snabbintroduktion till latex - lthbme.lth.se/fileadmin/biomedicalengineering/staff... ·...

4
Rapporter skrivna i Word… har tyvärr högst varierande format som gör dem svårlästa, och därmed svårbedömd a. Hur åstadkomma en lättläst och strukturerad rapport som ser likadan ut för alla studenter? IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 12, NO. 3, MAY 2008 399 An Efficient Motion-Resistant Method for Wearable Pulse Oximeter Yong-Sheng Yan and Yuan-Ting Zhang, Fellow, IEEE Abstract—Reduction of motion artifact and power saving are crucial in designing a wearable pulse oximeter for long-term telemedicine application. In this paper, a novel algorithm, mini- mum correlation discrete saturation transform (MCDST) has been developed for the estimation of arterial oxygen saturation (SaO2), based on an optical model derived from photon diffusion analysis. The simulation shows that the new algorithm MCDST is more ro- bust under low SNRs than the clinically verified motion-resistant algorithm discrete saturation transform (DST). Further, the exper- iment with different severity of motions demonstrates that MCDST has a slightly better performance than DST algorithm. Moreover, MCDST is more computationally efficient than DST because the former uses linear algebra instead of the time-consuming adaptive filter used by latter, which indicates that MCDST can reduce the required power consumption and circuit complexity of the imple- mentation. This is vital for wearable devices, where the physical size and long battery life are crucial. Index Terms—Blood oxygen saturation, motion artifact, photo- plethysmography (PPG), pulse oximetry, telemedicine, wearable medical device. I. INTRODUCTION W EARABLE medical devices are capable of continuously monitoring an individual’s vital signs in real time. These devices are particularly important to the world’s increasingly aging population, whose health conditions have to be assessed regularly or monitored continuously. In order to make wear- able devices practical, a series of technical problems have to be solved. For example, these devices need to be miniature in size, must possess a user-friendly interface, be efficient in power consumption, and be motion resistant when monitoring the vital signs under normal daily life. The pulse oximeter has been widely used for measuring blood oxygen saturation (SpO2) in clinical situation. Reduction of motion artifact and power saving are crucial in designing a wearable pulse oximeter for long-term telemedicine applica- tion. Actually, a number of attempts have been made in the past decade to improve the accuracy of pulse oximeter when subjects move [1]–[8]. The typical method is based on an independent measure of motion. For example, one or more transducers (e.g., piezo or optical sensors) are employed to record the user’s mo- tion. By assuming that the artifact is a linear addition to the Manuscript received December 6, 2005; revised December 5, 2006. This work was supported in part by the Hong Kong Innovation and Technology Fund, in part by Standard Telecommunications Ltd., and in part by Jetfly Technology Ltd. The authors are with the Joint Research Centre for Biomedical Engineering, Department of Electronic Engineering, The Chinese University of Hong Kong, Hong Kong (e-mail: [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TITB.2007.902173 pulsatile photoplethysmography (PPG) signal, the original sig- nal can be reconstructed from the corrupted signal [6], [7]. This hypothesis is, however, often doubted when inspecting PPG signals under typical artifact-producing forces [8]. This obser- vation drives researchers to develop more realistic models for the PPG signal or the artifact. A recently reported motion-resistant algorithm, developed by Masimo called discrete saturation transform (DST), is able to detect the SpO2 during low perfusion and motion using an adap- tive filter based on the Masimo pulse oximetry model derived from Beer–Lambert law [9], [10]. A number of studies have demonstrated that the technique may have a significantly lower failure rate and a lower false positive alarm rate than conven- tional techniques [9]–[13]. However, some researchers reported that the accuracy of Masimo pulse oximeter deteriorated at deep hypoxia levels (e.g., SaO2 80%) compared to its performance at higher SaO2, like common pulse oximeters [3]. The phenomenon can be partly explained by the results of previous theoretical studies based on the photon diffusion analysis, which suggests that changes in physiological factors (i.e., hematocrit, arterial venous saturation differences, etc.) that affect tissue scattering and absorption can significantly affect the accuracy of the pulse oximeters, espe- cially at low SaO2 levels [14], [15]. The derived pulse oximeter model from photon diffusion theory, which accounts for both effects of tissue scattering and absorption, has been proven to be a valuable tool to predict the response of pulse oximeter. In this paper, based on the model derived from photon dif- fusion analysis, we propose a novel time-efficient algorithm, minimum correlation discrete saturation transform (MCDST), for motion-resistant measurement of SpO2. Section II describes the principle of the conventional pulse oximeter and the pro- posed motion-resistant algorithm MCDST. Section III presents the simulation results of SaO2 estimation by MCDST, as well as DST algorithm. Section IV presents the experiment results. Finally, Section V gives the discussion and conclusion. II. METHODOLOGY A. Conventional Pulse Oximetry The traditional algorithms for estimating SpO2 detect peaks and troughs of the PPG signal in the time domain. Based on the Beer–Lambert law, which relates the optical path length and effective absorbance to the intensity of transmitted light, the relationship between intensity of transmitted light and SpO2 is commonly described as I(λ,t)=I0(λ) exp[(sεHbO2(λ) + (1 s)εHb(λ))cd(t)] (1) 