sigInspectComputeFeatures.m

function [featVals, featNames] = sigInspectComputeFeatures(segment, featNames, samplingFreq)
% [featVals featNames] = sigInspectComputeFeatures(segment, featNames)
%   compute vector of features for MER artifact detection on signal segment
% IN
%   segment      - signal segment (multi-channel also supported)
%   featNames    - identifiers of M features to be calculated (all features if empty)
%   samplingFreq - sampling frequency for computation of
%                  frequency-dependent features (default: segment length)
% OUT
%   featVals  - row vector of M values (Or Nch x M matrix, where Nch is number of channels)
%   featNames - feature identifiers (same as input or all feature names if input featNames empty)
% 
%  E. Bakstein 2015-06-29
%
% UPDATES
%  2016-02-12 EB: support for multi-channel features + segments


    % -- supported features
    % features from raw signal
    rawFeatNames = {'pow','sigP90','sigP95','sigP99','ksnorm','powDiff','maxCorr'};
    % features from PSD
    psdFeatNames ={'maxNormPSD','stdNormPSD','psdP75','psdP90','psdP95','psdP99','maxAbsDiffPSD','psdF100'...
                    'psdBase','psdPow','psdFreq','psdMaxStep'};
    % features from time domain
    timeFeatNames = {'energyEntrophy5', 'energyEntrophy15','peakToRMS',...
                'HjorthMobility', 'HjorthComplexity', 'numPeaks', 'meanPeakHeight','peakFreq',...
                'peakRMSRatio', 'avgPeakWidth', 'energyRatio', 'sparseness', 'irregularity', 'zeroUpCrossingPeriod','sigLen'};
                
                
    if(nargin<2 || isempty(featNames))
        featNames = [rawFeatNames psdFeatNames timeFeatNames];
    end
    % 
    % if(nargin<3 || isempty(samplingFreq))
    %     samplingFreq=length(segment);
    % end
    
    % number of channels
    [Nch, ~] = size(segment);
    
    % init feature values
    featVals = nan(Nch,length(featNames));
    
    if(isempty(featNames))
        return;
    end

    
    % Remove NaNs and preprocess
    % identify NaNs
    isn=isnan(segment);
    if(all(sum(~isn,2)==0))
        warning('segment contains only NaNs, skipping')
        return
    end
    % trim trailling all-nan rows
    segment = trimTrailNans(segment);
    %N = size(segment,2);
    % subtract DC component
    segment = double(segment); 
    segment = segment - mean(segment, 2, 'omitnan'); %repmat(nanmean(segment,2),1,N);
    
    % Compute raw features
    rawInds = ismember(featNames, rawFeatNames);
    if any(rawInds)
        rawFeatureList = featNames(rawInds);
        rawVals = computeRawFeatures(segment, rawFeatureList, samplingFreq);
        featVals(:, rawInds) = rawVals;
    end
    
    % Compute time-domain features
    timeInds = ismember(featNames, timeFeatNames);
    if any(timeInds)
        timeFeatureList = featNames(timeInds);
        timeVals = computeTimeDomainFeatures(segment, timeFeatureList, samplingFreq);
        featVals(:, timeInds) = timeVals;
    end
    
    % Compute PSD features
    NFFT = 2048; % fft length in PSD algorithm
    psdInds = ismember(featNames, psdFeatNames);
    if any(psdInds)
        psdFeatureList = featNames(psdInds);
        psdVals = computePSDFeatures(segment, psdFeatureList, samplingFreq, NFFT);
        featVals(:, psdInds) = psdVals;
    end
end


function rawVals = computeRawFeatures(segment, featureList, samplingFreq)
    % Compute raw features
    [Nch, ~] = size(segment);
    rawVals = nan(Nch, length(featureList));
    segmentNorm = segment ./ sqrt(mean(segment.^2, 2, 'omitnan')); % Normalize by RMS
    
    for fi = 1:length(featureList)
        fn = featureList{fi};
        switch fn
            case 'pow'
                rawVals(:, fi) = mean(segment.^2, 2, 'omitnan');
            case 'sigP90'
                rawVals(:, fi) = prctile(segmentNorm', 90)';
            case 'sigP95'
                rawVals(:, fi) = prctile(segmentNorm', 95)';
            case 'sigP99'
                rawVals(:, fi) = prctile(segmentNorm', 99)';
            case 'ksnorm'
                rawVals(:, fi) = compKS(segmentNorm);
            case 'powDiff'
                rawVals(:, fi) = compPowDiff(segmentNorm, samplingFreq);
            case 'maxCorr'
                rawVals(:, fi) = compMaxCorr(segmentNorm, samplingFreq);
        end
    end
end

function psdVals = computePSDFeatures(segment, featureList, samplingFreq, NFFT)
    % Compute PSD features
    [Nch, ~] = size(segment);
    psdVals = nan(Nch, length(featureList));

