Sprint 5

[lkorczowski]

SPRINT 5

Bruxism:

• [x] PR #43 (add image)
• [x] PR #38
• [x] redone entirely reporting with adaptation (1min, 240 epochs) (see issue #45 )

Middle Ear(th): detect events and remove artifacts

signal is a linear mixture of pulse, respiration and middle ear activation. 80% of activation should occur in REM in healthy subjects (10-15% before REM, only 5-10% outside)

3 events:

• activation tensor tympani, 1 positive sharp activation + transient to baseline (0,75 s) with negative (optional)
• activation stapedien, 1 sharp negative flat step + transient to baseline (1-1.5 s)
• activation spamodic, repetition of tensor tympani activation (min 2 tensors tympani)

  2 artifacts:

• sensor lost: flat electrode
• compressed sensor: cardiac pulse dominance

Tasks

• [x] have list of edf files with signals middle ear @RobinGuillard https://github.com/lkorczowski/Tinnitus-n-Sleep/issues/44#issuecomment-619996418
• [x] qualify signals properties (amplitude range, timing, etc.)
• [x] AmplitudeThresholding + is_good() (detect and remove artifacts)
• [x] iterate with better algorithm
• [x] visual validation
• [ ] validation of sleep phase

End of sprint Friday 1rd of May at 14:00

Backlog

• [ ] get labeled bruxism data from expert @RobinGuillard
• [x] compare masking volume data with episodes @RobinGuillard
• [x] compare EVA data with episodes @RobinGuillard
• [ ] #21 Detect bruxism events with noise
• [ ] #24 Detect awake moments
• [ ] #25 API consistency: duration, interval and other time-related parameters
• [ ] #26 Benchmarking methods: data simulation
• [ ] #29 Patient selection consensus
• [ ] #30 Are healthy subject of any interest for bruxism analysis?
• [ ] #31 Interesting notes on tinnitus variation
• [ ] #37 Unconsistent number of epoch. see issue for investigation protocol
• [ ] #41 Merging Impedance_thresholding_sliding and AmplitudeThresholding
• [ ] detection of OMA (with accelerometer channel "activity" + artefacts on EOG)
• [x] #45 ind_pick_channels
• [x] #47 case insensitive hell
• [ ] #49 channels bruxism
• [ ] #51 Add ratio of power
• [ ] #54 invalid channels in valid_chans*.pk
• [ ] #55 slow reporting function

[RobinGuillard]

Here is the complete list of nights that are containing middle ear recordings: La channel à regarder pour ces enregistrements où est enregistrée l'oreille moyenne s'appelle "airflow". Je t'ai envoyé par mail un petit notebook qui permet de visualiser rapidement des recordings avec MNE avec les signaux et notamment l'oreille moyenne assez bien calibrée.

Nuits patients : 1DA15_nuit_hab.edf 1RA17_nuit_hab.edf 1GB18_nuit_son.edf 1MF19_nuit_hab.edf HZB_nuit_1.edf HZB_nuit_2.edf HZB_nuit_3.edf (nuit partielle) SCHMIDTLIN_nuit_1_dec_OD0to0.edf SCHMIDTLIN_nuit_3_dec_OD4to0to2.edf SCHMIDTLIN_nuit_4_dec_OD3to3.edf SCHMIDTLIN_nuit_5_dec_OD0to1.edf SCHM_nuit_1.edf (nuit partielle) SCHM_nuit_2.edf SCHM_nuit_3.edf Schmidtlin_nuit_2_dec_3to0to4.edf Unger_2.edf unger_nuit_1.edf

Nuits contrôles sujets sains : Sophie_mema.edf tom_mema.edf jon_mema.edf jose_mema.edf robin_mema_nuit_1.edf robin_mema_nuit_2.edf

[RobinGuillard]

The paper about Middle Ear Muscle Activity (MEMA) is: Spontaneous Middle Ear Muscle Activity in Man: A Rapid Eye Movement Sleep Phenomenon Author(s): Michael A. Pessah and Howard P. Roffwarg Source: Science, New Series, Vol. 178, No. 4062 (Nov. 17, 1972), pp. 773-776 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1735855 It has been added to the file "Notebook/Resources"

[lkorczowski]

Review of the sprint:

• classification of the 3 types of MEMA events is not that easy
• counting the number of MEMA episodes (all types) seems to be positively correlated with tinnitus modulation (EVA p<0,05 for many thresholding, masking volume p<0,05 for fewer).
• inter-subject is validated but more work is required:
  • instead of separating all 3 MEMA, we should at least focus independent of spasmodic
  • several patients should be removed because of very poor night
  • for one given subject, we can predict EVA questionnaire based on MEMA activity (across 7 nights) we should investigate that for other patients
  • even high amplitude thresholds are correlated, i.e. maybe spasmodic ? We should investigate.
• Bruxism_and_MEMA_combine.ipynb combines MEMA in two categories: Pure MEMA and combined. Distinction between Pure Bruxism, and combined should be added
• Even with low amplitude thresholding, adaptive algorithm may not detect all epsiode, we should investigate
• We should investigate forward-backward amplitude thresholding for bruxism