1089-7771/$25.00 © 2008 IEEE 400 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 12, NO. 3, MAY2008 where εHbO2 and εHb are the extinction coefficients of oxy- genated and deoxygenated hemoglobin, ands, c, anddrepresent SpO2, total concentration of hemoglobin, and the optical path length, respectively. By using two light sources—red (RD) and infrared (IR) lights—and calculating a normalized ratio of the ac compo- nent to the dc component for each light source, SpO2 can be computed from the ratio of ratios R, i.e., the normalized ratio of the red to the infrared transmitted light intensity. That is R= (Iac/Idc)RD (Iac/Idc)IR = (IRD/IIR)ac (IRD/IIR)dc (2) and s = εHb(λR) εHb(λIR)R εHb(λR) εHbO2(λR)+[εHbO2(λIR) εHb(λIR)]R . (3) By calculating the ratio of the ac components and the ratio of the dc components of the two light sources, SpO2 can be obtained from every single pulse of a PPG signal. To stabilize the reading, the weighted moving average (WMA) is often used [5], [16]. Typical averaging methods, e.g., the median averaging and standard arithmetic averaging, are applied to every several samples or samples in every several second intervals. B. MCDST Algorithm Although the operation of pulse oximeters can be understood qualitatively using an analysis based on Beer–Lambert law, this approach fails to account for the effects of intense scattering of light in tissue [14], [15]. In this paper, we employ photon diffusion theory to derive the pulse oximeter model relating to the finger motion (see the Appendix). The model has the same form as the Masimo pulse oximetry model [10] except the different ratio of ratios Ra RD=RaS +RvN IR =S +N (4) where RD and IR represent normalized red and infrared light intensity, S and N are the pulsatile component and motion artifact, and Ra and Rv are the ratio of ratios relating to SaO2 and venous blood oxygen saturation (SvO2), respectively. Based on this model, a novel algorithm, MCDST, is proposed using a subspace-based technique for SaO2 estimation from motion-corrupted signals. With the assumption that the pulsatile component (S) is un- correlated with motion artifact (N), the inner product S, Ncan be neglected compared with the inner product S, Sand N, N. Then, the constrained relationship betweenRa andRv could be found roughly by solving the following equations: A11 ≡⟨IR, IR=S, S+N,N+2S, N≈⟨S, S+N,NA22 ≡⟨RD, RD=R2 aS, S+R2 vN,N+2RaRvS, NR2 aS, S+R2 vN,NA12 ≡⟨IR, RD=RaS, S+RvN,N+(Ra +Rv)S, NRaS, S+RvN,N(5) Fig. 1. Illustration of the MCDST spectrum. The local minimums represent SaO2 andSvO2 values corresponding to the ratios Ra andRv. where A11, A22, and A12 can be calculated directly accord- ing to the normalized ac component of received PPG signal, respectively. Solving (5), we get Ra = A22 A12Rv A12 A11Rv . (6) It means that the discrete choice of Ra will simultaneously determine the value of Rv. Choosing the value of ra stepping through a range of 0–100% of SaO2 [the corresponding value of rv can be calculated using (6)], the reference pulsatile signal and noise are generated by RS =rvIRRD RN=raIRRD. (7) With the different pair of reference signals generated, a subspace-based technique was used to identifyRa andRv based on the minimum correlation between S and N. For the two vec- tors or subspaces RS and RN, the angle between them is defined as cos1(RS×RN). The angle is a good indicator of indepen- dence or correlation. For example, if the angle between the two subspaces is small, the two spaces are nearly linearly dependent. In our study, based on the assumption that S and N are inde- pendent, the angle between the two subspaces RS and RN will be nearest to π/2 when ra =Ra or rv =Rv. To stabilize the optimization process, we construct a cost function considering not only the equal time dependence but also the time-delayed dependence between the subspaces T =abs [angle(RS(t), RN(t)) π/2] + τ=1, ..., n abs [angle(RS(t), RN(t +τ)) π/2] . (8) Here, the function angle(x, y) finds the angle between two subspaces specified by the columns of x and y, the parameter τ is selected fromN =1 to N =20 in this study, which is an approximate number of points in one period of PPG signal. As the value of r is chosen discretely according to the limited sat- uration level 0–100%, the optimization problem of minimizing (8) could be easily solved with stepping through all the possi- bilities. The ratios Ra and Rv will occur at the local minimums in the MCDST spectrum (Fig. 1). In summary, the MCDST algorithm comprises the following steps. 1) Collect the red and infrared signals, and calculate the nor- malized pulsatile component of either signal. Så snyggt blir det i LaTeX! Snabbintroduktion till LaTeX LaTeX är… en familj med program designade för att producera konsekventa dokument med professionell typsättning. speciellt bra för att skapa strukturerade dokument, t.ex. rapporter i klinnovation. speciellt bra för dokument med matematiska symboler och ekvationer. mjukvara för gratis nerladdning. allmänt utbredd bland ingenjörer, men även andra grupper. ett mycket använt format för publicering i ingenjörsvetenskap, vare sig det handlar om tidskrifter eller böcker. Nödvändiga komponenter Ett LaTeX-dokument måste innehålla åtminstone följande tre komponenter (allt annat är valfritt, inklusive brödtexten): \documentclass{article} \begin{document} \end{document} Den första satsen talar om för LaTeX typen av dokument som ska behandlas. Reglerna för typsättning beror på den dokumenttyp som valts. Här används article som dokumentklass men andra klasser är möjliga, däribland report, book och letter. LaTeX mjukvara Såväl gratis som kommersiell mjukvara finns tillgänglig. TeXShop för Mac OSX och MikTeX för Windows är populära, och båda kan enkelt laddas ner och användas. OBS! Distributionen är några Gb. Mjukvara för referenshantering ingår vanligtvis, men dess användning är inte nödvändig för att kunna slutföra rapporten i klinnovation.

