Siberian programmers help rehabilitate speech in cancer patients
According to the developers, multiple errors in determining the fundamental tone have hindered full-blown voice rehabilitation so far
Jul 03, 2015
Researchers at the Tomsk State University of Control Systems and Radio Electronics (TUSUR) and the Tomsk Research Institute of Oncology, both in Siberia, are working to improve their noteworthy software solution enabling cancer patients after larynx or laryngopharynx ablation surgery to restore their normal voice much faster than it is possible today, Nanonewsnet.ru reported.
The alpha version of the software was first used at the Institute of Oncology as far back as 2004. It facilitated the development of voice rehabilitation methodology that was up-to-date enough at that time.
Voice rehab consists of several computer-aided stages, the developers say. The stage enabling the assessment of speech loudness and duration was successfully completed with the initial version of the software. What requires polishing and refining is the next more complex stage, which is the determination of a voice’s fundamental tone frequency on a real time basis.
According to the developers, multiple errors in determining the fundamental tone have hindered full-blown voice rehabilitation so far. For example, a male patient who has failed to restore a normal fundamental tone of 150Hz (after just 20Hz immediately following larynx ablation) is likely to always speak in a very low voice.
"We have managed to lessen errors in determining the fundamental tone to as low as 1-3% from an undue 6-15% we had with the prior version of the software, and we have done so in real time. The algorithm only just requires 15 milliseconds to identify a frequency at which a patient has uttered a 50-millisecond sound. Which means the software takes virtually no time to respond with a notification written on the computer screen about whether the patient has taken the frequency required or not," said Sergei Kharchenko, the software project leader and a postgraduate at TUSUR’s integrated computer data security chair.
This level of diagnostics accuracy is expected to help patients cut back their rehab period required to return to normal speech. With the TUSUR-assisted methodology specialists at the Tomsk Research Institute of Oncology are said to have already been successful in taking something between eight to 22 days to fully restore speech functions in patients after larynx ablation. Without applying the automated solution the average rehab period takes two-to-four months, the source said.
At the moment, the developers are said to be upgrading their software complex for it to be downloadable to any computer. Already Tomsk physicians reportedly use the new software for clinical trials; their feedback is seen as critical in improving the new interface.
The alpha version of the software was first used at the Institute of Oncology as far back as 2004. It facilitated the development of voice rehabilitation methodology that was up-to-date enough at that time.
Voice rehab consists of several computer-aided stages, the developers say. The stage enabling the assessment of speech loudness and duration was successfully completed with the initial version of the software. What requires polishing and refining is the next more complex stage, which is the determination of a voice’s fundamental tone frequency on a real time basis.
According to the developers, multiple errors in determining the fundamental tone have hindered full-blown voice rehabilitation so far. For example, a male patient who has failed to restore a normal fundamental tone of 150Hz (after just 20Hz immediately following larynx ablation) is likely to always speak in a very low voice.
"We have managed to lessen errors in determining the fundamental tone to as low as 1-3% from an undue 6-15% we had with the prior version of the software, and we have done so in real time. The algorithm only just requires 15 milliseconds to identify a frequency at which a patient has uttered a 50-millisecond sound. Which means the software takes virtually no time to respond with a notification written on the computer screen about whether the patient has taken the frequency required or not," said Sergei Kharchenko, the software project leader and a postgraduate at TUSUR’s integrated computer data security chair.
This level of diagnostics accuracy is expected to help patients cut back their rehab period required to return to normal speech. With the TUSUR-assisted methodology specialists at the Tomsk Research Institute of Oncology are said to have already been successful in taking something between eight to 22 days to fully restore speech functions in patients after larynx ablation. Without applying the automated solution the average rehab period takes two-to-four months, the source said.
At the moment, the developers are said to be upgrading their software complex for it to be downloadable to any computer. Already Tomsk physicians reportedly use the new software for clinical trials; their feedback is seen as critical in improving the new interface.