    % mean clean PSD spectrum - calculated on clean segments of 60 train
    % signals from commonly-annotated EMBC 2015 paper signalIds set
    meanClnPSD=[4.77562431760460e-05,0.000166178838654735,0.000227806114901376,0.000146040306929677,0.000142911335033798,0.000192389466398494,0.000241298241799244,0.000313328240515836,0.000551530514429587,0.000676116046498073,0.000589870009775301,0.000666057276492933,0.000976870506074863,0.00153008829774771,0.00111734213373207,0.00118978582786262,0.00138826592498130,0.00183274813159608,0.00166025979142410,0.00172365391321772,0.00187407420905555,0.00232046231406939,0.00231718653722749,0.00223308922838443,0.00229458106215795,0.00249993684340288,0.00265075493463280,0.00270928586528559,0.00290230525509276,0.00311006754409378,0.00343268066344955,0.00329037019721431,0.00334454943324877,0.00344422750530643,0.00341031274248160,0.00341195297684773,0.00349486459313133,0.00359357475245854,0.00396922454483106,0.00406929313602893,0.00411443190955744,0.00417929900513955,0.00399602431438698,0.00411053910110060,0.00413623833438308,0.00422796555655805,0.00442006297486236,0.00450079497340097,0.00449572150989367,0.00459150757377476,0.00480359422910404,0.00477445815618844,0.00467302146324857,0.00475741028179341,0.00466343919152143,0.00479140294729342,0.00490962350523759,0.00495305358928402,0.00492313075478619,0.00494184122580270,0.00507271776515933,0.00497804382483539,0.00491620260712145,0.00497576630307890,0.00504328104327651,0.00500239299106261,0.00505212342917972,0.00500623040375661,0.00511808393864830,0.00517149453283253,0.00515532861107924,0.00525727966706852,0.00532564371679703,0.00524950588753595,0.00513195437617655,0.00515258273909534,0.00525297852364360,0.00518068254349220,0.00516886501298688,0.00525421453752356,0.00521933405230834,0.00522291129028569,0.00515097660980339,0.00513469256030222,0.00509685952940660,0.00513033954767364,0.00510540184120212,0.00511239023673139,0.00510007133642861,0.00515262337645407,0.00509412434890254,0.00499609041421214,0.00498431690873961,0.00499109119680796,0.00500562129446170,0.00495958712223161,0.00502210600501136,0.00508880992250116,0.00517377931021268,0.00505011303168902,0.00505528149348956,0.00500098263183940,0.00499810504945096,0.00492261532910077,0.00496610863134506,0.00497472013897678,0.00490580239360861,0.00482089626038923,0.00475508596774923,0.00474499271537251,0.00477645929755003,0.00475545145715365,0.00478105644166975,0.00465488152469854,0.00471409717536129,0.00471048021369165,0.00475915457598463,0.00468940956222786,0.00462034906737799,0.00449382096630521,0.00451091562820079,0.00454705830525373,0.00455982028848946,0.00446689456067388,0.00445675377991204,0.00444786821089221,0.00439386079321514,0.00447871981464816,0.00442405879331474,0.00439621058676603,0.00437314571485769,0.00423853279637452,0.00432281038873557,0.00428664394427408,0.00420520265904031,0.00420277442594741,0.00419779497156240,0.00416879314864072,0.00410729046859493,0.00400428675858715,0.00405876046320252,0.00410178417379235,0.00397467089452320,0.00391046801293594,0.00391752187150546,0.00388483532227158,0.00386087601599158,0.00385741600500733,0.00382291959822417,0.00383491328601276,0.00376932951838254,0.00368331069678440,0.00371123558130991,0.00366310738526172,0.00364366432217688,0.00362413382217811,0.00354570588820700,0.00358475706353929,0.00366862215868109,0.00351006826610550,0.00345978657800265,0.00342900994917641,0.00339385866148339,0.00336125790170430,0.00339453771900013,0.00331911911350546,0.00329870399182654,0.00328714338438837,0.00328874618744437,0.00327686272574081,0.00324735483354771,0.00322155179713891,0.00315570229917142,0.00315326110916976,0.00319684641646357,0.00314736171834294,0.00306681677500525,0.00301049225149239,0.00300742249482694,0.00302020068853759,0.00298394947205322,0.00295828496246546,0.00295793639653616,0.00294729960938506,0.00288618218091725,0.00287939586512066,0.00288281375613166,0.00286836338449627,0.00285755463853944,0.00279416086371548,0.00270718061793415,0.00271371249989810,0.00270919503275142,0.00269729330913122,0.00269408511581507,0.00267720221524766,0.00264801544357892,0.00263377346083732,0.00262111731761184,0.00263595262004538,0.00262626968521818,0.00261661460063328,0.00250243478137951,0.00248250812259610,0.00245990029590071,0.00245886