Upload: others

Post on 23-Apr-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Snabbintroduktion till LaTeX - LTHbme.lth.se/fileadmin/biomedicalengineering/Staff... · Abstract—Reduction of motion artifact and power saving are ... motion artifact and power

Rapporter skrivna i Word…

har tyvärr högst varierande format som gör dem svårlästa, och därmed svårbedömda.

Hur åstadkomma en lättläst och strukturerad rapport som ser likadan ut för alla studenter?

IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 12, NO. 3, MAY 2008 399

An Efficient Motion-Resistant Methodfor Wearable Pulse Oximeter

Yong-Sheng Yan and Yuan-Ting Zhang, Fellow, IEEE

Abstract—Reduction of motion artifact and power saving arecrucial in designing a wearable pulse oximeter for long-termtelemedicine application. In this paper, a novel algorithm, mini-mum correlation discrete saturation transform (MCDST) has beendeveloped for the estimation of arterial oxygen saturation (SaO2 ),based on an optical model derived from photon diffusion analysis.The simulation shows that the new algorithm MCDST is more ro-bust under low SNRs than the clinically verified motion-resistantalgorithm discrete saturation transform (DST). Further, the exper-iment with different severity of motions demonstrates that MCDSThas a slightly better performance than DST algorithm. Moreover,MCDST is more computationally efficient than DST because theformer uses linear algebra instead of the time-consuming adaptivefilter used by latter, which indicates that MCDST can reduce therequired power consumption and circuit complexity of the imple-mentation. This is vital for wearable devices, where the physicalsize and long battery life are crucial.

Index Terms—Blood oxygen saturation, motion artifact, photo-plethysmography (PPG), pulse oximetry, telemedicine, wearablemedical device.

I. INTRODUCTION

W EARABLE medical devices are capable of continuouslymonitoring an individual’s vital signs in real time. These

devices are particularly important to the world’s increasinglyaging population, whose health conditions have to be assessedregularly or monitored continuously. In order to make wear-able devices practical, a series of technical problems have tobe solved. For example, these devices need to be miniature insize, must possess a user-friendly interface, be efficient in powerconsumption, and be motion resistant when monitoring the vitalsigns under normal daily life.

The pulse oximeter has been widely used for measuring bloodoxygen saturation (SpO2) in clinical situation. Reduction ofmotion artifact and power saving are crucial in designing awearable pulse oximeter for long-term telemedicine applica-tion. Actually, a number of attempts have been made in the pastdecade to improve the accuracy of pulse oximeter when subjectsmove [1]–[8]. The typical method is based on an independentmeasure of motion. For example, one or more transducers (e.g.,piezo or optical sensors) are employed to record the user’s mo-tion. By assuming that the artifact is a linear addition to the

Manuscript received December 6, 2005; revised December 5, 2006. This workwas supported in part by the Hong Kong Innovation and Technology Fund, inpart by Standard Telecommunications Ltd., and in part by Jetfly TechnologyLtd.

The authors are with the Joint Research Centre for Biomedical Engineering,Department of Electronic Engineering, The Chinese University of Hong Kong,Hong Kong (e-mail: [email protected]; [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TITB.2007.902173

pulsatile photoplethysmography (PPG) signal, the original sig-nal can be reconstructed from the corrupted signal [6], [7]. Thishypothesis is, however, often doubted when inspecting PPGsignals under typical artifact-producing forces [8]. This obser-vation drives researchers to develop more realistic models forthe PPG signal or the artifact.

A recently reported motion-resistant algorithm, developed byMasimo called discrete saturation transform (DST), is able todetect the SpO2 during low perfusion and motion using an adap-tive filter based on the Masimo pulse oximetry model derivedfrom Beer–Lambert law [9], [10]. A number of studies havedemonstrated that the technique may have a significantly lowerfailure rate and a lower false positive alarm rate than conven-tional techniques [9]–[13].

