Three group comparison2.doc

A COMPARISON OF PERFORMANCE AMONG PATIENTS FIT WITH THE
CII HI-RESOLUTION, 3G AND TEMPO+ PROCESSORS
Anthony J. Spahr & Michael F. Dorman, Arizona State University Presented at the 2003 Conference on Implantable Auditory Prostheses BACKGROUND
The three implants available in the U.S. differ in terms of signal processing strategies and in the hardware that implements those strategies. It is likely that some of the differences among devices lead to differences in patient performance. In the study reported here we describe the outcome of an experiment in which the perception of speech, voice and music was assessed in patients fit with (i) the Advanced Bionics Corporation’s Hi-Resolution system, (ii) Cochlear Corporation’s 3G system and (iii) the Med El Tempo + system. Our aim was to look for differences in performance in hopes of finding aspects of implant design that lead to better performance. Patients were matched on CNC word scores in quiet. For example, a CII patient with a 70% CNC score was matched with a 3G patient and Tempo+ patient who scored 70% correct. A CII patient with a 50% CNC score was matched with a 3G patient and Tempo+ patient with a 50% correct score. The patients were then tested on material presented in quiet and in noise. At issue was whether the patients differed in performance on tests of detailed spectral resolution, detailed temporal resolution, performance in noise or performance at low signal levels. SUBJECTS
10 CII, 10 3G, and 10 Tempo+ patients were included in this analysis. Nine triads were matched within 4 percentage points. One triad was matched within 6 percentage points. Patients did not differ in mean age (CII=53.3 years, 3G=54.1 years, Tempo+=53.8 years ), mean length of deafness (CII=11.9 years, 3G=14.4 years, Tempo+=12.3), or mean length of experience with electrical stimulation (CII=1.5 year, 3G=1.6 years, Tempo+=1.8 years). Patients were recruited by letter from implant centers in the United States and Canada. Patients needed to score greater than 40% correct on the CNC words to be included in our sample. Patients with scores lower than this showed a floor effect when tested at +10 and at +5 dB SNR and were excluded from our sample. A total of 24 CII patients, 31 3G patients, and 12 Med El patients have been tested. All patients were tested with their ‘everyday’ settings of processors. Patients could not change settings during the tests, i.e., they could not increase or decrease sensitivity or change to a ‘whisper’ or noise reduction setting. STANDARD TEST MATERIALS
CNC words. All patients were tested with the same 50-item CNC word list in our laboratory.
City University of New York (CUNY) Sentences – A total of 24 sentences (two lists) were
used in each condition. All lists were taken from the Cochlear Corporation Investigational Test Battery CD (Boothroyd, Hanin & Hnath, 1985). NEW TEST MATERIALS
To ensure that patients from one group did not have more experience with the test materials than patients from the other group, we created a new battery of tests.
The AzBio sentences. We recorded 500 sentences ranging in length from 6 to 10 words. A total
of 5 speakers (2 male & 3 female) were used. All sentences were normalized to be of approximately equal SPL. The sentences were then processed as a 5-channel, cochlear-implant simulation and presented to 10 normal-hearing subjects for identification. Mean percent correct scores were then calculated for each of the 500 sentences. Nine lists of 40 sentences each were constructed. Sentences from 4 speakers (2 male & 2 female) were included. The mean intelligibility of the lists (89%), for normal hearing subjects listening to the 5-channel simulation, differed by less than 2 percentage points. Implant patients were asked to repeat back the sentences and were encouraged to guess when unsure. All individual sentences and lists were scored as words correct and overall percent correct was computed. Talker discrimination. The stimuli were drawn from a digital database developed at the Speech
Research Laboratory at Indiana University, Bloomington. 108 words produced by 5 males and 5 females were selected. Subjects were presented with pairs of words. Within each condition, half of the pairings were produced by the same talker and half were produced by different talkers. The words in the parings always differed, e.g., one male talker might say ‘ball’ and the other male talker might say ‘brush’. Across the different talker pairs, each talker was paired with every other talker an equal number of times. Participants responded “same” or “different” by pressing one of two buttons. Responses were scored as the percent of correct responses (Kirk, et al., 2002). Melody recognition. Each subject selected five familiar melodies from a list of 33 simple
melodies. Each melody consisted of 16 equal-duration notes and synthesized with MIDI software that
used samples of a grand piano (Smith et al., 2002). The frequencies of the notes ranged from 277 Hz to
622 Hz. The average note was concert A (440 Hz) +/- 1 semitone. The melodies were created without
distinctive rhythmic information. After the presentation of a melody subjects responded by pressing a
button from a list containing the 5 pre-selected melodies.