232108565,0.00246224867222939,0.00242636884710659,0.00239949345834174,0.00237552764431891,0.00239456292005979,0.00235664221929828,0.00230486147795030,0.00247973206258655,0.00268951184804001,0.00228710246732938,0.00220864511020846,0.00218157917992252,0.00218259776769702,0.00214869175208088,0.00212158964037666,0.00208024229172271,0.00212793967087608,0.00206507769806619,0.00207025116067468,0.00203882462468301,0.00203359939070705,0.00202612148388391,0.00204754318184759,0.00196389620383922,0.00195354774577354,0.00197032387955965,0.00195452996560773,0.00192208273341569,0.00191009257630677,0.00191076341976409,0.00187829312137760,0.00182625944094375,0.00179086603108449,0.00179871090759119,0.00177741456196684,0.00176465492671573,0.00174715201493156,0.00177208801878110,0.00175460360420476,0.00171475916784424,0.00171461191245323,0.00171989254168275,0.00169053441597643,0.00165043547530270,0.00168051444830999,0.00166285226395527,0.00162966127418428,0.00162595616198859,0.00163267839345235,0.00159762766965056,0.00161062958055545,0.00157638671573888,0.00154936250712022,0.00153402998560929,0.00156841336528123,0.00154897631742393,0.00148062230251802,0.00151891986972174,0.00151300911069890,0.00148996111047519,0.00147552541465048,0.00145337081960274,0.00147913373957561,0.00152326697458419,0.00144671943556326,0.00139510403541311,0.00136375026650744,0.00136915971522884,0.00136445738926030,0.00135653074777702,0.00135125142196236,0.00137363677471252,0.00135755037872982,0.00129294316898353,0.00129724586299263,0.00128041227861883,0.00129219670710612,0.00127100355594330,0.00124674711239799,0.00123654274371625,0.00123468554428219,0.00121521862163789,0.00119839984048907,0.00118088546307052,0.00119330814308405,0.00117635913832556,0.00114275440482225,0.00113522811181218,0.00114878803924825,0.00115640819464204,0.00113779751342715,0.00112207344399758,0.00110816205369944,0.00109700301580710,0.00109091009632187,0.00108894759507681,0.00107272826146688,0.00110889342564740,0.00106181762777779,0.00103718414969366,0.00103782216972429,0.00101438481238316,0.000999398724652642,0.00100855972082798,0.000992670464595343,0.000993106274414697,0.000987135064253253,0.000966204342528641,0.000971652513735865,0.000968032044291694,0.000958601124217830,0.000951318696509442,0.000945095679679914,0.000946593729861222,0.000925784758955504,0.000886789004767534,0.000902230823285457,0.000882729607039601,0.000880630403924362,0.000867110609988293,0.000872032444304620,0.000870070347674444,0.000877204091176647,0.000865341380362729,0.000847223092148180,0.000835187040099669,0.000839695954697677,0.000845348070450712,0.000823489555381757,0.000831918270618146,0.000805011115725262,0.000803496327487231,0.000784889528471347,0.000772008578721483,0.000770513995547345,0.000750230422750691,0.000742106641658372,0.000752693293885436,0.000733240046195243,0.000713192459157332,0.000723622827320018,0.000723554305401711,0.000711056990146928,0.000697823180835042,0.000686916484669332,0.000687069406791615,0.000688355141769348,0.000683167806269009,0.000656005693033487,0.000644314773263546,0.000638489836566494,0.000651305582588892,0.000646758030301500,0.000640562414538154,0.000622987735346499,0.000610607783523218,0.000618507347065954,0.000619508065463372,0.000606345491819543,0.000603250619787423,0.000592051355089845,0.000571299147710659,0.000564521927073995,0.000562252070357333,0.000548058553576822,0.000567593554269263,0.000559049591044400,0.000546368184672955,0.000534202832480463,0.000521001303156587,0.000509118486151177,0.000503513649718214,0.000498335208055070,0.000489919143470776,0.000489839565999609,0.000482608022147641,0.000474500695441070,0.000456016012191429,0.000452992804030727,0.000447008353698699,0.000442436060154468,0.000429780785250373,0.000415298807303484,0.000413535174192959,0.000405719045548631,0.000391300610860691,0.000380704557441276,0.000376823742965357,0.000366443461994963,0.000360427105607389,0.000356257553553216,0.000343963017867418,0.000339594141435343,0.000327416262669204,0.000311785071102165,0.000296233652795617,0.000289778595145242,0.000280723531277823,0.000270074021530794,0.000263095315827241,0.000252779736596250,0.000243343133141444,0.000233664551961454,0.000222888038805312,0.000215788801633336,0.000205419463213638,0.000195807888998832,0.000187787179005757,0.000