Signal Extraction Technology

Size: px

Start display at page:

Download "Signal Extraction Technology"

Marilynn Webb
6 years ago
Views:

1 Extaction Technology Intoduction Masimo SET pulse oximety is a new and fundamentally distinct method of acquiing, pocessing and epoting ateial oxygen satuation and pulse ate. As illustated below, Masimo SET technology enables the of adaptive filtes to be applied to eal-time physiologic monitoing by utilizing popietay techniques to accuately establish a noise efeence in the detected physiologic signal, thus enabling the diect calculation of ateial oxygen satuation and pulse ate. Because it is not bound by a conventional ed ove infaed atio appoach, the Masimo SET system substantially eliminates the poblems of motion atifact, low peipheal pefusion and most low signal-to-noise situations. This geatly extends the utility of SpO2 in high motion, low signal and noise intensive envionments. The supeio pefomance of Masimo SET pulse oximety has been clinically validated. The gaphs below summaize the esults of one such study. Demonstated Accuacy with Adults 1 Demonstated Accuacy with Infants 2 Masimo SET Nellco N-600 GE TuSat Masimo SET Nellco N-395 Philips Vivida (FAST Rev. B 100% 83% % 40 Pecent of Events 60% 40% 20% 0 3% 43% Missed Tue Events* False Alams** This study measued the occuence ate of missed tue events duing 40 low blood oxygen episodes and false alams duing 120 fully oxygenated episodes, both duing conditions of motion. 5% 28% 18% * The failue of the monito to detect a physiological change that should tigge the monito s alam. ** The eoneous activation of the monito s alam without an appopiate tiggeing physiological event. 1 Shah N, Estanol L. Compaison of thee new geneation pulse oximetes duing motion & low pefusion in voluntees. Anesthesiology. 2006; 105: A Hay WW, Rodden DJ, Collins SM, Melaa DL, Hale KA, Fashaw LM. Reliability of conventional and new oximety in neonatal patients. Jounal of Peinatology. 2002; 22: Numbe of Studies Missed Tue Events* Discete Satuation Tansfomation (DST Algoithm IR RD Geneato Adaptive Filte Tial Satuation SpO2 = 95% (ange 1 100% Powe 6 This study measued missed tue events and false alams ove a 28 hou peiod, compaing Masimo SET to othe motion esistant pulse oximety technologies. 2 False Alams* 12 ( v SpO2 ( a Masimo SET's most ful algoithm is DST. All algoithms depend upon assumptions. The moe assumptions, the weake the algoithm. DST makes only one assumption - that ateial blood has a highe oxygenation than venous - making it the most ful pulse oximety algoithm.

