accurate respiration measurement using dc-coupled continuous-wave radar sensor for motion-adaptive...

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 59, NO. 11, NOVEMBER 2012 3117

Accurate Respiration Measurement UsingDC-Coupled Continuous-Wave Radar Sensor for

Motion-Adaptive Cancer RadiotherapyChangzhan Gu, Student Member, IEEE, Ruijiang Li, Member, IEEE, Hualiang Zhang, Member, IEEE,

Albert Y. C. Fung, Carlos Torres, Steve B Jiang, and Changzhi Li, Member, IEEE

AbstractAccurate respiration measurement is crucial inmotion-adaptive cancer radiotherapy. Conventional methods forrespiration measurement are undesirable because they are eitherinvasive to the patient or do not have sufficient accuracy. In ad-dition, measurement of external respiration signal based on con-ventional approaches requires close patient contact to the physicaldevice which often causes patient discomfort and undesirable mo-tion during radiation dose delivery. In this paper, a dc-coupledcontinuous-wave radar sensor was presented to provide a noncon-tact and noninvasive approach for respiration measurement. Theradar sensor was designed with dc-coupled adaptive tuning archi-tectures that include RF coarse-tuning and baseband fine-tuning,which allows the radar sensor to precisely measure movement withstationary moment and always work with the maximum dynamicrange. The accuracy of respiration measurement with the pro-posed radar sensor was experimentally evaluated using a physicalphantom, human subject, and moving plate in a radiotherapy en-vironment. It was shown that respiration measurement with radarsensor while the radiation beam is on is feasible and the mea-surement has a submillimeter accuracy when compared with acommercial respiration monitoring system which requires patientcontact. The proposed radar sensor provides accurate, noninva-sive, and noncontact respiration measurement and therefore has agreat potential in motion-adaptive radiotherapy.

Index TermsCancer radiotherapy, dc information, dc offset,moving tumor, radar, respiration.

Manuscript received February 15, 2012; revised April 28, 2012; acceptedJune 18, 2012. Date of publication June 29, 2012; date of current versionOctober 16, 2012. This work was supported by the Cancer Prevention ResearchInstitute of Texas under Grant RP120053 and National Instruments. Asteriskindicates corresponding author.

C. Gu is with the Department of Electrical and Computer Engineering, TexasTech University, Lubbock, TX 79409 USA (e-mail: [email protected]).

R. Li is with the Department of Radiation Oncology, Stanford University,Stanford, CA 94305 USA (e-mail: [email protected]).

H. Zhang is with the Department of Electrical Engineering, University ofNorth Texas, Denton, TX 76207 USA (e-mail: [email protected]).

A. Y. C. Fung and C. Torres are with the Department of Radiation On-cology, Southwest Cancer Treatment and Research Center, University MedicalCenter, Lubbock, TX 79417 USA (e-mail: [email protected];[email protected]).

S. B. Jiang is with the Center for Advanced Radiotherapy Technologies andthe Department of Radiation Medicine and Applied Sciences, University ofCalifornia San Diego, La Jolla, CA 92093 USA (e-mail: [email protected]).

C. Li is with the Department of Electrical and Computer Engineering, TexasTech University, Lubbock, TX 79409 USA (e-mail: [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/TBME.2012.2206591

I. INTRODUCTION

CANCER is the second leading cause of deaths in the U.S.According to the American Cancer Society, 1.6 millionnew cancer cases will be diagnosed and about 600 000 peoplewill die from the disease in 2012. Radiation therapy is a majormodality for treating cancer patients. Studies have shown thatan increased radiation dose to the tumor will lead to improvedlocal control and survival rates. However, in many anatomicsites (e.g., lung and liver), the tumors can move significantly(23 cm) with respiration. The respiratory tumor motion hasbeen a major challenge in radiotherapy to deliver sufficient radi-ation dose without causing secondary cancer or severe radiationdamage to the surrounding healthy tissue [1], [2].