180919580096046,0.000170097584386522,0.000158303917583322,0.000149189914694997,0.000142768041927102,0.000135775822618333,0.000128520745336136,0.000116678582371879,0.000109189924972610,0.000102296436265799,9.70852943037977e-05,8.99677696901637e-05,8.37943404557717e-05,7.84590810288204e-05,6.98675682491148e-05,6.37113507154509e-05,6.02438323624790e-05,5.41515998931824e-05,4.94706771062464e-05,4.59255080687316e-05,4.11370318253974e-05,3.72350614959839e-05,3.26683097946813e-05,2.92831773481897e-05,2.57157730167172e-05,2.34101274978662e-05,2.13581789005654e-05,1.85068535340041e-05,1.62579772348467e-05,1.41084871984134e-05,1.24022267251321e-05,1.09745684720539e-05,9.49371390203570e-06,8.13492590269067e-06,6.92512690619028e-06,5.89814033220041e-06,4.95101609390562e-06,4.10480825875449e-06,3.43481776593233e-06,2.88106330105423e-06,2.35491029065358e-06,1.91646119690768e-06,1.53589565395398e-06,1.25092631588289e-06,1.00505342772475e-06,8.03861393774004e-07,6.64701374447933e-07,5.48189579375301e-07,4.55966023111770e-07,3.90410096705582e-07,3.53826424888033e-07,3.41742399633712e-07,3.45212243845645e-07,3.45394260859671e-07,3.51233748378259e-07,3.71856158341656e-07,3.86058023920602e-07,4.13228272070997e-07,4.34293625655231e-07,4.82681055375979e-07,5.14072929203551e-07,5.26494309026015e-07,5.33654909090267e-07,5.58642050558073e-07,5.70769509206151e-07,5.72148112840766e-07,5.92208682381448e-07,6.00174293994549e-07,6.04443036260513e-07,6.03471315488289e-07,5.98817274790035e-07,5.93448636484945e-07,5.74484831076828e-07,5.67372196217104e-07,5.64618694395925e-07,5.49553802910645e-07,5.36937738816994e-07,5.24375371781841e-07,5.11454367770285e-07,4.77871921869496e-07,4.61715561992907e-07,4.56400082911407e-07,4.49690180876640e-07,4.30204413224156e-07,4.06516943215753e-07,3.86949554790614e-07,3.75085286153522e-07,3.60583674200204e-07,3.41043116702204e-07,3.30067911516630e-07,3.13429906260035e-07,3.03720992821877e-07,2.88634710657772e-07,2.74799653104203e-07,2.67996698449539e-07,2.50877347511189e-07,2.38160059940251e-07,2.24474283108719e-07,2.13162954474516e-07,2.04690547648461e-07,1.97697618138098e-07,1.89927450984560e-07,1.83938643504457e-07,1.72762371458191e-07,1.64211542935680e-07,1.58424950852324e-07,1.48938592041789e-07,1.42641571367666e-07,1.34572556308535e-07,1.29430646707875e-07,1.22186982111589e-07,1.15755851008638e-07,1.10503513257390e-07,1.06783255252658e-07,1.01277532057332e-07,9.43621401994685e-08,8.66895939014789e-08,8.23475942103369e-08,7.97010750236329e-08,7.48238266736953e-08,7.04990670080386e-08,6.52992692480372e-08,6.17550059214575e-08,5.73278496908541e-08,5.31377495252991e-08,5.00311110386191e-08,4.63971148128278e-08,4.33219182688595e-08,3.97138505763498e-08,3.71520554399501e-08,3.37006920901795e-08,3.08539514317643e-08,2.87646697479823e-08,2.63504305446115e-08,2.37999427855522e-08,2.26860013245252e-08,2.07936334332681e-08,1.91024946637942e-08,1.71944675795184e-08,1.62698568387615e-08,1.50620072583190e-08,1.44180531321305e-08,1.35820987957281e-08,1.29454604124832e-08,1.28282120623197e-08,1.23396218153985e-08,1.21356049534891e-08,1.18671309527197e-08,1.20677562631827e-08,1.22190433261332e-08,1.24970185552656e-08,1.23250292614887e-08,1.28894980268185e-08,1.32419481042036e-08,1.32836219702095e-08,1.40207016066353e-08,1.42720190651168e-08,1.43011046999729e-08,1.47083805268668e-08,1.50885941818194e-08,1.51494929643599e-08,1.55160116772730e-08,1.54338856281701e-08,1.59034038815075e-08,1.57449150174302e-08,1.56965233255726e-08,1.54552179687527e-08,1.53323576980381e-08,1.53362533641807e-08,1.49360311905061e-08,1.48058183017036e-08,1.44744893660825e-08,1.40426349594431e-08,1.36678290350521e-08,1.32187808076580e-08,1.31226276165554e-08,1.25431828789748e-08,1.23085838994648e-08,1.17589589882313e-08,1.12252869636442e-08,1.08940473409769e-08,1.05443612856623e-08,1.01453596802462e-08,9.83780973879624e-09,9.44304531580569e-09,9.19552233955886e-09,8.95100578193612e-09,8.60956034442690e-09,8.46217704007130e-09,8.35243881831049e-09,8.22177144249874e-09,8.14217850702200e-09,8.05408736615532e-09,8.03754153554013e-09,8.11531930744293e-09,8.17660998599579e-09,8.20165333874107e-09,8.31174471659749e-09,8.41420071579722e-09,8.70405129860911e-09,8.82685208964075e-09,8.84609415341114e-09,9.12490891707173e-09,9.44775904868302e-