However, some researchers reported that the accuracy ofMasimo pulse oximeter deteriorated at deep hypoxia levels (e.g.,SaO2 ≤ 80%) compared to its performance at higher SaO2 , likecommon pulse oximeters [3]. The phenomenon can be partlyexplained by the results of previous theoretical studies basedon the photon diffusion analysis, which suggests that changes inphysiological factors (i.e., hematocrit, arterial venous saturationdifferences, etc.) that affect tissue scattering and absorption cansignificantly affect the accuracy of the pulse oximeters, espe-cially at low SaO2 levels [14], [15]. The derived pulse oximetermodel from photon diffusion theory, which accounts for botheffects of tissue scattering and absorption, has been proven tobe a valuable tool to predict the response of pulse oximeter.

In this paper, based on the model derived from photon dif-fusion analysis, we propose a novel time-efficient algorithm,minimum correlation discrete saturation transform (MCDST),for motion-resistant measurement of SpO2 . Section II describesthe principle of the conventional pulse oximeter and the pro-posed motion-resistant algorithm MCDST. Section III presentsthe simulation results of SaO2 estimation by MCDST, as wellas DST algorithm. Section IV presents the experiment results.Finally, Section V gives the discussion and conclusion.

II. METHODOLOGY

A. Conventional Pulse Oximetry

The traditional algorithms for estimating SpO2 detect peaksand troughs of the PPG signal in the time domain. Based onthe Beer–Lambert law, which relates the optical path length andeffective absorbance to the intensity of transmitted light, therelationship between intensity of transmitted light and SpO2 iscommonly described as

I(λ, t) = I0(λ) exp[(−sεHbO2 (λ) + (1 − s)εHb(λ))cd(t)](1)

1089-7771/$25.00 © 2008 IEEE

400 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 12, NO. 3, MAY 2008

where εHbO2 and εHb are the extinction coefficients of oxy-genated and deoxygenated hemoglobin, and s, c, and d representSpO2 , total concentration of hemoglobin, and the optical pathlength, respectively.

By using two light sources—red (RD) and infrared (IR)lights—and calculating a normalized ratio of the ac compo-nent to the dc component for each light source, SpO2 can becomputed from the ratio of ratios R, i.e., the normalized ratioof the red to the infrared transmitted light intensity. That is

R =(Iac/Idc)RD

(Iac/Idc)IR=

(IRD/IIR)ac

(IRD/IIR)dc(2)

and

s =εHb(λR ) − εHb(λIR)R

εHb(λR ) − εHbO2 (λR ) + [εHbO2 (λIR) − εHb(λIR)]R.

(3)By calculating the ratio of the ac components and the ratio

of the dc components of the two light sources, SpO2 can beobtained from every single pulse of a PPG signal. To stabilizethe reading, the weighted moving average (WMA) is often used[5], [16]. Typical averaging methods, e.g., the median averagingand standard arithmetic averaging, are applied to every severalsamples or samples in every several second intervals.

B. MCDST Algorithm

Although the operation of pulse oximeters can be understoodqualitatively using an analysis based on Beer–Lambert law, thisapproach fails to account for the effects of intense scatteringof light in tissue [14], [15]. In this paper, we employ photondiffusion theory to derive the pulse oximeter model relatingto the finger motion (see the Appendix). The model has thesame form as the Masimo pulse oximetry model [10] except thedifferent ratio of ratios Ra

RD = RaS + RvN

IR = S + N (4)

where RD and IR represent normalized red and infrared lightintensity, S and N are the pulsatile component and motionartifact, and Ra and Rv are the ratio of ratios relating to SaO2and venous blood oxygen saturation (SvO2), respectively.

Based on this model, a novel algorithm, MCDST, is proposedusing a subspace-based technique for SaO2 estimation frommotion-corrupted signals.

With the assumption that the pulsatile component (S) is un-correlated with motion artifact (N ), the inner product ⟨S, N⟩can be neglected compared with the inner product ⟨S, S⟩ and⟨N, N⟩. Then, the constrained relationship between Ra and Rv

could be found roughly by solving the following equations:⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

A11 ≡ ⟨IR, IR⟩ = ⟨S, S⟩ + ⟨N,N⟩ + 2⟨S,N⟩≈ ⟨S, S⟩ + ⟨N,N⟩

A22 ≡ ⟨RD, RD⟩ = R2a⟨S, S⟩ + R2

v ⟨N,N⟩ + 2RaRv ⟨S,N⟩≈ R2

a⟨S, S⟩ + R2v ⟨N,N⟩

A12 ≡⟨IR, RD⟩= Ra⟨S, S⟩+ Rv ⟨N,N⟩ + (Ra + Rv )⟨S,N⟩≈ Ra⟨S, S⟩ + Rv ⟨N,N⟩

(5)

Fig. 1. Illustration of the MCDST spectrum. The local minimums representSa O2 and Sv O2 values corresponding to the ratios Ra and Rv .

where A11 , A22 , and A12 can be calculated directly accord-ing to the normalized ac component of received PPG signal,respectively. Solving (5), we get

Ra =A22 − A12Rv

A12 − A11Rv. (6)

It means that the discrete choice of Ra will simultaneouslydetermine the value of Rv .