Motion-adaptive radiotherapy explicitly accounts for andtackles the issue of tumor motion during radiation dose de-livery, in which respiratory gating and tumor tracking are twopromising approaches. Respiratory gating limits radiation ex-posure to a portion of the breathing cycle when the tumor is ina predefined gating window [3]. Tumor tracking, on the otherhand, allows continuous radiation dose delivery by dynamicallyadjusting the radiation beam so that it follows the real-time tu-mor movement. For either technique to be effective, accuratemeasurement of the respiration signal is required. Conventionalmethods for respiration measurement are undesirable becausethey are either invasive to the patient or do not have sufficientaccuracy. For instance, measurement based on fiducial mark-ers requires an invasive implantation procedure and involvesserious risks to the patient, e.g., pneumothorax for lung cancerpatients [4]. On the other hand, measurement of external res-piration surrogates using infrared reflective marker, spirometer,or pressure belt etc., generally lacks sufficient accuracy to inferthe internal tumor position, because they only provide a pointmeasurement or a numerical index of the respiration [5]. In ad-dition, these devices have to be in close contact with the patientin order to function. This often brings discomfort to the patientand can lead to additional patient motion during dose delivery.To that end, accurate respiration measurement that does not re-quire invasive procedures or patient contact is urgently neededin order to realize the potential of motion-adaptive radiotherapy.

Continuous-wave (CW) radar sensor provides a noncontactand noninvasive approach for respiration measurement [6], [7].Instead of measuring the marker, it directly measures the peri-odic motion of the body, which has better correlation with thelung tumor motion. Moreover, the radar system is insensitive toclothing and chest hair, due to microwave penetration, making

0018-9294/$31.00 2012 IEEE

3118 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 59, NO. 11, NOVEMBER 2012

it better than the existing contact devices that are sensitive to thesurrounding environment. In radar respiration measurement, theradar sensor suffers from dc offset at the RF front-end output,which is mainly caused by the reflections from stationary objectssurrounding the body. The dc offset may saturate or limit thedynamic range of the following stages of baseband amplifiers.To overcome this demerit, ac coupling has been commonly usedin radar sensors. However, due to the high-pass characteristicsof the coupling capacitor, ac coupling leads to significant signaldistortion when the target motion has a very low frequency ora dc component. Respiration is such a motion that is low fre-quency of less than 0.5 Hz and tends to rest for a while at theend of expiration (EOE), i.e., there is a short stationary mo-ment after lung deflation. This is a problem in radar respirationmeasurement. To deal with it, several approaches, such as highRF-LO isolation mixers [8], have been introduced to employdc coupling in radar sensors. However, these approaches are ei-ther cumbersome to implement or do not completely remove dcoffsets and limit the dynamic range of the baseband amplifiers.

In this paper, a dc-coupled CW radar sensor was presentedto provide a noncontact and noninvasive approach for accu-rate respiration measurement. The radar sensor was designedwith dc-coupled adaptive tuning architectures that include RFcoarse-tuning and baseband fine-tuning. The RF tuning was im-plemented using a path of an attenuator and a phase shifter at theRF front end of the radar sensor [9], [10]. It adds a portion of thetransmitter signal to the receiver signal to cancel out most of thedc offset. To further calibrate the remaining dc offset, the base-band fine-tuning architecture was used to adaptively adjust theamplifier bias to the desired level that allows both high gain am-plification and maximum dynamic range at the baseband stage.With the aforementioned dc tuning architectures, the radar sen-sor is able to precisely measure the low-frequency respirationmotions with stationary moment. The radar sensor was tested inthe lab environment to demonstrate its ability of accurate dis-placement measurement that preserves dc information of sta-tionary moment. Moreover, the radar sensor was integrated andtested with the linear accelerator (LINAC) to validate its clinicaluse. In order to achieve accurate radiation beam targeting usingrespiration measurement, a correlation model needs to be builtand validated between the internal tumor target and external res-piration signal. This can be decoupled and is generally treatedas a separate issue from respiration measurement. In fact, var-ious approaches have been explored to infer the internal targetposition from external measurement [11], [12]. Therefore, thisstudy has focused on the accurate measurement of respirationin a noninvasive and noncontact way.