09,9.64971357160747e-09,9.72401539043036e-09,1.00187933936491e-08,1.02678582540819e-08,1.05104007125309e-08,1.07460652729190e-08,1.07871011375819e-08,1.13239233386507e-08,1.14653475222328e-08,1.15296232228472e-08,1.18196331828017e-08,1.18536626984965e-08,1.21385096165467e-08,1.20915144977500e-08,1.22568348906121e-08,1.24253158753117e-08,1.25372475115458e-08,1.26193749899074e-08,1.28427184390675e-08,1.27432261895472e-08,1.28462947972924e-08,1.28228797202864e-08,1.29973179406871e-08,1.27921495227504e-08,1.29075425248766e-08,1.28995319056928e-08,1.29709180973061e-08,1.28791251861747e-08,1.27825994117986e-08,1.25591450387766e-08,1.25037161511047e-08,1.27707207778715e-08,1.24419052183699e-08,1.25334990197956e-08,1.25267466697345e-08,1.24055902914037e-08,1.20883572914270e-08,1.20073043632533e-08,1.17584840588113e-08,1.16727319594448e-08,1.17705262750697e-08,1.17615607735210e-08,1.13581827572531e-08,1.13368145754561e-08,1.11384282499791e-08,1.11123604009849e-08,1.06350105487156e-08,1.07806886417339e-08,1.05623384704122e-08,1.04610385989021e-08,1.01872286308010e-08,1.00197613484506e-08,9.80916729595086e-09,9.94106944424440e-09,9.71844196047031e-09,9.50948157130208e-09,9.34881651187146e-09,9.19857009634992e-09,9.08593945946998e-09,8.94666852394253e-09,8.81584705546601e-09,8.47887904490651e-09,8.43763948323309e-09,8.47578608706971e-09,8.20805448408339e-09,8.04959627803338e-09,7.84022642398165e-09,7.88581266629828e-09,7.64107442233025e-09,7.57222855187598e-09,7.36665872514071e-09,7.38253114771021e-09,7.26505053330269e-09,7.07223027787267e-09,6.97021929587131e-09,6.82425127515634e-09,6.89736401900291e-09,6.66992859851545e-09,6.61460954851629e-09,6.51181651707937e-09,6.44508637143702e-09,6.45610332520672e-09,6.33046832788904e-09,6.22176140330527e-09,6.23598448275473e-09,6.23021095482203e-09,6.19798658376121e-09,6.10641243808578e-09,6.06554162741670e-09,6.14635625989937e-09,6.13081229415009e-09,6.11704256352033e-09,6.06576068736526e-09,6.11832492212053e-09,6.20357047068721e-09,6.21936302905819e-09,6.18616289930780e-09,6.19894716262241e-09,6.36989667130753e-09,6.37804191636784e-09,6.46874586300163e-09,6.42734006284377e-09,6.59468354611630e-09,6.70432160192223e-09,6.78514776354703e-09,6.81921858533582e-09,6.83336346702455e-09,7.00721488220545e-09,7.12715360212327e-09,7.19884166791996e-09,7.29298968200818e-09,7.37593907234373e-09,7.58409929532260e-09,7.53692484699357e-09,7.63296993221228e-09,7.74074461662520e-09,7.87269788886191e-09,8.05635194592090e-09,8.05375949411308e-09,8.12503548049438e-09,8.24422667999210e-09,8.31708852186548e-09,8.52765399339300e-09,8.38559918116659e-09,8.58277024682485e-09,8.58890400572100e-09,8.73539821148376e-09,8.61320150986839e-09,8.86869457819456e-09,8.83553720407036e-09,8.84036576883360e-09,8.93678673155361e-09,8.83537435622520e-09,8.66191753336350e-09,8.74620849243848e-09,8.80683997372379e-09,8.84845382469954e-09,8.76622457210192e-09,8.64224715794218e-09,8.66401931722051e-09,8.55409678838542e-09,8.38508967371981e-09,8.41196006075755e-09,8.27836987488605e-09,8.10045081013938e-09,8.09115226926859e-09,7.90874378055291e-09,7.77347770899612e-09,7.64843398071162e-09,7.55662371538550e-09,7.40679467685124e-09,7.24472275254694e-09,7.12644907886771e-09,6.96680505735342e-09,6.84188165505771e-09,6.68927411177299e-09,6.53581434145592e-09,6.41730849205438e-09,6.32676874873490e-09,6.13973202133388e-09,5.99921405689694e-09,5.90849635617375e-09,5.85470756505375e-09,5.70703402873957e-09,5.53367094089815e-09,5.50358603053393e-09,5.45976517378417e-09,5.38562174726508e-09,5.29361162951415e-09,5.22379964705863e-09,5.24084670383661e-09,5.22610720569617e-09,5.20665969878838e-09,5.17826778495070e-09,5.19715915338678e-09,5.28765417450591e-09,5.33341019142194e-09,5.31596831456025e-09,5.39105748724983e-09,5.51586455101943e-09,5.67045124414651e-09,5.73225445695453e-09,5.76281304203984e-09,5.98635003146130e-09,6.19189943243678e-09,6.27766966135523e-09,6.40576792111652e-09,6.64312373074044e-09,6.75145770650152e-09,7.12153363492708e-09,7.07716218402933e-09,7.59847105842650e-09,7.97625278320909e-09,8.21153383547577e-09,8.08484979314034e-09,8.38121173281176e-09,8.47795729324677e-09,8.81052226335894e-09,9.04825340298944e-09,9.08938064330689e-09,9.38435028109522e-09,9.50563544636009e-09,9.73347885002146e-09,9.92784637557