Electrical-Impedance biofeedback instrument for swallowing rehabilitation

Dysphagia is the difficulty or abnormality of swallowing. It is usually a consequence of another health condition and may be present in any of the phases of swallowing. The cause may be structural or neurological. The effects of dysphagia range in severity; milder symptoms range from discomfort to difficulty swallowing. More severe effects include expelling food or liquid out through the mouth and nose or aspiration of material into the lungs, resulting in heavy bouts of coughing and increasing the risk of developing pneumonia. Treatment for dysphagia, if applicable, therefore depends on the type and cause of dysphagia. Evaluation charts, barium swallow, pharyngeal manometry, and endoscopy are some of the available tools in the diagnosis and classification of dysphagia. Patients diagnosed with neurogenic dysphagia may undergo motor-training exercises for swallowing rehabilitation. However, literature suggests that motor-training exercises are more effective when the patient is presented with some form of interpretable feedback of their motor activity – known as biofeedback. Patients themselves have expressed the view that biofeedback gives something to aim for. To this end various techniques for biofeedback have been formally reviewed, including surface electromyography, acoustic, endoscopy and ultrasound amongst others. Pharyngeal manometry is an invasive procedure which measures pressure along specific sections of an endoscopic-style probe and provides measurements that identify pharyngeal muscle activity sequencing. These measurements have been used as biofeedback information in rehabilitative treatment. However, such techniques are typically performed in a clinical setting and only administered by suitably trained staff. Bio-impedance has been researched and demonstrated in other literature to be a suitable tool for assessing swallowing function, comparisons have been made to manometry with mixed results. Previous Masters students in the University of Canterbury have investigated the suitability of a bio- impedance sensing device to perform as an easy-to-use, non-invasive alternative to intrusive pharyngeal sequence measuring devices. Though impedance measurement changes were obtained from a human subject, the research concluded that the number of measuring channels must be expanded. The device concept was named GULPS (Guided Utility for Latency in Pharyngeal Swallowing). This project aims to expand the number of channels of the GULPS prototype whilst retaining the sensitivity and signal gain of the original. The previous implementations of GULPS took a multi-frequency approach as a means to create channels with 40 kHz and 70 kHz as the selected nominal current injection frequencies. This approach makes adding further channels very challenging as each channel is in effect a complete impedance measurement device, each with their own signal injection circuit, amplifiers, filters and detectors. Each signal injection circuit must have its power supply isolated from the rest posing further design considerations. Channel multiplexing was determined to be the most efficient method by which to add additional channels to the bio-impedance module. This presents its own set of challenges, specifically the amplitude detector board must settle on a steady output considerably faster than the channel multiplexing rate, to allow sufficient time for multiple samples of the channel value to be taken for averaging. A tetrapolar electrode measuring scheme was selected as this approach has proven successful in previous implementations of GULPS, as well as other bio-impedance projects. The number of channels was expanded to sixteen and comparable sensitivity was demonstrated on simulated loads of 120 Ω subjected to a 10% drop in nominal impedance. A test chamber was constructed with channel electrodes spaced vertically along a column filled with saline solution to simulate conditions comparable to that of a human neck. A narrow conductive cylinder attached to an insulating rod was lowered and raised through the path of the current injection and voltage measurement electrodes. This resulted in sufficiently large voltage swings in the corresponding channel with minimal cross- talk or interference to other channels. Though the settling time for the new detector was measured to be sufficiently fast for the desired sampling rate of 1000 Hz per channel, the sampling rate had to be lowered to 700 Hz in the final implementation of the GULPS hardware. The cumulative effect of series-connected resistances meant filtering capacitors had to be lowered to almost parasitic values to try to maintain the required time constants, until the delays could be reduced no further. The per-channel bandwidth is limited to 10 Hz after additional filtering at the output stage. The prototype is initiated and configured via USB by an application developed in Visual Studio. The application allows for frequency selection for the constant current source from 70 kHz to 2 MHz so that the most suitable injection frequency could be determined experimentally. Due to time constraints the prototype was not tested on a test subject nor comparisons made to manometry readings. Testing with a simulated load designed to mimic human tissue demonstrated sufficient signal gain and low inter-channel interference, suggesting the device would be suitable to go to human trials