The radar motion-adaptive radiotherapy system will be in-troduced in Section II. Section III discusses the dc-coupledradar sensor with tuning architectures. Experimental evalua-tions would be presented in Section IV. A conclusion will bedrawn in Section V.

II. RADIOTHERAPY SYSTEM WITH RADAR SENSINGThe motion-adaptive radiotherapy system based on radar res-

piration sensing is shown in Fig. 1. The radiotherapy process

Fig. 1. Motion-adaptive radiotherapy based on radar respiration sensing. Theprocess includes two steps: treatment preparation and treatment execution. In-sets: (a) multiple radars, (b) beam-scanning radar, and (c) designed 2.4-GHzminiature radar sensor seating with a quarter.

includes two steps: treatment preparation, which consists ofpatient simulation and treatment planning, and treatment ex-ecution, which delivers radiation dose to the patient. At thepatient simulation stage, the patient and tumor geometrical in-formation is collected through computed tomography (CT) scanand then a 3-D patient model is built for the target tumor andorgans at risk [1], [2]. The patients breathing pattern is alsoexamined at this stage. Those patients who cannot exhibit stablebreathing patterns would be excluded from the motion-adaptiveradiotherapy. Treatment planning is a virtual process that de-signs the patient treatment using the patient model built at thesimulation stage. During the radiotherapy execution stage, amedical LINAC would be working together with two radar sen-sors that dynamically monitor the chest wall and the abdomento provide the real-time motion information. The LINAC couldalso be integrated with a radar sensor having beam-scanningcapability, shown as inset (b) in Fig. 1, which makes it pos-sible to use one radar sensor to simultaneously measure thebreathing motions at multiple body locations [13]. In the thirdstep, the advanced tumor tracking algorithm combines the chestwall and abdomen motion information together with the pre-collected patient model to extract the tumor locations in realtime. Then, a controller utilizes the extracted tumor locationinformation to control the LINAC to either perform gated ra-diotherapy or steer the radiation beam to track the tumor. Theinset (c) shows the designed 2.4-GHz miniature radar sensorwith the size of 5 cm 5 cm. The radar sensor was config-ured with a ZigBee module for wireless data transmission,which allows wireless monitoring of the respiration outside thetreatment room and relieves from a bunch of cables that mayconstrain the radar installation to the LINAC. The radar sen-sor will be working in the arctangent-demodulated microwave


Fig. 2. Mechanism of (a) respiratory-gated radiotherapy and (b) tumortracking.

interferometry mode to achieve a submillimeter motion sensingaccuracy [14].

In the proposed radar radiotherapy system, the patient wouldlie on the treatment couch under the radar sensor and breathenormally. The radar sensor measures the breathing signal andwirelessly transmits it to the laptop with a LabView signal pro-cessing interface running in real time. By looking at the real-timevisual feedback from the laptop screen, or following some kindof audio guidance, the patient would be coached to breathe tomaintain a relatively regular breathing pattern [2].

In respiratory gated radiotherapy, the mechanism of which isshown in Fig. 2(a), the radiation dose is delivered to the tumoronly when it moves into the radiation coverage. When the tumormoves out of the radiation coverage, the radiation is turned off.Gating of the radiation beam can be based on either amplitude orphase of the respiration signal. Duty cycle and residual motionare two important parameters that characterize a gated treatment.Duty cycle is the percentage of time that radiation is turned onduring a breathing cycle. Residual motion is the amount of tumormotion during radiation beam on. In respiratory gating, there isa tradeoff between duty cycle and residual motion. Longer dutycycle means longer radiation exposure to the tumor and shorteroverall treatment time, but it also means larger residual motionof the tumor, which will expose more healthy tissues to theradiation. The duty cycle is determined by the doctor at thetreatment preparation stage, according to specific patients anddifferent treatment strategies.