983e-09,9.96571831108112e-09,1.05311779049360e-08,1.04903051005866e-08,1.06203705674889e-08,1.08197323533492e-08,1.04997926013933e-08,1.08516466217419e-08,1.08861566269503e-08,1.09685091289644e-08,1.08684882552757e-08,1.09582446029681e-08,1.07842766136097e-08,1.08480377409314e-08,1.07428651399853e-08,1.09526960615159e-08,1.07352632196625e-08,1.06191465323148e-08,1.06236888636880e-08,1.02460398270652e-08,1.02834603855961e-08,1.00371269282241e-08,9.89008060986722e-09,9.54797743998147e-09,9.61963981017928e-09,9.04470260689983e-09,9.07945279478662e-09,8.69997562122384e-09,8.69057447081210e-09,8.26653964572654e-09,8.06627416330421e-09,7.80509806063779e-09,7.51186615411658e-09,7.25840353036627e-09,7.09361384584830e-09,6.82831848518548e-09,6.56243839778335e-09,6.34295305289336e-09,6.04666269223983e-09,5.93682671663685e-09,5.74446355164421e-09,5.54973502276983e-09,5.31688627800498e-09,5.20610701051850e-09,5.08001443598997e-09,5.05519242097039e-09,4.93162440245895e-09,4.78552802877058e-09,4.79345453693508e-09,4.78917127095902e-09,4.79257973561460e-09,4.76655689776503e-09,4.80418929575097e-09,4.93527232381691e-09,5.07231797107576e-09,5.07825585610543e-09,5.27191644419243e-09,5.42054782750199e-09,5.62482196680140e-09,5.80128031119815e-09,6.01830756240572e-09,6.28587818034269e-09,6.62412846808276e-09,6.73340052065807e-09,7.11722751835444e-09,7.35532647050431e-09,7.71416808300485e-09,8.17955239988027e-09,8.39964529178026e-09,8.66542660211900e-09,8.92971638395389e-09,9.53123352138477e-09,9.61954499589397e-09,1.00376030890240e-08,1.02678353165108e-08,1.06551591120823e-08,1.10958254039842e-08,1.11447374735411e-08,1.14744328850929e-08,1.15167847714678e-08,1.19266276784467e-08,1.22362183035677e-08,1.22784121563390e-08,1.23182858107873e-08,1.24762715605983e-08,1.26782535392525e-08,1.29397619399789e-08,1.29797028046751e-08,1.30276880095548e-08,1.27134958989925e-08,1.28926520351367e-08,1.27201683728450e-08,1.26751598660526e-08,1.24890319147621e-08,1.25375848975444e-08,1.22260465221358e-08,1.20334796208192e-08,1.19448287165050e-08,1.18858189534083e-08,1.13985137035275e-08,1.10919126934910e-08,1.08717275062873e-08,1.06618258633244e-08,1.03052339160006e-08,9.90114622220627e-09,9.75503435268087e-09,9.17658172007369e-09,8.66843837636719e-09,8.61348099566140e-09,8.03208697817769e-09,7.67325341787941e-09,7.36643203484623e-09,7.05963267193484e-09,6.75443328963404e-09,6.50977730092499e-09,6.05833288255153e-09,5.88000514723227e-09,5.68359185216403e-09,5.40299237832263e-09,5.12948986417307e-09,4.99264114848373e-09,4.82457852698581e-09,4.71325995827692e-09,4.56436306277787e-09,4.44468687548332e-09,4.45690397514216e-09,4.43844524480389e-09,4.46350425181594e-09,4.43285197193375e-09,4.51301338475480e-09,4.63395465720549e-09,4.78307798514555e-09,4.90559122539794e-09,5.05896769629320e-09,5.30638960773179e-09,5.50620149506042e-09,5.71924053555511e-09,6.02402919224847e-09,6.13690671988587e-09,6.62343066818935e-09,6.83619643296632e-09,7.03556867018822e-09,7.47615935748243e-09,7.79938513696424e-09,8.20461176873868e-09,8.36951360564259e-09,8.87511596941587e-09,8.96317600856258e-09,9.36884377314376e-09,9.44881371016363e-09,9.84826687485888e-09,1.01067505390981e-08,1.04865512538349e-08,1.06742911302100e-08,1.08826804772582e-08,1.07988655370995e-08,1.10489480715200e-08,1.11055704416854e-08,1.10603971025097e-08,1.10018346780983e-08,1.09139009949028e-08,1.09566689303234e-08,1.13830636789298e-08,1.12998127330269e-08,1.12280609118899e-08,1.09986542189594e-08,1.05601915982285e-08,1.01652518484794e-08,1.00009039306147e-08,9.94635035435038e-09,9.87834071074689e-09,9.67543651388748e-09,9.48667508095968e-09,9.52165756328639e-09,8.82192675636523e-09,8.49250115127243e-09,8.07158319281818e-09,7.79319831868151e-09,7.44868801987736e-09,7.06343528506092e-09,6.97039510804949e-09,6.71626396887560e-09,6.72576427001867e-09,6.03171045432870e-09,5.87240298484062e-09,5.68176916797240e-09,5.67641154128319e-09,5.26312509202078e-09,5.04807718443033e-09,4.99790111811965e-09,5.05659178386406e-09,4.81835745270881e-09,4.59081163930200e-09,4.57714811555838e-09,4.57078446230199e-09,4.53998782144989e-09,4.43289735546490e-09,4.40486489459850e-09,4.42317316988629e-09,4.45853802563628e-09,4.40110475476902e-09,4.37467329232380e-09,4.41037858050479e-09,2.21845864714005e-09;];    