Choosing the value of ra stepping through a range of 0–100%of SaO2 [the corresponding value of rv can be calculated using(6)], the reference pulsatile signal and noise are generated by

RS = rv IR − RD

RN = ra IR − RD. (7)

With the different pair of reference signals generated, asubspace-based technique was used to identify Ra and Rv basedon the minimum correlation between S and N . For the two vec-tors or subspaces RS and RN, the angle between them is definedas cos−1(RS′ × RN). The angle is a good indicator of indepen-dence or correlation. For example, if the angle between the twosubspaces is small, the two spaces are nearly linearly dependent.In our study, based on the assumption that S and N are inde-pendent, the angle between the two subspaces RS and RN willbe nearest to π/2 when ra = Ra or rv = Rv . To stabilize theoptimization process, we construct a cost function consideringnot only the equal time dependence but also the time-delayeddependence between the subspaces

T = abs [angle(RS(t), RN(t)) − π/2]

+∑

τ =1, ..., n

abs [angle(RS(t), RN(t + τ)) − π/2] . (8)

Here, the function angle(x, y) finds the angle between twosubspaces specified by the columns of x and y, the parameterτ is selected from N = 1 to N = 20 in this study, which is anapproximate number of points in one period of PPG signal. Asthe value of r is chosen discretely according to the limited sat-uration level 0–100%, the optimization problem of minimizing(8) could be easily solved with stepping through all the possi-bilities. The ratios Ra and Rv will occur at the local minimumsin the MCDST spectrum (Fig. 1).

In summary, the MCDST algorithm comprises the followingsteps.

1) Collect the red and infrared signals, and calculate the nor-malized pulsatile component of either signal.

Så snyggt blir det i LaTeX!

Snabbintroduktion till LaTeX

LaTeX är…❖ en familj med program designade för att producera

konsekventa dokument med professionell typsättning.

❖ speciellt bra för att skapa strukturerade dokument, t.ex. rapporter i klinnovation.

❖ speciellt bra för dokument med matematiska symboler och ekvationer.

❖ mjukvara för gratis nerladdning.

❖ allmänt utbredd bland ingenjörer, men även andra grupper.

❖ ett mycket använt format för publicering i ingenjörsvetenskap, vare sig det handlar om tidskrifter eller böcker.

Nödvändiga komponenterEtt LaTeX-dokument måste innehålla åtminstone följande tre komponenter (allt annat är valfritt, inklusive brödtexten):

\documentclass{article}\begin{document}\end{document}

Den första satsen talar om för LaTeX typen av dokument som ska behandlas. Reglerna för typsättning beror på den dokumenttyp som valts. Här används article som dokumentklass men andra klasser är möjliga, däribland report, book och letter.

LaTeX mjukvara

❖ Såväl gratis som kommersiell mjukvara finns tillgänglig.

❖ TeXShop för Mac OSX och MikTeX för Windows är populära, och båda kan enkelt laddas ner och användas. OBS! Distributionen är några Gb.

❖ Mjukvara för referenshantering ingår vanligtvis, men dess användning är inte nödvändig för att kunna slutföra rapporten i klinnovation.

Page 2: Snabbintroduktion till LaTeX - LTHbme.lth.se/fileadmin/biomedicalengineering/Staff... · Abstract—Reduction of motion artifact and power saving are ... motion artifact and power

Avsnitt (sections)Kommandot för att skapa ett avsnitt med numrering är:

\section{sectiontitle} Detta kommando skapar ett separat avsnitt med en titel som har större typsnitt. Kommandot kan förses med information så att hänvisning kan göras till detta avsnitt (korsreferens), se nästa bild.

Om så behövs kan delavsnitt kan skapas med kommandot:

\subsection{subsectiontitle}

Korsreferenser

❖ Refererens till ett numrerat avsnitt görs genom att först lägga till \label{name} efter \section{} där name är en unik mnemonic som används på ett annat ställe i dokumentet för hänvisning: Section \ref{name}.

❖ Vid typsättning ersätts \ref{name} med det nummer som avsnittet kopplat till kommandot \label{name}.

❖ Dokumentet måste typsättas två gånger för att kors-referensen ska bli komplett.

Några matematiska uttryck

❖ Exponent: x^{2a}

❖ Index: x_{2k}

❖ Bråk: \frac{x^n-1}{x-1}

❖ Funktioner: \sin(x),\ln(x)

❖ Summation: \sum_{k=0}^{10} k

❖ Grekiska bokstäver: \alpha,\delta

sin(x), ln(x)

x

n�1x�1

↵, �

P10k=0 k

x2k

x

2a

Inbäddade ekvationer

Ekvationer kan vara inbäddade i texten (“inline equations”), exemplifierat av texten:

Följande samband är välkänt från litteraturen: $E=mc^2$

som i typsatt form blir:

Foljande samband ar valkant fran litteraturen: E = mc2

Numrerade ekvationerVanligare är dock att ekvationer ges sitt eget utrymme (dvs. på egna rader), oftast numrerad:

\begin{equation}\frac{x^n-1}{x-1} = \sum_{k=0}^{n-1}x^k\label{eqSum}\end{equation}

som typsatt blir

Hänvisning till ekvation görs med \eqref{eqSum}.