III. DC-COUPLED RADAR SENSORFig. 3 shows the block diagram of the dc-coupled radar sensor

system with RF coarse-tuning and baseband fine-tuning archi-tectures. The radar sensor was designed with homodyne directconversion architecture. In radar applications, dc offset occurs atthe RF output, which may saturate the amplifiers in the followingbaseband stages or limit their dynamic range performance. Thedc offset mainly comes from: 1) direct coupling from transmit-ter to receiver and the reflections from stationary objects aroundthe patient, which is difficult to deal with since it depends ondifferent patients and different test environments, and 2) circuitimperfections such as the self-mixing of LO and interferers [15].As shown in Fig. 3, capacitors have been commonly placed be-tween the RF output and baseband amplifiers for ac couplingto remove the dc offset. However, the high-pass characteristicsof the coupling capacitor lead to significant signal distortionwhen the target motion has a very low frequency or a dc com-

ponent. Respiration is a motion that has very low frequency andcontains a short period of stationary moment after the lung de-flation. So there is practical need for the radar sensor to be ableto accurately measure breathing motion with dc coupling.

The designed dc-coupled radar sensor eliminates the use ofcoupling capacitors by employing dc tuning architectures thatnot only allow the baseband amplifiers to work with sufficienthigh gain but also preserve the stationary dc information. AnRF coarse-tuning signal path was added at the RF front end ofthe radar sensor to remove most of the dc offset, and a basebandfine-tuning feedback patch was used to calibrate the remainingdc offset and reach the largest dynamic range. The RF coarse-tuning path consists of a voltage-controlled attenuator and avoltage-controlled phase shifter. It collects a portion of the trans-mitter (Tx) signal and adds it to the receiver (Rx) signal. In thisway, by properly tuning the attenuator and the phase shifter, thesignals directly coupled from Tx to Rx and bounced back fromthe stationary objects were canceled out [9]. Therefore, the dcoffset at the mixer output was greatly reduced. However, due tocircuit imperfections, the phase variety of the received signalsand the limited resolution of the attenuator and phase shifter, thefeedback path at the RF front end cannot completely remove allthe dc offsets, which limits the dynamic range of the basebandamplifiers. To deal with this problem, a baseband fine-tuning ar-chitecture was added to adaptively adjust amplifier bias to a levelthat allows both high gain amplification and maximum dynamicrange. As shown in Fig. 3, unlike the conventional ac-coupledradar sensor biased at a fixed dc point, the proposed basebandfine-tuning architecture allows the external power supply to ad-just the biasing level at the baseband amplifier. The basebandamplifiers outputs VQ OUT /VI OUT are dynamically read bythe microcontroller (TI MSP430) that transmits the digitizeddata via a ZigBee node to another node connected to the laptop.Take the Q channel, for example, assuming that the amplifierhas infinite open-loop gain, the dc level of the amplifier outputis

VQ OUT = VT Q +R2R1

(V +Q V Q ) (1)

where VT Q is the tuning voltage from the power supply, VQ +and VQ are the dc offsets of the differential channels, andR2/R1 is the closed-loop gain. Depending on the dc valueof VQ OUT /VI OUT , the laptop configures the power supply(Tektronics PS2521G), through a GPIB interface, to increasethe biasing voltage VT Q /VT I when the output is close to thelower rail, or decrease it when the output is close to the upperrail. This process continues until the amplifier output reachesthe desired dc level that allows both high-gain application andmaximum dynamic range. The GPIB interface can be removedafter the dc tuning process is finished. A wireless tuning loop isalso possible by using the ZigBee communications.

Experiments were carried out to validate the dc calibrationcapability of the proposed dc-coupled radar sensor. In the firstexperiment, the RF coarse-tuning loop was verified to removemost of the dc offset and relieve the voltage level for the base-band fine-tuning. As shown in Fig. 4, the original dc differenceof the differential Q channel at the mixer output was 0.07 V.


Fig. 3. Block diagram of the dc-coupled radar sensor system with RF coarse-tuning and baseband fine-tuning architectures. The radar sensor is able to wirelesslytransmit data and receive command via the ZigBee protocols.

