Aditya Pandya

Subject independent identification of breath sounds components using multiple classifiers

Breath sounds have been shown very valuable for diagnosis of obstructive sleep apnea. In this study, we present a subject independent method for automatic classification of breath and related sounds during sleep. An experienced operator manually labelled segments of breath sounds from 11 sleeping subjects as: inspiration, expiration, inspiratory snoring, expiratory snoring, wheezing, other noise, and non-audible. Ten features were extracted and fed into 3 different classifiers: näıve Bayes, Support Vector Machine, and Random Forest. Leave-one-out method was used in which data from each subject, in turn, is evaluated using models trained with all other subject. Mean accuracy for concurrent classification of all 7 classes reached 85.4%. Mean accuracy for separating data into 2 classes, snoring and non-snoring, reached 97.8%. To our knowledge, these are the highest accuracies achieved in automatic classification of all breath sounds components concurrently and for snoring, in a subject independent model.

Spatially resolved, diffuse reflectance imaging for subsurface pattern visualization toward development of a lensless imaging platform: phantom experiments

Abstract. A portable, spatially resolved, diffuse reflectance lensless imaging technique based on the charge-coupled device or complementary metal-oxide semiconductor sensor directly coupled to the fiber optic bundle is proposed for visualization of subsurface structures such as superficial microvasculature in the epithelium. We discuss an experimental method for emulating a lensless imaging setup via raster scanning a single fiber-optic cable over a microfluidic phantom containing periodic hemoglobin absorption contrast. To evaluate the ability of the technique to recover information about the subsurface linear structures, scattering layers formed of the Sylgard® 184 Silicone Elastomer and titanium dioxide were placed atop the microfluidic phantom. Thickness of the layers ranged from 0.2 to 0.7 mm, and the values of the reduced scattering coefficient (μs′) were between 0.85 and 4.25  mm−1. The results demonstrate that fiber-optic, lensless platform can be used for two-dimensional imaging of absorbing inclusions in diffuse reflectance mode. In these experiments, it was shown that diffuse reflectance imaging can provide sufficient spatial sampling of the phantom for differentiation of 30  μm structural features of the embedded absorbing pattern inside the scattering media.

Evaluation of spatially resolved diffuse reflectance imaging for subsurface pattern visualization towards applicability for fiber optic lensless imaging setup: phantom experiments and simulation

A portable, spatially resolved diffuse reflectance (SRDR) lensless imaging technique based on the charge coupled device (CCD), or complementary metal-oxide semiconductor (CMOS) sensor directly coupled with fiber optic bundle can be proposed for visualization of subsurface structures such as intrapapillary capillary loops (IPCLs). In this article, we discuss an experimental method for emulating a lensless imaging setup via raster scanning a single fiberoptic cable (where image is relayed onto the sensor surface through a fiber-optic cable equivalent to coupling a fiber optic conduit directly onto the sensor surface without any lenses) over a microfluidic phantom containing periodic hemoglobin absorption contrast. For mimicking scattering properties of turbid media, a diffusive layer formed of polydimethylsiloxane (PDMS) and titanium dioxide (TiO2) was placed atop of the microfluidic phantom. Thickness of the layers ranged from 0.2-0.7mm, and the μs` value of the layers were in the range of 0.85 mm-1 – 4.25mm-1. The results demonstrate that a fiber-optic bundle/plate coupled lensless imaging setup has a high potential to recover intensity modulations from the subsurface patterns. Decreasing of the interrogation volumes leads to enhanced spatial resolution of diffuse reflectance imaging, and hence, can potentially overcome the scattering caused blurring.

Diffuse reflectance measurements using lensless CMOS imaging chip

To assess superficial epithelial microcirculation, a diagnostic tool should be able to detect the heterogeneity of microvasculature, and to monitor qualitative derangement of perfusion in a diseased condition. Employing a lensless CMOS imaging chip with an RGB Bayer filter, experiments were conducted with a microfluidic platform to obtain diffuse reflectance maps. Haemoglobin (Hb) solution (160 g/l) was injected in the periodic channels (grooves) of the microfluidic phantom which were covered with ~250 μm thick layer of intralipid to obtain a diffusive environment. Image processing was performed on data acquired on the surface of the phantom to evaluate the diffuse reflectance from the subsurface periodic pattern. Thickness of the microfluidic grooves, the wavelength dependent contrast between Hb and the background, and effective periodicity of the grooves were evaluated. Results demonstrate that a lens-less CMOS camera is capable of capturing images of subsurface structures with large field of view.