    for ci = 1:Nch
        seg = trimTrailNans(segment(ci, :));
        % segment shorter than FFT length - return without computing spectral features
        if length(seg) < NFFT || any(isnan(seg))
            continue;
        end
        % compensate different sampling freq.        
        if(samplingFreq~=24000)
            seg = resample(seg,24000,samplingFreq);
        end

        psd = pwelch(seg, NFFT, NFFT/2, NFFT)';
        npsd = psd / sum(psd);
        fpsd = (0:NFFT/2) * samplingFreq / NFFT;
        
        for fi = 1:length(featureList)
            fn = featureList{fi};
            switch fn
                case 'maxNormPSD'
                    psdVals(ci, fi) = max(npsd);
                case 'stdNormPSD'
                    psdVals(ci, fi) = std(npsd);
                case 'psdP75'
                    psdVals(ci, fi) = prctile(npsd,75);
                case 'psdP90'
                    psdVals(ci, fi) = prctile(npsd,90);
                case 'psdP95'
                    psdVals(ci, fi) = prctile(npsd,95);
                case 'psdP99'
                    psdVals(ci, fi) = prctile(npsd,99);
                case 'maxAbsDiffPSD'
                    psdVals(ci, fi) = max(abs(npsd-meanClnPSD));
                case 'psdF100'
                    fbnd = floor(100*NFFT/samplingFreq); % 100Hz bin
                    psdVals(ci, fi) = max(npsd(2:fbnd));
                     % psdBase feature
                case 'psdBase' % PSD maximum in the range 1-60Hz divided by mean of 1000-3000Hz range
                    denInd = 1000 <= fpsd & fpsd <=3000;
                    fBound = find(fpsd>60,1,'first');
                    psdVals(ci, fi) = max(npsd(2:fBound))/mean(npsd(denInd)); % normalize by the rest of the spectrum
                case 'psdPow' % PSD maximum in the range 60-600Hz divided by mean of the 1000-3000Hz range
                    numInd = 60 < fpsd & fpsd <= 600;
                    denInd = 1000 <= fpsd & fpsd <=3000;
                    psdVals(ci, fi) = max(npsd(numInd))/mean(npsd(denInd));      
                case 'psdFreq'
                    psdVals(ci, fi) = max(npsd)/median(npsd(fpsd<5000)); % max/median of the non-zero part of the spectrum
                case 'psdMaxStep'
                    psdVals(ci, fi) = max(abs(diff(npsd(2:end))));
            end
        end
    end
end