Assume that the baseband amplifier works with a gain of 200,which is necessary to boost the radar signals that are usually veryweak, the amplifier dc output results in 14 V, which saturates theamplifier since the dc level is actually way above the amplifierpower supply of 3.3 V. To adjust it to half of the power supply,baseband fine-tuning would need VT Q = 12.35 VDC to beapplied to the amplifier, as shown in (1). The high tuning volt-age may potentially damage the baseband amplifier. However,after RF coarse-tuning, the dc difference drops significantly to0.005 V. In this case, the baseband fine-tuning only needs VT Q= 0.65 VDC to pull the amplifier dc output to the desired level. Inthe second experiment, the two tuning architectures were com-bined to validate the dc calibration functionality. Fig. 5 showsthe dc offset tuning process. It is seen that both I/Q channelswere originally at the lower end of the amplifier power supplyrails. The coarse-tuning architecture was used to adaptively pullup the I/Q channels step-by-step to reasonable levels that allowfurther baseband fine-tuning. It should be known that the RFoutput I/Q channels are 90 out of phase, and the coarse-tuningcannot compensate for the phase offset for both channels at thesame time. After coarse-tuning, the fine-tuning feedback loopprecisely adjusts the I/Q dc offsets to the desired levels, with afine step of 0.01 V, owing to the high resolution of the powersupply. In Fig. 5, the baseband amplifiers outputs were pre-cisely tuned to 1.65 V, which is half way between the basebandamplifier supply rails, so that the maximum dynamic range isguaranteed. It should be noted that in real application, the tuningprocedure can be made very fast (within a few milliseconds).Fig. 6 shows the radar measurement results before and after afast dc tuning. It is seen that the amplifier was originally af-fected by large dc offset such that the Q channel was totallysaturated on the top and the I channel was partially clampedon the bottom. However, after dc tuning, both channels exhibitsatisfactory amplification on signals. The I/Q channels have dif-ferent amplitudes due to the different residual phase [16].

The proposed dc-coupled radar sensor is able to measure tar-get motions with stationary moment. That is, it can preserve thedc information of the stationary moment. It can also preservethe dc information that comes from the nonlinear cosine expan-sion, which is necessary for accurate arctangent demodulation.In a word, owing to the all-pass characteristic of the dc-coupledstructure, the proposed radar sensor is able to maintain the signal

Fig. 4. DC offset difference of the differential Q channel before and after thecoarse-tuning. Coarse-tuning relieves the headroom requirement for fine-tuning.

Fig. 5. Systems dc offsets were adaptively tuned to the desired level by theprocess of coarse-tuning and fine-tuning.

integrity by preserving all the dc information. In Fig. 7, the ac-coupled radar and the dc-coupled radar were used to measure thesame sinusoidal motion of an actuator. The measurement resultsare shown in a constellation graph to compare the I/Q trajecto-ries. It is seen that the trajectory of the dc-coupled radar matcheswell with the unit circle, while the ac-coupled radar tends to de-viate from the circular arc to form a ribbon-like shape. This isbecause of the radar signal amplitude variation that is caused bythe ac coupling capacitors charging and discharging. Since thedc-coupled radar has a trajectory that is much closer to an idealarc, it can be used for more accurate dc calibration [17], andtherefore leads to more accurate arctangent demodulation. To


Fig. 6. Radar measurement result before and after dc tuning.

Fig. 7. Constellation graph of the signals measured by ac radar and dc radar.Insets show the time-domain radar signals.

Fig. 8. Programmed actuator movement compared with the movements mea-sured by ac-coupled radar and dc-coupled radar in an electronic lab.

verify the dc-coupled radars ability to preserve the dc informa-tion of stationary moment, both the dc radar and ac radar wereused to measure the actuator that was programmed to move si-nusoidally but with a stationary moment between two adjacentcycles. The measurement results were shown in Fig. 8. It is seenthat the dc-coupled radar successfully preserves the stationaryinformation by precisely matching with the programmed actu-ator motion. However, the ac radar measured movement startsto deviate from the ground truth when the stationary momentbegins. This is because the ac coupling capacitors cannot holdthe charge for a long time and they tend to discharge over thestationary moment. The proposed dc-coupled radar sensor is ex-pected to more precisely measure the respiration motion that hasvery low frequency and has a short stationary moment duringits cycle.