Holographic interferometry imaging monitoring of photodynamic (PDT) reactions in gelatin biophantom

Heat and photochemical reactions with human hemoglobin and photosensitizer were monitored by holography interference method in gelatin phantom. The method has successfully facilitated monitoring the reactions as a highresolution refraction index mapping in real time video regime. Methylene Blue was exploited as a photosensitizer.

The effect of continuous positive airway pressure on spectral encephalogram characteristics in stroke patients with obstructive sleep apnea

OBJECTIVES To evaluate if treatment with continuous positive airway pressure (CPAP) compared to usual care in stroke patients with obstructive sleep apnea (OSA) over one month reduces delta and alpha oscillations on quantitative electroencephalography (EEG) in association with improvements in cognitive or functional outcomes. METHODS Spectral EEG analysis was performed in patients with subacute stroke and OSA randomized to usual care or CPAP treatment from a previous study. RESULTS A total of 23 subjects were included. Compared to CPAP (n = 14), those in the control (n = 9) group demonstrated a significant increase in alpha power (p = 0.042). There was no between group differences for delta, theta or beta power. No significant correlation was demonstrated between the change in alpha power and indices of OSA severity or sleepiness. The increase in alpha power did not correlate with improvements in outcomes. CONCLUSION Contrary to expectations CPAP treatment of OSA did not significantly decrease alpha and delta oscillations in stroke subjects.

Non-labeled lensless micro-endoscopic approach for cellular imaging through highly scattering media

We describe an imaging approach based on an optical setup made up of a miniature, lensless, minimally invasive endoscope scanning a sample and matching post processing techniques that enable enhanced imaging capabilities. The two main scopes of this article are that this approach enables imaging beyond highly scattering medium and increases the resolution and signal to noise levels reaching single cell imaging. Our approach has more advantages over ordinary endoscope setups and other imaging techniques. It is not mechanically limited by a lens, the stable but flexible fiber can acquire images over long time periods (unlike current imaging methods such as OCT etc.), and the imaging can be obtained at a certain working distance above the surface, without interference to the imaged object. Fast overlapping scans enlarge the region of interest, enhance signal to noise levels and can also accommodate post-processing, super-resolution algorithms. Here we present that due to the setup properties, the overlapping scans also lead to dramatic enhancement of non-scattered signal to scattered noise. This enables imaging through highly scattering medium. We discuss results obtained from in vitro investigation of weak signals of ARPE cells, rat retina, and scattered signals from polydimethylsiloxane (PDMS) microchannels filled with hemoglobin and covered by intralipids consequently mimicking blood capillaries and the epidermis of human skin. The development of minimally invasive procedures and methodologies for imaging through scattering medium such as tissues can vastly enhance biomedical diagnostic capabilities for imaging internal organs. We thereby propose that our method may be used for such tasks in vivo.

Contrast enhancement for diffuse reflectance imaging by microbubbles

Microbubble contrast agents can potential improve optical imaging diagnostics based on increasing light scattering. Detection and quantification of microcirculation have been improved in diffuse reflectance imaging by administrating microbubbles

Identifying/digital staining of diseased regions by monitoring disease specific marker molecules using Raman spectral libraries

We have collected Raman spectral libraries for obtaining operator independent diagnostics under ex-vivo and in-vivo conditions and we have tested this approach using multivariate analysis on known solutions with Raman active components. Article not available.

MICROVASCULAR PATTERNS MONITORING VIA LENSLESS CMOS CHIP

Microvascular pattern was estimated via lens-less image acquisition from intralipid/microfluidic phantom. The CMOS chip provided proficient images of subsurface structures - microchannels filled out by haemoglobin disguised under layer of Intralipid.