function timeVals = computeTimeDomainFeatures(segment, featureList, samplingFreq)
    % Compute time-domain features
    [Nch, ~] = size(segment);
    timeVals = nan(Nch, length(featureList));

    segmentNorm = segment ./ sqrt(mean(segment.^2, 2, 'omitnan')); % Normalize by RMS
    
    smoothWindowDuration = 0.005;
    warning('off', 'signal:findpeaks:largeMinPeakHeight');
    [numPeaks, meanPeakHeight, peakFreq, peakRMSRatio, avgPeakWidth] = compPowerPeakFeatures(segment, samplingFreq, smoothWindowDuration, false);
    for ci = 1:Nch
        seg = segment(ci, :);
        segNorm = segmentNorm(ci, :);
        for fi = 1:length(featureList)
            fn = featureList{fi};
            switch(fn)
                case 'energyEntrophy5' %
                    fv = compShannonEntropy(seg, 5);
                case 'energyEntrophy15'
                    fv = compShannonEntropy(seg, 15);
                case 'peakToRMS'
                    fv = compPeakToRMS(segNorm);
                case 'HjorthMobility'
                    fv = compHjorthMobility(seg);
                case 'HjorthComplexity'
                    fv = compHjorthComplexity(seg);
                case 'numPeaks'
                    fv = numPeaks(ci);
                case 'meanPeakHeight'
                    fv = meanPeakHeight(ci);
                case 'peakFreq' %
                    fv = peakFreq(ci);
                case 'peakRMSRatio'
                    fv = peakRMSRatio(ci);
                case 'avgPeakWidth'
                    fv = avgPeakWidth(ci);
                case 'energyRatio' %
                    fv = computeEnergyRatio(seg, samplingFreq);
                case 'sparseness'
                    fv = compSparseness(segNorm);
                case 'irregularity'
                    fv = compIrregularityFactor(segNorm);
                case 'zeroUpCrossingPeriod'
                    fv = compZeroUpCrossingPeriod(seg, samplingFreq, smoothWindowDuration);
                case 'sigLen'
                    fv = compSignalLength(segNorm);
            end
            % store to output
            timeVals(ci, fi) = fv;
        end
    end
end

% compute value of the maxCorr feature: small windows, max of correlation
% coefficients between given channel and all other channels
function f=compMaxCorr(seg,samplingFreq)
    % one channel: 0
    if(size(seg,1)==1)
        f=0;
        return;
    end

    WIN=ceil(samplingFreq/20);
    STEP=round(WIN/2);
    N = size(seg,2);
    
    Cp=[];
    for wi=1:ceil(N/STEP)
        wind = (1:WIN)+(wi-1)*STEP;
        wind = wind(wind<=N);
        C = corr(seg(:,wind)');
        
        % coefficients for channels
        Cp(:,end+1) = max(C-eye(size(C)),[],2);
    end
    % overall maximum for each channel
    f = max(Cp,[],2);   
end

% compute value of powDiff feature - amplitude changes in smaller windows
function f = compPowDiff(seg,samplingFreq)
    WLEN = ceil(samplingFreq/20);    
    [Nc,N] = size(seg);
    
    f = nan(Nc,1);
    
    % segment too short
    if(N<2*WLEN)
         return;
    end    
    
    md = mod(N,WLEN);
    if(md)
        % pad with nans
        % seg = [seg nan(1,md)];
        % truncate instead
        seg = seg(:,1:end-md);
    end
    % compute for channels
    for ci=1:Nc
        S = reshape(seg(ci,:),WLEN,[]);
        pow = nanmean(S.^2);
        if(isnan(pow))
            continue;
        end
        
        Pwdv = pow(1:end-1)./pow(2:end); 
        Pwdv(Pwdv<1) = 1./Pwdv(Pwdv<1);

%         %  multi-channel version
%         Pwdv = pow(:,1:end-1)'./Pw(:,2:end)'; Pwdv(Pwdv<1) = 1./Pwdv(Pwdv<1);
%         Pwdv = abs(diff(Pwdv));

        
        f(ci) = max(abs(diff([1 Pwdv])));
    end
end
    
% compute value of the Kolmogorov-Smirnov statistic
function fv = compKS(segment)
    fv=nan(size(segment,1),1);
    for chi=1:size(segment,1)
        chan = trimTrailNans(segment(chi,:));
        if(isempty(chan)) % skip all-nan channel
            continue;
        end
        [~,~,fv(chi),~] = kstest(chan); % statistic from Kolmogorov-Smirnov normality test
    end
end
    
% trim trailling nans (columns with all nans)   
function segment = trimTrailNans(segment)
    isn = isnan(segment);
    % first all-nan column
    [frNr, frNc] = find(mean(isn,1)==1,1,'first'); 
    %nans at the end only? > trim!
    if(~isempty(frNr) && all(all(isn(:,frNc:end)))) % trim if signal contains nans till the end
        segment = segment(:,1:frNc-1);
    end 
end