Fig. 9. Experimental setups: (a) motion phantom measured by camera andradar. Inset shows the infrared camera, (b) human subject simulating patientbreathing measurement, and (c) Respiratory Gating Platform measured by radar.

IV. EXPERIMENTAL EVALUATIONS AND DISCUSSIONS

The accuracy of the proposed dc-coupled radar sensor for res-piration measurement has been evaluated with physical experi-ments in a radiotherapy environment. During the experiments,the radar sensor was placed on the fixation frame crossed overthe treatment platform, as shown in Fig. 9. The respiration datawere wirelessly transmitted to a ZigBee receiver end connectedto the laptop that was placed outside the treatment room.

Three experiments were carried out to validate the use ofradar respiration sensing for clinical applications in radiother-apy. First, a physical motion phantom performed a sinusoidal-like movement with a 5-s period during each cycle. The radarsensor and the real-time position management (RPM) system(Varian Medical Systems, Palo Alto, CA) were used to mea-sure the same motion phantom on the treatment platform. TheRPM system is a widely used respiration monitoring approachin radiotherapy, which employs an infrared camera to track anexternal reflective marker put on the patients chest or abdomen.An infrared marker was put on top of the phantom in order forthe RPM system to track its motion. The infrared camera wasmounted on the wall and is in the line of sight of the marker.This experiment evaluated the accuracy performance of the radarsensor. The setup of this test is shown in Fig. 9(a). To simulatethe real clinical situation, the treatment radiation was turnedon during the phantom test. Under some ideal situations, theRPM system using an infrared camera and reflective marker canprovide accurate measurement of respiratory signal of a singlepoint, and thus was used as the gold standard to assess the per-formance of our radar sensor. However, because of the complex-ity of human respiration, the internal tumor location cannot be


Fig. 10. Phantom motion measured by radar with the LINAC radiation beamturned on, compared with the same phantom motion measured by Varians RPMsystem.

Fig. 11. (a) Radar measured respiration signal. (b) Generated gating signal.The shaded area indicates the coach reference area, and the red line representsthe reference for gating.

derived from only a point measurement with sufficient accuracy.This requires simultaneous measurement of multiple externalsurrogates, which can be achieved by our radar system. In thesecond experiment, a human subject laid on the treatment plat-form to simulate the patients respiration. The setup is shownin Fig. 9(b). In the third experiment, to further account for theclinically relevant breathing patterns, the Respiratory GatingPlatform (Standard Imaging, Middleton, WI) was used to sim-ulate breathing motions with various amplitudes and periods.The Respiratory Gating Platform simulates breathing motionsfor training, quality assurance, and dose verification in radio-therapy. Radar measured the different motion patterns duringthe experiment. The setup is shown in Fig. 9(c).

The measurement results are illustrated in Fig. 10 for thephantom case. It is seen that the radar sensor measurementmatches very well with that from the RPM system. This demon-strates that the proposed radar sensor is not interfered by thehigh-energy radiation dose and can work compatibly with theLINAC during the radiation dose delivery process. For the ex-periment on the human subject, the radar measured respirationis shown in Fig. 11(a). The subject was coached to dynamicallyadjust his breathing to put the EOE position within the shadedarea, so as to generate reproducible respiration signals, fromwhich, gating signals would be easy to be obtained. The accu-rate measurement of the dc-coupled radar sensor also allows thecoaching reference area to be chosen near the position of end ofinspiration (EOI), depending on specific clinical situations. Theradar measured reproducible respiration signals allow both am-plitude gating and phase gating [18]. The red line in Fig. 11(a)

Fig. 12. Radar measured gating platform motions with amplitudes of 20, 10,and 5 mm, and periods of (a) 4 s, (b) 5 s, and (c) 6 s.

