Windows AEP & IHP Protocol Martyn Hyde Kris Madsen Jill Witte Mount Sinai Hospital, Toronto mhyde@mtsinai.on.ca 1
Agenda Updated Assessment Protocol Major system errors AD Dx issues Click artifact 60 Hz review Residual Noise Level ABR report formatting AOB Report formatting demo 2
Updated Assessment Protocol Integrated document with all updates Embedded protocol changes: Min levels, adjustment factors, AD, RNL, 60 Hz, 5dB steps, etc Embedded NavPro/Win AEP setup (appendix) Added Conditioned Play Audiometry section Index of Updates 3
Major System Errors Windows AEP/gold headbox incompatibilities Replaced with silver headbox, Win AEP 6.1.0 prolonged test time (deep timing error) Loss of average, reopen AEP data collection (headbox comm. error) Loss of average, reconnect? YES Blank screen on startup Power down, restart,?static,?replace if frequent Notify mhyde@mtsinai.on.ca 4
5
AD issues & protocol changes 6
Website Secure Section IHP.mtsinai.on.ca Website managers: Leslie Garbett/Jill Witte Contact: lgarbett@mtsinai.on.ca Staff Support: Login/Password protocol AN/AD Review Rance, Trends in Amp 43pp 7
AD Prevalence Funnel Plot, Rance 2005 1000 size of cohort 100 median 5.0 % mean 7.2 % 10 0 5 10 15 20 25 AD prevalence %, 90% CI 8
Classic CM example in AD 9
Problematic CM/AEP records in AD/?AD big CM, small wave V small CM, no ABR big CM, wave I? + slow wave 10
?AD, 2k 65 db,.5k 45 db 190143 11
12
Rare-cond replication vs subaveraging 13
14
15
Protocol changes - draft Do AD protocol if ABR 2k >70 db even if sharp Rar & Con clicks 85 db, 2x4000 or RNL < 20 Choose best ABR rar or con Bracket threshold for best r/c, rare if same Repeat highest neg 2k level with best r/c pip Notify mhyde@mtsinai.on.ca 16
Interpretation OAE! & no clk ABR rar and con at 85 db = AD+ OAE! & ABR 2k >! 70 dbnhl = AD? No OAE, clk ABR best r/c > 70, CM85 > ABR = AD? No OAE, no clk ABR r/c 85, no CM, ABN tymp = AD? No OAE, no clk ABR r/c 85, no CM, N tymp = AD- No OAE, sharp clk ABR r/c 85, CM < ABR = AD- Absent OAE does not R/O AD Abnormal tymp threatens CM absence CM85 > 2x clk ABR best r/c implies AD strongly If clk r/c differ & > alt clk, 2k & 4k true thresholds may be better than alt 2k and alt 4k thresholds If AD? or +, true thresholds may be better 17
Click stimulus artifact 18
large click artifact 19
Click stimulus artifact Why now is new equipment more sensitive? New Bio-logic inserts more shielded Acoustically equivalent to old inserts (CAL!) USE NEW INSERTS & PATIENT CABLE Interim fix: add 1.0 ms post-stim delay 20
60 Hz problems & actions 21
22
23
Unilateral L, equal ipsi/contra BC, R masked BC Intermittent low 60 Hz artifact 24
60 Hz proof and remedies Slow bend in 0 db 2kHz runs If unilateral, complete test on no-60 Hz side Check sources & electrode lead positions Check Z, is 60 Hz bigger on high-z mastoid? If high Z mastoid > 10 k, replace, do 0 db run If 0 db 60 Hz > 5 divs, abandon test If < 5 divs, notch ON, do 0 db run, proceed. If residual 60 Hz, focus on 2kHz. Look for early bend and 0 db waveform match at 500 Hz. Do NOT use the notch filter routinely! Why? 25
Residual Noise level What is it? Why is it useful? RNL/average mismatch 60 Hz Management of Max Sweeps & RNL criterion 26
What is RNL RNL is the estimated standard deviation of the deflections in the average record, calculated in such a way that response is absent RNL is calculated every 256 accepted sweeps It is displayed & is part of the tag strings Autostop if criterion met for ALL channels 27
Fsp = N x Var of Average Var of Single Point 28
Why is RNL useful? If criterion is met, averaging may be stopped Increases consistency of averaged noise, relative to ritual use of 2000 sweeps Increases reliability of ABR+ and ABR decisions, especially in single averages Must replicate for actual threshold bracket 29
30
RNL/average mismatch 60 Hz voltage changes from sweep to sweep at the single point used and inflates the RNL Average may/may not show 60 Hz deflection, depending on the rep rate, no. of accepted sweeps, and the size of the interference Can get big RNL but smooth average record Notch filter IN will tend to give true EEG RNL 31
Unilateral, normal R, masked BC Intermittent low 60 Hz, RNL-average mismatch 32
Max sweeps and RNL strategy Exceeding 4k max to meet RNL is tedious, every 256 sweeps.. RNL may not match visual impression may want to continue even if criterion satisfied Solution: set max sweeps to 8000 and RNL to 20 nv, ALWAYS stop averages manually 33
IHP ABR Reporting Format Key elements are required, for province-wide consistency and protocol description DON T SUPERIMPOSE, USE MATCH ALWAYS EQUALIZE PLOT WIDTH ( default ) RNL 2-digit O/S update Recommend 0.4 uv/div scaling.. 34
35