Electroencephalogram Connectivity for the Diagnosis of Psychogenic Non-epileptic Seizures.

Psychogenic non-epileptic seizures (PNES) are attacks that resemble epilepsy but are not associated with epileptic brain activity and are regularly misdiagnosed. The current gold standard method of diagnosis is expensive and complex. Electroencephalogram (EEG) analysis with machine learning could improve this. A k-nearest neighbours (kNN) and support vector machine (SVM) were used to classify EEG connectivity measures from 48 patients with PNES and 29 patients with epilepsy. The synchronisation method - correlation or coherence - and the binarisation threshold were defined through experimentation. Ten network parameters were extracted from the synchronisation matrix. The broad, delta, theta, alpha, beta, gamma, and combined 'all' frequency bands were compared along with three feature selection methods: the full feature set (no selection), light gradient boosting machine (LGBM) and k-Best. Coherence was the highest performing synchronisation method and 0.6 was the best coherence threshold. The highest balanced accuracy was 89.74%, produced by combining all six frequency bands and selecting features with LGBM, classified by the SVM. This method returned a comparatively high accuracy but at a high computation cost. Future research should focus on identifying specific frequency bands and network parameters to reduce this cost. Clinical relevance - This study found that EEG connectivity and machine learning methods can be used to differentiate PNES from epilepsy using interictal recordings to a high accuracy. Thus, this method could be an effective tool in assisting clinicians in PNES diagnosis without a video- EEG recording of a habitual seizure.

Binary Online Learned Descriptors.

We propose a novel approach to generate a binary descriptor optimized for each image patch independently. The approach is inspired by the linear discriminant embedding that simultaneously increases inter and decreases intra class distances. A set of discriminative and uncorrelated binary tests is established from all possible tests in an offline training process. The patch adapted descriptors are then efficiently built online from a subset of features which lead to lower intra-class distances and thus, to a more robust descriptor. We perform experiments on three widely used benchmarks and demonstrate improvements in matching performance, and illustrate that per-patch optimization outperforms global optimization.

High correlation between performance on a virtual-reality simulator and real-life cataract surgery.

PURPOSE: To investigate the correlation in performance of cataract surgery between a virtual-reality simulator and real-life surgery using two objective assessment tools with evidence of validity. METHODS: Cataract surgeons with varying levels of experience were included in the study. All participants performed and videorecorded three standard cataract surgeries before completing a proficiency-based test on the EyeSi virtual-reality simulator. Standard cataract surgeries were defined as: (1) surgery performed under local anaesthesia, (2) patient age >60 years, and (3) visual acuity >1/60 preoperatively. A motion-tracking score was calculated by multiplying average path length and average number of movements from the three real-life surgical videos of full procedures. The EyeSi test consisted of five abstract and two procedural modules: intracapsular navigation, antitremor training, intracapsular antitremor training, forceps training, bimanual training, capsulorhexis and phaco divide and conquer. RESULTS: Eleven surgeons were enrolled. After a designated warm-up period, the proficiency-based test on the EyeSi simulator was strongly correlated to real-life performance measured by motion-tracking software of cataract surgical videos with a Pearson correlation coefficient of -0.70 (p = 0.017). CONCLUSION: Performance on the EyeSi simulator is significantly and highly correlated to real-life surgical performance. However, it is recommended that performance assessments are made using multiple data sources.

Predicting complications with pretreatment testing in infantile haemangioma treated with oral propranolol.

BACKGROUND: Since 2008, orally administered propranolol has rapidly gained acceptance as the preferred therapy for haemangiomas, and is usually initiated by ophthalmologists, dermatologists or plastic surgeons who do not routinely use propranolol for any other indication. During the initial years when experience was limited, most healthcare professionals justifiably adopted a cautious approach when initiating and monitoring treatment. A consensus recommendation from the American Society of Dermatologists suggests routine observation, monitoring and cardiology assessments prior to propranolol initiation. AIM: This study aims to analyse treatment initiation in a large tertiary children's hospital and investigate the value of pretreatment testing in predicting commonly seen adverse reactions of propranolol. METHOD: 104 eligible patients treated between January 2009 and July 2012 were included. All patients underwent pretesting either with protocol A (administration of test dose with routine observations) or protocol B (cardiology clinic assessment, including two-dimensional echocardiography without test dose). RESULTS: 38.5% (40/104) of patients developed adverse reactions during treatment; however, there were no severe or life-threatening reactions. Protocol A has a sensitivity of 0 (95% CI 0 to 0.17) and specificity of 0.95 (95% CI 0.83 to 0.99). Protocol B has a sensitivity of 0.07 (95% CI 0 to 0.34) and specificity of 0.86 (95% CI 0.63 to 0.96). CONCLUSIONS: The predictive values of both protocols for the commonly observed adverse reactions are low. In this series, there is no evidence to suggest that routine pretreatment testing before propranolol initiation is of any value in otherwise healthy children.

"PhacoTracking": an evolving paradigm in ophthalmic surgical training.

Motion analysis has been validated as a tool to evaluate surgical skill. We investigated a novel computer vision-based tool for the evaluation of surgical movements during cataract surgery. A prospective cohort analysis of 2 groups was performed. Ten videos of junior surgeons (ie, those with <200 cases) and 10 videos of senior surgeons (ie, those with >1000 cases) were analyzed. Movement parameters were measured over an entire procedure. Significant statistical differences were found between novice and expert surgeons for total path length (P = .002), number of movements (P = .05), and total time (P = .004). Our study has shown that computer vision-based motion analysis can be successfully applied to video recordings of cataract surgery to provide robust measurements of instrument motion. Further work needs to be done to evaluate its usefulness in training and feedback.