Formattering av text

❖ Fetstil: \textbf{Hejhopp}

❖ Kursiverad: \textit{Hejhopp}

❖ Punktlistor: \begin{itemize} \item Text för punkt 1 \item Text för punkt 2 \end{itemize}

❖ Var sparsmakad med dessa typer av formattering!

Page 3: Snabbintroduktion till LaTeX - LTHbme.lth.se/fileadmin/biomedicalengineering/Staff... · Abstract—Reduction of motion artifact and power saving are ... motion artifact and power

Figurer

\begin{figure}\begin{center}\includegraphics[width=8.5cm]{figure.eps}\caption{Figurtext…}\label{figExempel}\end{center}\end{figure}

En figur inkluderas i dokumentet med följande kommandon (figuren lagrad i filen “figure.eps”):

Figuren, som är centrerad i kolumnen, har figurtext definierad i \caption{}. Formatet är Encapsulated Postscript (EPS) och kan åstadkommas i de flesta program, däribland Matlab.

Referenser❖ I slutet av dokumentet läggs bibliografin, i följande

exempel med endast en artikel:

❖ En referens hänvisas till med kommandot \cite{reflabel} där reflabel är mnemonic.

\begin{thebibliography}{99}\bibitem{reflabel} A.~Author1, B.~Author2. ``Title of article,'' \textit{Journal}, vol.~65, pp.~1499--1510, 2004.\end{thebibliography}

Rapportmall

❖ Rapportmall finns att ladda ner som används som utgångspunkt för att skriva din rapport!

❖ Mallen visar hur alla de vanliga LaTeX-elementen ska användas.

❖ Vanliga frågor och tillhörande svar (FAQ) ingår.

❖ Det finns gott om information på nätet om LaTeX. Du kan också kontakta Leif S för hjälp.

1

Titel pa projektrapport i klinnovationFornamn Efternamn (BME–XX), Fornamn Efternamn (BME–XX)

Sammanfattning—Sammanfattningen ar kort och koncist ut-formad (maximalt 250 ord), och ger lasaren en snabborienteringom rapportens innehall. Sammanfattningen rymmer bara uppgif-ter som redan ingar i rapporten och far inte hanvisa till ovrigadelar i rapporten. En lamplig struktur pa sammanfattningenar som foljer: kort bakgrund till behandlat problem, syfte medprojektet, var projektet genomforts, kort beskrivning av metod,viktigaste resultat och dess betydelse. Tank pa att nar mansoker pa en rapport eller artikel ar sammanfattningen/abstractetdet som gor att man bestammer sig for om skall fortsattalasa rapporten eller soka vidare pa annat hall. Det ar alltsaviktigt att inkludera alla huvudresultat och beratta vad mankommit fram till. Undvik ocksa formuleringar av typen ”. . . harundersokts”, ”. . . har byggts” eller ”. . . har visats”, som tenderaratt ge sammanfattningen intrycket av en innehallsforteckning itextform. For att ge en uppskattning av hur mycket textmassa 250ord omfattar har till och med sista ordet i meningen utnyttjats153 ord.

I. INTRODUKTION

SYFTET med detta avsnitt ar att fa lasaren att forsta att detar vart besvaret att lasa artikeln eller rapporten. Typiskt

bestar introduktionen av tre delar: en bakgrundsteckning, entes (pastaende) och en agenda.

A. BakgrundsteckningBakgrundsteckningen ar en beskrivning av tidigare arbeten

och forskning inom det aktuella omradet (litteraturstudie). IKlinnovationsrapporten skall avsnittet avspegla att forfattarnaskaffat sig betydande kunskap inom omradet vad galler savalnya metoder som tillhorande viktiga resultat — kunskap somskall formedlas till lasaren pa ett begripligt satt. De kallor somanvants skall tydligt redovisas i form av referenshanvisningar(formatet pa de vanligaste hanvisningarna ar som foljer: tid-skriftsartikel [?] och bok [?]), dar referenslistan naturligt blirplacerad i slutet av rapporten.Det gar ganska latt att fa ihop en sadan historik, det svara

ar att fa den att hanga ihop logiskt och att naturligt leda framtill din tes.