2 signal extaction technology Conventional Filtes While pulse oximety is eadily accepted as a standad of cae in the Opeating Room, Recovey Room and most Intensive Cae Units 1, its pefomance in high motion envionments o in patients with low pefusion is substantially less than ideal. The epoted high incidence of false alams due to motion atifact 2 and the inability of conventional pulse oximety systems to povide infomation duing times of cisis have led to its chaacteization as a fai weathe fiend. 3 Confonted with the poblem of motion atifact, false alams and poo "signal to noise" envionments, medical equipment manufactues have utilized band-pass filteing in an attempt to addess these confounding clinical poblems. Band-pass filtes, whethe in analog o digital fom, ae designed to allow only a physiologic window of inteest to pass while ejecting fequencies outside the desied fequency band. With the advent of Digital Pocessing (Digital Filteing, the pefomance of band-pass filteing was impoved, but was still unable to addess the poblem of noise occuing within the bandwidth of inteest. Band-Pass Filteing In-Band Noise This appoach has been widely used in the telecommunications and aeospace industies whee a suitable noise efeence is accessible. Pobes ae utilized to obtain a noise efeence that can then be used in conjunction with an adaptive noise cancelle to extact a desied signal potion fom a composite signal containing both desied and undesied signal potions. The poblem in applying this technique to physiological monitoing is that a noise efeence is aely available. In addition, both the noise and the desied signal vay fom patient to patient and ae quickly and continually changing in tems of fequency, amplitude and phase, even within the same patient. In pulse oximety, the noise efeence signal equied to make an adaptive noise cancelle wok in eal time was unavailable until the advent of Masimo Extaction Technology. Conventional Pulse Oximety The conventional "ed ove infaed" appoach measues the diffeential optical density of ed ( and infaed ( light as pojected though a vascula bed and calculates a atio ( of the optical densities. Utilizing the optical density atio, an ateial oxygen satuation (SpO2 value is empiically epoted based on the atio obtained. Amplitude Ambient Light Electo Lines 660nm 940nm d i Emittes 0.5Hz 5Hz 60Hz Fequency 120Hz Ateial Supply Venous Retun Vascula Bed Photo - Detecto Detecto Capillaies Adaptive Filtes To addess the confounding issue of "in-band" noise, a class of filtes known as adaptive digital filtes has evolved. These filtes take advantage of the fact that the constuction of the filte itself is contained within the memoy of the micopocesso, allowing its multiplication coefficients, symbolized as W 0, W 1,...W n-1, to be changed in eal time, hence alteing the filte's chaacteistic. Thus, the filte can be tuned on the fly. The multiplication coefficients detemine whethe the fequency components of an input signal should be cancelled (e.g., multiplied by zeo o allowed to pass (e.g., multiplied by one. Given that the filte's coefficients can be apidly changed, adaptive filtes deive thei name in thei ability to change thei filteing chaacteistics in esponse to changing in-band noise. The detected physiologic signal is geneally composed of both desied signal (S and undesied signal (N o noise potions. To emove the effects of the undesied signal, some knowledge of the noise chaacteistics, o equivalently its noise efeence (N', must be known. The adaptive filte will adjust its filteing chaacteistics, so that the noise efeence input is tansfomed into an estimate of the undesied signal potion (N^ of the physiologic signal. A subtacte subsequently emoves the undesied signal fom the physiologic signal to yield an estimate of the desied signal potion (S^. The combination compising the adaptive filte and the subtacte is commonly called an adaptive noise cancelle (. N S+N ADAPTIVE ALGORITHM w 0 w 1 w 2 wn-2 wn-1 N Adapative Noise Cancelle ( block diagam S & Basis Fo Measuement: S d + N d S i + N i Ratio ( % SpO2 In the pesence of patient motion, the optical densities of ed and infaed light contain noise potions (N d, N i, theeby falsely alteing the optical density atio and poviding an inaccuate satuation value. Duing peiods of outine patient motion o low pefusion, the noise components within the physiologic signals can be much lage than the desied signals (S d, S i. In these cases, the optical density atio is pimaily detemined by the noise contibutions. This epesents a situation wheeby the noise is simply downing out the desied signal. In a lage noise envionment, conventional wisdom holds that pulse oximety will yield an optical density atio substantially equivalent to "noise ove noise" o a atio of one. This is equivalent to a satuation value of appoximately 82% in most conventional systems. If: N>>S, N d Then: 1 82% SpO2 N i ~ Confonted with the poblems of ovewhelming noise and pevented fom utilizing adaptive digital filtes, pulse oximety manufactues have esoted to managing false alams. This can include extending aveaging times o employing a decision matix to feeze when it decides it has detected motion. If the motion pesists, it epots zeo. 4