TABLE IACCURACY OF THE RADAR MEASUREMENT FOR VARIOUS MOTION PATTERNS

shows the reference for amplitude gating. The respiration signaltriggers the radiation on once its amplitude falls below the refer-ence line. Fig. 11(b) shows the gating signals with a duty cycleof 40.5%. For the Respiratory Gating Platform measurement,the radar measured data are shown in Fig. 12. The platformwas configured to move at amplitudes of 5, 10, and 20 mm, andwith periods of 4, 5, and 6 s, respectively. These motion pat-terns represent the common breathing parameters in the clinicalradiotherapy. It is seen that the radar sensor was able to accu-rately capture all the respiration motion patterns with variousamplitudes and periods. The radar measurement accuracy forvarious motion patterns is illustrated in Table I. The RMS er-ror was obtained by comparing the radar measured movementwith the actual motion pattern of the gating platform. The mea-surement error is mainly due to the radar system noise, such asquantization noise, electronic noise, and environmental noise,which introduces variations to the measured signal. However,even with noise, it is shown that the proposed radar sensor hasa submillimeter measurement accuracy.


The proposed radar sensor system has the potential applica-tion of real-time tumor tracking for motion-adaptive radiother-apy [3]. As mentioned in Section II, respiratory gating has toleverage the tradeoff between duty cycle and residual tumor mo-tion. However, this demerit is eliminated in the tumor trackingtype of treatment, in which, the radiation beam is always on anddynamically follows the moving tumor in real time, as shownin Fig. 2(b). It is technologically more challenging than respi-ratory gating treatments. For radiotherapy treatments based onLINACs, tumor tracking can be implemented by tracking thetumor motion using a dynamic multileaf collimator that shapesthe radiation beam [3]. In order for the radiation beam to dynam-ically follow the tumor, the location of the tumor must be knownwith high accuracy and in real time. Finding the accurate tumorlocation from external respiration measurement is currently anactive research topic [11] and is beyond the scope of this paper.

V. CONCLUSIONA dc-coupled CW radar sensor has been presented to provide

a noncontact and noninvasive approach for accurate respirationmeasurement in motion-adaptive radiotherapy. The proposedradar sensor is configured with adaptive dc tuning architec-tures and able to precisely measure movements with stationarymoment, such as the respiration motion. The concept of a ra-diotherapy system with radar sensing has been introduced anddescribed in the context of respiratory gating and tumor track-ing for motion-adaptive radiotherapy. The dc-coupled CW radarsensor was designed and its dc tuning capability was tested.Experiments were carried out to validate the radar sensor formeasuring movements with stationary moment. The accuracyof respiration measurement using dc-coupled radar sensor hasbeen experimentally evaluated using both physical phantomand human subject in a radiotherapy environment. It has beenshown that respiration measurement with radar sensor whilethe radiation beam is on is feasible and the measurement hasa submillimeter accuracy when compared with the commercialRespiratory Gating Platform. The proposed radar sensor pro-vides accurate, noninvasive, and noncontact respiration mea-surement and therefore has a great potential in motion-adaptiveradiotherapy.

ACKNOWLEDGMENT

The authors would like to acknowledge the Cancer PreventionResearch Institute of Texas (CPRIT) for the grant support, andNational Instruments/AWR for providing funding support aswell as RF instruments and design software.

REFERENCES

[1] S. B. Jiang, Radiotherapy of mobile tumors, Semin. Radiat. Oncol.,vol. 16, no. 4, pp. 239224, Oct. 2006.

[2] S. B. Jiang, Technical aspects of image-guided respiration gated radiationtherapy, Med. Dosim., vol. 31, no. 2, pp. 141151, 2006.

[3] C. Gu, R. Li, C. Li, and S. B. Jiang, A multi-radar wireless system forrespiratory gating and accurate tumor tracking in lung cancer radiother-apy, in Proc. 33rd Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Boston,MA, Aug. 2011, pp. 417420.