Vowel recognition without duration cues. 13 vowels were created with the use of KLATT
software in /bVt/ format (“bait, Bart, bat, beet, Bert, bet, bit, bite, boat, boot, bought, bout, but”). The vowels were synthesized with brief vowels (90 ms) of equal duration so that vowel length would not be a cue to identity (Dorman, Dankowski, McCandless & Smith, 1989). There were 5 repetitions of each stimulus. The order of the items was randomized in the test list. Consonants in /e/ environment. 20 consonants were recorded in /eCe/ format, e.g., ‘a bay, ‘a
day’, a gay’, etc. A single male talker made five productions of each token. The pitch and vocalic portion of each token was intentionally varied. The order of items was randomized in the test list. CNC Words & Sentences in
Percent Correct
CNC word scores were used to match subjects. Although we tested a total of 24 CII subjects, 31 3G subjects and 12 Tempo+ subjects, only 10 triads could be made using our matching criteria. All subjects included in this analysis were able to score between 90 and 100 percent correct on The Az Bio sentences in quiet yielded a range of scores between 35 and 97 percent correct, a range very similar to that seen in the CNC word test (42 to 94 percent correct). Tests of Temporal and Spectral Resolution
Percent Correct
No significant differences were seen between groups for any test of temporal or spectral Difficult Listening Environments
Percent Correct
The two environments most commonly identified as “difficult” by cochlear implant patients are noisy backgrounds and soft speech. No significant differences were seen between groups for the carefully articulated CUNY sentences in +10 dB SNR, or Az Bio sentences presented at 74 dB in +10 dB SNR, 64 dB in quiet, and 54 dB in quiet. A significant difference was seen for Az Bio sentences presented at 74 dB in +5 dB SNR (indicated by a box surrounding the test name). Difference in scores in noise (+10 dB SPL) and low level (54 dB SPL) conditions
CII Subjects
Percentage Points 10
3G Subjects
Percentage Points 10
Subjects
Percentage Points 10
This analysis is intended to show how “balanced” subjects are for listening in two difficult environments (74 dB at +10 dB SNR & 54 dB Quiet). Each bar represents the performance of one patient. The mean difference in scores was 15.4 percentage points for CII subjects, 15.6 percentage points for Tempo+ subjects, and 32.5 percentage points for 3G subjects. This measure suggests that both CII and Tempo+ patients are more balanced for listening in two unique and difficult environments. That is, they will perform similarly in both situations, without making changes to their device. Robustness Index
Average of Difficult Conditions / Quiet Condition ((74 dB @ +10 dB SNR) + (54 dB in Quiet)) / 2 Robustness Score
The robustness index is a measure used to describe drops in performance in difficult listening situations (74 dB in +10 SNR and 54 dB in quiet), relative to performance in an “easy” listening situation (74 dB in quiet). Subjects can achieve a score of 100 by showing no drop in either the noise or the low input level conditions, and a score of 0 if performance should drop to 0 in both difficult conditions. The performance level of both the CII and Tempo+ patients was significantly higher than that of the 3G patients. That is to say, the CII and Tempo+ patients were less affected by the difficult listening conditions than the 3G patients. Research funded by the NIDCD and by contributions from Advanced Bionics Corp., Cochlear Corp. and

Source: http://www.intelligent.gr/userfiles/file/Dorman_%CE%92%CE%A4%CE%95_comparison_ENG.pdf