3 signal extaction technology Decision Matix Repot New Value Feeze & Repot Old Value Repot Zeo The attempt to teat the "symptom" athe than the coe poblem does not povide clinicians with continuous eal-time infomation and can be uneliable in citical medical situations. Masimo SET Pulse Oximety Masimo Extaction Technology ejects the conventional wisdom and begins with an undestanding that duing patient motion the venous blood, being at a elatively low pessue, is quite susceptible to the local effects of petubation duing motion. Consideing the finge fo example, the venous blood in the vascula bed will be easily defomed duing motion, epesenting a significant souce of in-band noise within the fequency bandwidth of inteest. In addition, the venous blood is a stong absobe of light. Hence, it can epesent a significant contibuto to the total optical density duing motion episodes. Futhemoe, the venous blood satuation is nomally lowe than the ateial blood satuation. This explains why satuation values tend to dop in conventional pulse oximete systems duing episodes of patient motion. Duing outine patient motions (shiveing, waving, tapping, etc., the esulting noise can be quite substantial and can easily ovewhelm a conventional atio based oximety system. Having identified the venous blood as a significant contibuto to noise duing motion, it follows that if the noise efeence coesponding to the venous component could be measued, then an adaptive noise cancelle might be utilized to cancel its contibution. Ateial Supply > 80 mmhg Venous Retun > 20 mmhg & d i Detecto Vascula Bed Capillaies Geneating a Noise The detected physiologic signals in esponse to both ed ( and infaed ( light consist of desied signal potions (S d, S i as well as undesied signal potions (N d, N i. It is commonly undestood in pulse oximety that the desied signal potions ae popotional to one anothe though the ateial optical density atio ( a. This suggests that one should simply subtact the poduct of the ateial optical density atio and the physiologic signal due to infaed light fom the physiologic signal due to ed light. The esultant is a efeence signal that contains only noise potions. This is the noise efeence signal (N. Physiological (S + N If the ateial optical density atio is known, one can easily calculate the noise efeence as just descibed. Howeve, if it wee known, one could simply calculate the ateial oxygen satuation diectly. One would not need to utilize the adaptive noise cancellation pocess. How does one then use the of adaptive filtes and noise efeence signals fo pulse oximety? The answe lies in the Discete Satuation Tansfom algoithm. Discete Satuation Tansfom The Discete Satuation Tansfom algoithm allows one to sepaate and, consequently, calculate the optical density atios that coespond to both the ateial oxygen satuation ( a and an estimate of the venous oxygen satuation ( v. These optical densities ae not known befoehand but ae equied to obtain the appopiate efeence signals fo adaptive noise cancellation. Evey optical density atio, coesponding to the patient s physiological ange (SpO2 = 1% to 100% must be consideed. Theefoe, the DST algoithm not only uses a noise efeence signal, but a whole family of efeence signals. Each efeence signal is used in the adaptive noise cancellation pocess and each yields infomation egading the oxygen satuation content of the physiological signals. If: 1 = S d + N d 2 = S i + N i S 3 a = d S i S d = a S i THEN: Id-[ a ] = [S d + N d ] - [S i a + N i a ] Substituting S i a fo S d, we get: = [S i a + N d ]-[S i a + N i a ] = N d -N i a = N (Noise A family of efeence signals, N'(, is geneated simila to that of a noise efeence signal. The efeence signal, as discussed ealie, is the diffeence between the physiologic signal due to ed light ( and the poduct of an abitay optical density atio ( and the physiologic signal due to infaed light (. Although thee is a family of efeence signals, based on the selected optical density atio, thee ae only thee distinct cases to conside. If one selects an optical density atio that does not coespond to eithe ateial o venous oxygen satuation (Case I, the efeence signal consists of a desied signal potion and an undesied signal potion. In the adaptive noise cancellation pocess, such a signal will not only emove the undesied signal potions of the physiologic signal, but also emove the desied signal potions. When an optical density atio that coesponds to the venous oxygen satuation is selected (Case II, the efeence signal only contains signal potions. Theefoe, the of the adaptive noise cancelle will consist of the undesied signal potions only. Similaly, when an optical density atio that coesponds to the ateial oxygen satuation is selected (Case III, the efeence signal only contains noise potions. Theefoe, the of the adaptive noise cancelle will consist of the desied signal potions only. = S d + N d, S d = a S i, : = S i + N i N d = v N i N ( = - Case I: a, v N ( = ( a - S i + ( v - N i : optical density atio a : ateial optical density atio v : venous optical density atio Venous Noise (N Adaptive Noise (S Ateial Case II: = v N ( v = ( a - v S i Case III: = a N ( a = ( v - a N i Fo each selected value of the optical density atio, the coesponding efeence signal is calculated and subsequently pocessed though an adaptive noise cancelle.

