Use of semi-synthetic data for catheter segmentation improvement.

In the era of data-driven machine learning algorithms, data is the new oil. For the most optimal results, datasets should be large, heterogeneous and, crucially, correctly labeled. However, data collection and labeling are time-consuming and labor-intensive processes. In the field of medical device segmentation, present during minimally invasive surgery, this leads to a lack of informative data. Motivated by this drawback, we developed an algorithm generating semi-synthetic images based on real ones. The concept of this algorithm is to place a randomly shaped catheter in an empty heart cavity, where the shape of the catheter is generated by forward kinematics of continuum robots. Having implemented the proposed algorithm, we generated new images of heart cavities with various artificial catheters. We compared the results of deep neural networks trained purely on real datasets with respect to networks trained on both real and semi-synthetic datasets, highlighting that semi-synthetic data improves catheter segmentation accuracy. A modified U-Net trained on combined datasets performed the segmentation with a Dice similarity coefficient of 92.6 ± 2.2%, while the same model trained only on real images achieved a Dice similarity coefficient of 86.5 ± 3.6%. Therefore, using semi-synthetic data allows for the decrease of accuracy spread, improves model generalization, reduces subjectivity, shortens the labeling routine, increases the number of samples, and improves the heterogeneity.

Estimating Vehicle Movement Direction from Smartphone Accelerometers Using Deep Neural Networks.

Characterization of driving maneuvers or driving styles through motion sensors has become a field of great interest. Before now, this characterization used to be carried out with signals coming from extra equipment installed inside the vehicle, such as On-Board Diagnostic (OBD) devices or sensors in pedals. Nowadays, with the evolution and scope of smartphones, these have become the devices for recording mobile signals in many driving characterization applications. Normally multiple available sensors are used, such as accelerometers, gyroscopes, magnetometers or the Global Positioning System (GPS). However, using sensors such as GPS increase significantly battery consumption and, additionally, many current phones do not include gyroscopes. Therefore, we propose the characterization of driving style through only the use of smartphone accelerometers. We propose a deep neural network (DNN) architecture that combines convolutional and recurrent networks to estimate the vehicle movement direction (VMD), which is the forward movement directional vector captured in a phone's coordinates. Once VMD is obtained, multiple applications such as characterizing driving styles or detecting dangerous events can be developed. In the development of the proposed DNN architecture, two different methods are compared. The first one is based on the detection and classification of significant acceleration driving forces, while the second one relies on longitudinal and transversal signals derived from the raw accelerometers. The final success rate of VMD estimation for the best method is of 90.07%.

Reviewing the connection between speech and obstructive sleep apnea.

BACKGROUND: Sleep apnea (OSA) is a common sleep disorder characterized by recurring breathing pauses during sleep caused by a blockage of the upper airway (UA). The altered UA structure or function in OSA speakers has led to hypothesize the automatic analysis of speech for OSA assessment. In this paper we critically review several approaches using speech analysis and machine learning techniques for OSA detection, and discuss the limitations that can arise when using machine learning techniques for diagnostic applications. METHODS: A large speech database including 426 male Spanish speakers suspected to suffer OSA and derived to a sleep disorders unit was used to study the clinical validity of several proposals using machine learning techniques to predict the apnea-hypopnea index (AHI) or classify individuals according to their OSA severity. AHI describes the severity of patients' condition. We first evaluate AHI prediction using state-of-the-art speaker recognition technologies: speech spectral information is modelled using supervectors or i-vectors techniques, and AHI is predicted through support vector regression (SVR). Using the same database we then critically review several OSA classification approaches previously proposed. The influence and possible interference of other clinical variables or characteristics available for our OSA population: age, height, weight, body mass index, and cervical perimeter, are also studied. RESULTS: The poor results obtained when estimating AHI using supervectors or i-vectors followed by SVR contrast with the positive results reported by previous research. This fact prompted us to a careful review of these approaches, also testing some reported results over our database. Several methodological limitations and deficiencies were detected that may have led to overoptimistic results. CONCLUSION: The methodological deficiencies observed after critically reviewing previous research can be relevant examples of potential pitfalls when using machine learning techniques for diagnostic applications. We have found two common limitations that can explain the likelihood of false discovery in previous research: (1) the use of prediction models derived from sources, such as speech, which are also correlated with other patient characteristics (age, height, sex,…) that act as confounding factors; and (2) overfitting of feature selection and validation methods when working with a high number of variables compared to the number of cases. We hope this study could not only be a useful example of relevant issues when using machine learning for medical diagnosis, but it will also help in guiding further research on the connection between speech and OSA.

Formant Frequencies and Bandwidths in Relation to Clinical Variables in an Obstructive Sleep Apnea Population.

OBJECTIVES: We investigated whether differences in formants and their bandwidths, previously reported comparing small sample population of healthy individuals and patients with obstructive sleep apnea (OSA), are detected on a larger population representative of a clinical practice scenario. We examine possible indirect or mediated effects of clinical variables, which may shed some light on the connection between speech and OSA. STUDY DESIGN: In a retrospective study, 241 male subjects suspected to suffer from OSA were examined. The apnea-hypopnea index (AHI) was obtained for every subject using overnight polysomnography. Furthermore, the clinical variables usually reported as predictors of OSA, body mass index (BMI), cervical perimeter, height, weight, and age, were collected. Voice samples of sustained phonations of the vowels /a/, /e/, /i/, /o/, and /u/ were recorded. METHODS: Formant frequencies F1, F2, and F3 and bandwidths BW1, BW2, and BW3 of the sustained vowels were determined using spectrographic analysis. Correlations among AHI, clinical parameters, and formants and bandwidths were determined. RESULTS: Correlations between AHI and clinical variables were stronger than those between AHI and voice features. AHI only correlates poorly with BW2 of /a/ and BW3 of /e/. A number of further weak but significant correlations have been detected between voice and clinical variables. Most of them were for height and age, with two higher values for age and F2 of /o/ and F2 of /u/. Only few very weak correlations were detected between voice and BMI, weight and cervical perimeter, wich are the clinical variables more correlated with AHI. CONCLUSIONS: No significant correlations were detected between AHI and formant frequencies and bandwidths. Correlations between voice and other clinical factors characterizing OSA are weak but highlight the importance of considering indirect or mediated effects of such clinical variables in any research on speech and OSA.

Speech Signal and Facial Image Processing for Obstructive Sleep Apnea Assessment.

Obstructive sleep apnea (OSA) is a common sleep disorder characterized by recurring breathing pauses during sleep caused by a blockage of the upper airway (UA). OSA is generally diagnosed through a costly procedure requiring an overnight stay of the patient at the hospital. This has led to proposing less costly procedures based on the analysis of patients' facial images and voice recordings to help in OSA detection and severity assessment. In this paper we investigate the use of both image and speech processing to estimate the apnea-hypopnea index, AHI (which describes the severity of the condition), over a population of 285 male Spanish subjects suspected to suffer from OSA and referred to a Sleep Disorders Unit. Photographs and voice recordings were collected in a supervised but not highly controlled way trying to test a scenario close to an OSA assessment application running on a mobile device (i.e., smartphones or tablets). Spectral information in speech utterances is modeled by a state-of-the-art low-dimensional acoustic representation, called i-vector. A set of local craniofacial features related to OSA are extracted from images after detecting facial landmarks using Active Appearance Models (AAMs). Support vector regression (SVR) is applied on facial features and i-vectors to estimate the AHI.