[4] F. Laurent, V. Latrabe, B. Vergier, M. Montaudon, J. Vernejoux, andJ. Dubrez, CT-guided transthoracic needle biopsy of pulmonary nodulessmaller than 20 mm results with an automated 20-gauge coaxial cuttingneedle, Clin. Radiol., vol. 55, pp. 281287, 2000.

[5] C. Li, C. Gu, R. Li, and S. B. Jiang, Radar motion sensing for accuratetumor tracking in radiation therapy, in Proc. IEEE Wireless Microw.Technol. Conf., Clearwater Beach, FL, Apr. 2011, pp. 16.

[6] J. C. Lin, Noninvasive microwave measurement of respiration, Proc.IEEE, vol. 63, no. 10, pp. 15301530, Oct. 1975.

[7] J. C. Lin, Noninvasive and remote wireless sensing of physiologic signa-tures and vital signals, presented at the XVII Argentinean Congr. Bioeng.,Rosario, Argentina, Oct. 2009.

[8] X. Zhao, C. Song, O. Boric-Lubecke, and V. M. Lubecke, DC coupleddoppler radar physiological monitor, in Proc. 33rd Annu. Int. Conf. IEEEEng. Med. Biol. Soc., Boston, MA, Aug. 2011, pp. 19091912.

[9] I. Mostafanezhad and O. Boric-Lubecke, An RF based analog lineardemodulator, IEEE Microw. Wireless Compon. Lett., vol. 21, no. 7,pp. 392394, Jul. 2011.

[10] W. Pan, J. Wang, J. Huangfu, C. Li, and L. Ran, Null point elimina-tion using RF phase shifter in continuous wave Doppler radar system,Electron. Lett., vol. 47, no. 21, pp. 11961198, Oct. 2011.

[11] R. I. Berbeco, S. Nishioka, H. Shirato, and S. B. Jiang, Residual motionof lung tumors in end-of-inhale respiratory gated radiotherapy based onexternal surrogates, Med. Phys., vol. 33, no. 11, pp. 41494156, 2006.

[12] H. Yan, F. Yin, G. Zhu, M. Ajlouni, and J. Kim, The correlation evaluationof a tumor tracking system using multiple external markers, Med. Phys.,vol. 33, no. 11, pp. 40734084, 2006.

[13] C. Gu, Z. Salmani, H. Zhang, and C. Li, Antenna array technology forradar respiration measurement in motion-adaptive lung cancer radiother-apy, in Proc. IEEE Top. Conf. Biomed. Wireless Technol., Netw., SensingSyst., Santa Clara, CA, Jan. 2012, pp. 2124.

[14] S. Kim and C. Nguyen, A displacement measurement technique usingmillimeter-wave interferometry, IEEE Trans. Microw. Theory Tech.,vol. 51, no. 6, pp. 17241728, Jun. 2003.

[15] B. Razavi, Design considerations for direct-conversion receivers, IEEETrans. Circuits Syst. I, vol. 44, no. 6, pp. 428435, Jun. 1997.

[16] A. D. Droitcour, O. Boric-Lubecke, V. M. Lubecke, J. Lin, and G. T.A. Kovac, Range correlation and I/Q performance benefits in single-chip silicon Doppler radars for noncontact cardiopulmonary monitoring,IEEE Trans. Microw. Theory Tech., vol. 52, no. 3, pp. 838848, Mar. 2004.

[17] B. K. Park, O. Boric-Lubecke, and V. M. Lubecke, Arctangent demodu-lation with DC offset compensation in quadrature Doppler radar receiversystems, IEEE Trans. Microw. Theory Tech., vol. 55, no. 5, pp. 10731079, May 2007.

[18] C. Gu, R. Li, C. Li, and S. B. Jiang, Doppler radar respiration mea-surement for gated lung cancer radiotherapy, in Proc. IEEE Top. Conf.Biomed. Wireless Technol., Netw., Sensing Syst., Phoenix, AZ, Jan. 2011,pp. 9194.

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/Description >>> setdistillerparams> setpagedevice

accurate respiration measurement using dc-coupled continuous-wave radar sensor for motion-adaptive...

Documents