4 signal extaction technology - [ ] = N ( O: SpO2 = 1,2, % When the selected value fo the optical density atio does not coespond to eithe the ateial o the venous oxygen satuation (Case I, the coesponding signal will contain little. When the selected value fo the optical density coesponds to eithe the venous oxygen satuation (Case II o the ateial oxygen satuation (Case III, the signal will contain significant. The of the adaptive noise cancelle epesents the pobability that the selected optical density atio, o its coesponding satuation value, is pesent in the physiologic signal. The o pobability value is plotted fo a seies of consecutive atio values geneating the DST tansfom. Duing peiods of no motion, a singula peak is geneated in the DST tansfom coesponding to the ateial oxygen satuation. The moe inteesting example occus duing peiods of motion when multiple peaks in the DST tansfom may be geneated. Repesentative examples with sample wavefom data ae shown in the next section. Output Powe DST Plot Typical No-Motion Data SpO2 ( a In summay, the pocedue fo detemining the ateial oxygen satuation utilizing Masimo SET pocessing is as follows: 1 Sweep all optical density atios that coespond to oxygen satuations of 1% to 100%. 2 Compute the efeence signal fo each optical density atio. 3 Measue the of the adaptive noise cancelle fo each efeence signal. 4 Identify the appopiate peak in the DST tansfom that coesponds to the ateial oxygen satuation (lagest SpO2 value. N ( The pocedue demonstates anothe impotant featue of Masimo SET pulse oximety. It is able to calculate the ateial oxygen satuation without fist extacting o detemining discete pulses in the physiologic data. Fo Masimo SET pocessing, the satuation algoithm is independent of the pulse ate algoithm. This is a significant distinction between Masimo SET systems and conventional pulse oximety systems whee the ecognition of a clean pulse is a peequisite fo the calculation of accuate ateial oxygen satuation. Anothe advantage of Masimo SET technology is that it can monito ateial oxygen satuation and pulse ate even if the motion stats befoe the pulse oximete is tuned on. It does not equie clean data duing instument stat-up. Discete Satuation Tansfom Example In Figue A, a noise cancellation pocess fo a selected optical density atio coesponding to an oxygen satuation of 36% (Case I is shown. Fo an SpO2 of 36%, the calculated efeence signal (shown in puple is substantially simila to the physiologic signal due to ed light (shown in ed. These signals being simila effectively cancel each othe within the adaptive noise cancelle, yielding an signal with little content in the DST tansfom. SpO2 = 36 % Output Powe ( v SpO2 ( a Figue A: Duing Motion DST plot at SpO2 of 36% In Figue B, a noise cancellation pocess fo anothe optical density atio coesponding to an oxygen satuation of 60% (Case II is shown. As is appaent, the efeence signal (shown in ed looks quite diffeent than the physiologic signal due to ed light (shown in ed. This yields an signal with significant content in the DST tansfom since the noise potion is not cancelled like the example in Figue A. SpO2 = 1, 2, % Adaptive Noise Cancelle ( Plot Possible DST Plot SpO2 = 60 % Output Powe ( v ( a SpO2 Figue B: Duing Motion DST plot at SpO2 of 60%

5 In Figue C, a noise cancellation pocess fo anothe optical density atio coesponding to an oxygen satuation of 95% (Case III is shown. In this paticula instance, the efeence signal is tuly the noise efeence. As is appaent, the noise efeence (shown in blue looks quite diffeent fom the physiologic signal due to ed light (shown in ed. This yields an signal with significant content in the DST tansfom since the signal potion is not cancelled like the examples in Figue A. SpO2 = 95 % Output Powe 2009 Masimo Copoation. All ights eseved. ( v ( a SpO2 Figue C: Duing Motion DST plot at SpO2 of 95% s 1. Eichhon JH. Pulse oximety as a standad of pactice in anesthesia. Anesthesiology Ma;78(3: Molle JT, Pedesen T, Rasmussen LS, Jensen PF, Pedesen BD, Ravlo O, Rasmussen NH, Espesen K, Johannessen NW, Coope JB, et al. Randomized evaluation of pulse oximety in 20,802 patients: I. Design, demogaphy, pulse oximety failue ate, and oveall complication ate. Anesthesiology Ma;78(3: Swan HJC. Retieved July 16, 2007 fom 4. SpO 2 Monitos with OISMART Advanced Pocessing and Alam Management Technology Pulse Oximety Note Numbe 9. Masimo Ameicas tel Masimo info-ameica@masimo.com Masimo Intenational tel info-intenational@masimo.com Close to the Heat S-1009

Signal Extraction Technology

Signal Extraction Technology Technical bulletin Introduction Masimo SET pulse oximetry is a new and fundamentally distinct method of acquiring, processing and reporting arterial oxygen saturation and pulse