B. TesDet viktigaste avsnittet i hela introduktionen (eller mojligen

hela rapporten) aterfinns lomskt nog i mitten. Det ar darman formulerar syftet med det arbete man presenterar ochvarfor det ar vart att genomfora. Den motiveringen skall foljanaturligt fran bakgrundsteckningen. Exempel: ”Skrivkramp aren akomma som drabbar over 90% av landets studenter. . .har harletts till anvandning av pennor. . . overdrivet skrivande

Inlamnat den 20 januari 2014Emejladress:{adress1, adress2)Teknisk handledare: Namn, institutionKlinisk handledare: Namn, institution

utan vila for handens muskler”. Efter denna beskrivning avproblemet foljer en viktig lank som gor klart var det finnsen brist i dagens kunskap eller teknik. ”Det finns idag ingethjalpmedel som kan begransa uppkomsten av skrivkramp. . . ”vilket leder fram till tesen: ”Vi presenterar har ett fing-erstod. . . ”. Vanligtvis skrivs rapporter av det har slaget i passivform, men det ar inte fel att anvanda ”vi” har for att markeraatt har tar historiken slut, och nu borjar det som vi tillforomradet.I kliniskt orienterade studier ar det vanligt att en hypotes

formuleras och testas. Om sa ar fallet for aktuellt projektskall denna hypotes inkluderas i slutet av detta avsnitt. Ex-empelvis kan hypotesen for det ovanstaende problemet varaatt ”anvandningen av fingerstod under genomforandet av enkurs minskar forekomsten av skrivkramp”. Denna visas da meden undersokning dar en grupp studenter anvander stodet, ochkontrollgruppen inte. En statistisk analys anger hur sannoliktdet ar att stodet fungerar.

C. AgendaDet sista stycket i introduktionen skall ge en kortfattad

beskrivning av hur rapporten ar upplagd for att visa din tes.Hall denna kort, och undvik att gora det till en upprakning.Tank istallet att agendan skall anvandas for att klargora dintes.

D. OvrigtOm sa onskas kan introduktionen delas upp i ett antal

underavsnitt med tillhorande, lampliga rubriker. Observera attunderrubrikerna som anvants har bara ar exempel. Man skriveraldrig ut de tre delarna, utan det ar mer som en tankehjalp narni skriver.Notera att rapportens totala langd skall vara pa 7–8 sidor

(varken kortare eller langre) nar den typsatts i LATEX, dvs.tvaspaltig text i IEEE-format som galler for detta dokument.Kravet pa total langd kan for vissa forfattare innebara att en ur-valsprocess behover tillampas dar de viktigaste resultaten barakan inkluderas i rapporten. En sadan urvalsprocess forkommeri vasentligen alla sammanhang efter universitetsstudierna!Texterna av Landes [?] och Claerbout [?] ar mycket lasvarda

och har tjanat som inspiration till tankarna kring sammanfatt-ning och introduktion.

II. DATAData som analyserats skall beskrivas i detta avsnitt, inklusive

det/de satt med vilket data samlats in. Det kan vara lampligt,och inte minst overskadligt, att sammanfatta de viktigasteegenskaperna hos data med hjalp av en tabell, exemplifierati tabell I.

2

Tabell IDATA FRAN FRISKA OCH HJARTSJUKA.

Grupp 1 Grupp 2

Antal friska 10 10Antal hjartsjuka 10 10Minsta alder (ar) 3 3Hogsta alder (ar) 8 9

Omfattningen pa detta avsnitt varierar fran projekt tillprojekt beroende pa hur avgorande data ar for projektetsgenomforande. Langden pa avsnittet skall i vilket fall inteoverstiga en halv sida.

III. METOD

De metoder som utvecklats eller implementerats inom ra-men for projektet skall beskrivas i detta avsnitt, liksom de-signen av de experiment som eventuellt har utforts. Stravanar att utforma texten sapass detaljerat att lasare med lampligabaskunskaper skall kunna genomfora det beskrivna arbetet ochna samma resultat (om an att denna stravan i de flesta fall farses som just en stravan!). Det ar viktigt att valet av metoderoch genomforda experiment ges bra motivation. Eventuellasamband som finns mellan existerande och nyutvecklade me-toder skall behandlas langre fram i avsnittet Diskussion.Beroende pa projektets karaktar kan detta avsnitt besta av

saval ren text som text som innehaller ett antal valvaldaekvationer, t.ex. som den i (1) nedan. Detta avsnitt, vars langdinte skall overskrida 3 sidor, kan indelas i underavsnitt och,om sa onskas, innefatta blockdiagram eller andra typer avillustrationer.

H(z) =1

1 + a1z−1 + a2z

−2(1)

IV. RESULTATDe resultat som framkommit under arbetets gang redovisas

har pa ett logiskt vis, och inte nodvandigtvis i den ordningde har framkommit under arbetets gang. Resultaten beskrivsi lopande text pa tydligt satt, med lankar till lampligt valdafigurer, diagram och tabeller for okad forstaelse; en sadanlank exemplifieras med Figur 1. Figurer och diagram farbara inkluderas i rapporten om tillstand (copyright) finns foranvandning.Detta avsnitt, vars langd inte skall overskrida 3 sidor, kan

innehalla korta kommentarer till resultaten, medan de meravidlyftiga tolkningarna, implikationerna och spekulationernahanskjuts till diskussionen i nasta avsnitt; slutsatserna franresultaten ska presenteras i ett eget avsnitt langre fram.

