characteristics of NIID in depth. We broaden the clinical spectrum of NIID and provide new insights into its differential diagnosis from the perspective of peripheral neuropathy.

Bearings are essential components of rotating machinery used in mechanical systems, and fault diagnosis of bearings is of great significance to the operation and maintenance of mechanical equipment. Deep learning is a popular method for bearing fault diagnosis, which can effectively extract the in-depth information of fault signals, thus achieving high fault diagnosis accuracy. However, due to the complex deep structure of deep learning, most deep learning methods require more time and resources for bearing fault diagnosis. This paper proposes a bearing fault diagnosis method combining feature engineering and fuzzy broad learning. First, time domain, frequency domain, and time-frequency domain features are extracted from the bearing signals. Then the stability and robustness indexes of these features are evaluated to complete the feature engineering. The features obtained by feature engineering are used as the input of the fault diagnosis model, and three sets of experimental data validate the model. The experimental results show that the proposed method can achieve the bearing fault diagnosis accuracy of 96.43% on the experimental bench data, 100% on the Case Western Reserve University dataset, and 100% on the centrifugal pump bearing fault dataset, with a time of approximately 0.28 s. The results show that this method has the advantages of accuracy, rapidity, and stability of bearing fault diagnosis.