V. DISKUSSIONDetta avsnitt, vars langd inte skall overskrida 1 sida, ar

platsen for vidlyftiga tolkningar och spekulationer rorande deresultat som erhallits under arbetets gang. Har kan eventuellasamband mellan existerande och nyutvecklade metoder disku-teras, och fordelar respektive nackdelar med den utvecklademetoden kan diskuteras.

0 1 2 3 4 5

−200

0

200

400

600

800

tid (s)

am

plit

ud

( µ

V )

V3

Figur 1. Exempel pa figur. Det ar viktigt att inte bara axlarna graderas utanatt aven storhet/enhet anges — i detta fallet ”amplitud (µV)” och ”tid (s)”.Undvik att anvanda automatskalning i Matlab!

Har kan vara vart att papeka att de flesta baxnar inforuppgiften att skriva en rapport. Det kan bottna i att man troratt man borjar med sammanfattningen och sedan fortsattermed introduktion, metod etc. Faktum ar att det lattaste ar attborja med metodavsnittet, och sedan arbeta sig ”utat”: resultat,diskussion, introduktion, slutsats. Sammanfattningen bor manvanta med till sist, nar man har hela rapporten ”i huvudet” ochdarfor lattare kan formulera en sammanfattning.

VI. SLUTSATSERDe viktigaste slutsatserna fran projektet redovisas har (6–8

rader). Observera att dessa bor kopplas ihop med den tes manformulerade i introduktionen.

VII. EFTERORDDe personer som, pa nagot satt bidragit, till projektets

genomforande omnamns har. Vidare skall en kort redogorelseav respektive forfattares arbetsinsats inkluderas.

BILAGALATEX kan laddas ner fran

http://www.tug.org/mactex/http://miktex.org/download

for Mac- resp. PC-anvandare (klicka pa den lank du foredraroch folj instruktionerna for installation). Information om hurman skriver i LATEX finns i Per Forebys handledning(DDG-gruppen pa LTH).Detta dokument kan och ska naturligtvis anvandas som

utgangspunkt for projektrapporten!

REFERENSER[1] S. Cavalcanti, A. Ciandrini, S. Severi, F. Badiali, S. Bini, A. Gattiani,

L. Cagnoli, and A. Santoro. “Model-based study of the effects of thehemodialysis technique on the compensatory response to hypovolemia,”Kidney Int., vol. 65, pp. 1499–1510, 2004.

[2] E. Niedermeyer and F. Lopes da Silva, Electroencephalography. BasicPrinciples, Clinical Applications and Related Fields. Baltimore: Willi-ams & Wilkins, 1998.

[3] K. K. Landes, A scrutiny of the abstract, AAPG Bulletin, vol. 35,p. 1660, 1951.

[4] J. Claerbout. ”Scrutiny of the introduction”, Geophysics, vol. 10, pp. 39–41, 1991.

Använd denna mall som

utgångspunkt

för att skriva din rapport!

En snabbtitt i LaTeX-filen…

Proceedings of Clinical InnovationBachelor of Science Projects in Biomedical Engineering, Lund UniversityVolume 1, 2014EDITORS: TOMAS JANSSON, LEIF SÖRNMO

9 7

8916

2 8

7740

8

Educational Report Series, Biomedical Engineering 2012:00

ISBN 000-00-00000-00-0ISSN 0000-0000

Proceedings of Clinical InnovationThis volume is the first in a series where Bachelor of Science project re-ports from the Biomedical Engineering programme at Lund University are collected. The Bachelor of Science projects in the present volume have all been conducted during the spring of 2014, in the format of a course, also including a lecture series on topics relevant project development, such as sustainability, ethics, and views from industry and the health care sector. Seventeen projects are presented, ranging on topics from retinal implants to app development. It is our hope that this will be the starting point for an increased interaction between on one hand, the Engineering faculty, and on the other the health care sector and the Medical faculty at Lund University

Printed by Tryckeriet i E-huset, LTH, Lund, Sw

eden

PROCEED

ING

S OF CLIN

ICAL INN

OVATIO

N Bachelor of Science Projects in Biom

edical Engineering

1

Page 4: Snabbintroduktion till LaTeX - LTHbme.lth.se/fileadmin/biomedicalengineering/Staff... · Abstract—Reduction of motion artifact and power saving are ... motion artifact and power

Hört om LaTeX från BME-studenter

❖ “Enligt vår handledare är LaTeX utdött”

❖ “Vi fick helt oförklarliga fel när vi typsatte vårt LaTeX-dokument”

❖ “Latex-föreläsningen gav ingenting”

❖ “Kommer dock ihåg föreläsningen om rapportskrivning vilken var bra”, “Det jag kommer ihåg av föreläsnings-serien var bra, t.ex. föreläsningen om LaTex” :)

Lycka till med rapportskrivandet!Innehållet är det viktigaste.