CLINICAL INVESTIGATION Breast TARGET VOLUME DELINEATION FOR PARTIAL BREAST RADIOTHERAPY PLANNING:CLINICAL CHARACTERISTICS ASSOCIATED WITH LOWINTEROBSERVER CONCORDANCER OSS P.P ETERSEN,B.S C.,*y P AULINE T.T RUONG,M.D.,C.M.,*y H OSAM A.K ADER,M.B.,B.S.,*yE RIC B ERTHELET,M.D.,*y J UNELLA C.L EE,B.S C.,*y M ICHELLE L.H ILTS,P H.D.,*zA DAM S.K ADER,B.S C.,*y W AYNE A.B ECKHAM,P H.D.,*z AND I VO A.O LIVOTTO,M.D.*y*Radiation Therapy Program,British Columbia Cancer Agency,Vancouver Island Centre,y University of BritishColumbia,and z University of Victoria,Victoria,BC,CanadaPurpose:To examine variability in target volume delineation for partial breast radiotherapy planning and evalu-ate characteristics associated with low interobserver concordance.Methods and Materials:Thirty patients who underwent planning CT for adjuvant breast radiotherapy formed thestudy ing a standardized scale to score seroma clarity and consensus contouring guidelines,three radi-ation oncologists independently graded seroma clarity and delineated seroma volumes for each case.Seroma geo-metric center coordinates,maximum diameters in three axes,and volumes were recorded.Conformity index(CI),the ratio of overlapping volume and encompassing delineated volume,was calculated for each case.Cases withCI#0.50were analyzed to identify features associated with low concordance.Results:The median time from surgery to CT was42.5days.For geometric center coordinates,variations from themean were0.5–1.1mm and standard deviations(SDs)were0.5–1.8mm.For maximum seroma dimensions,var-iations from the mean and SDs were predominantly<5mm,with the largest SDs observed in the medial–lateralaxis.The mean CI was0.61(range,0.27–0.84).Five cases had CI#0.50.Conformity index was significantly asso-ciated with seroma clarity(p<0.001)and seroma volume(p<0.002).Features associated with reduced concor-dance included tissue stranding from the surgical cavity,proximity to muscle,dense breast parenchyma,andbenign calcifications that may be mistaken for surgical clips.Conclusion:Variability in seroma contouring occurred in three dimensions,with the largest variations in themedial–lateral axis.Awareness of clinical features associated with reduced concordance may be applied towardtraining staff and refining contouring guidelines for partial breast radiotherapy trials.Ó2007Elsevier Inc.Breast cancer,Partial breast radiotherapy,Target volume,Interobserver variability,Conformity index.INTRODUCTIONProspective randomized trials are in progress to evaluate the efficacy of adjuvant partial breast radiotherapy(PBRT)after breast-conserving surgery in patients with early-stage breast cancer.The hypothesis that irradiation of the postsurgical excision cavity with a margin of adjacent breast tissue offers equivalent local control compared with standard whole breast radiotherapy is based on studies demonstrating that approxi-mately70–80%of in-breast recurrences occur at or near the tumor bed(1–4).Radiation treatment targeted to a limited volume using accelerated fractionation schedules over shorter durations has the potential to reduce normal tissue toxicities and improve convenience for breast cancer patients.Currently available PBRT modalities include interstitial brachytherapy(5–9),MammoSite balloon catheter systems(10–12),computerized tomography(CT)-based con-formal external beam techniques(13–15),proton beams(16), and intraoperative electron beam therapy(17).Regardless of the PBRT modality of choice,accurate delineation of target volumes is critical because these volumes provide the basis for all subsequent planning decisions.A potential caveat in PBRT planning is interobserver inconsistencies in delineating the target volume.Although the optimal target volume for PBRT remains to be estab-lished,most series in the PBRT literature have defined the postoperative seroma or tumor bed as the volume on which subsequent planning and prescription decisions are based (5–19).As margins are added to the seroma volume to create the clinical and planning target volumes,variations in itsReprint requests to:Pauline T.Truong,M.D.,C.M.,British Co-lumbia Cancer Agency,Vancouver Island Centre,2410Lee Ave., Victoria,BC V8R6V5,Canada.Tel:(250)519-5575;Fax:(250) 519-2018;E-mail:ptruong@bccancer.bc.caPresented in part at the2007United Kingdom Radiation Oncol-ogy Conference,March18–21,2007,Edinburgh,Scotland.Conflict of interest:none.Received Nov9,2006,and in revised form Jan27,2007. Accepted for publication Jan31,2007.41Int.J.Radiation Oncology Biol.Phys.,Vol.69,No.1,pp.41–48,2007CopyrightÓ2007Elsevier Inc.Printed in the USA.All rights reserved0360-3016/07/$–see front matterdoi:10.1016/j.ijrobp.2007.01.070delineation may ultimately result in critical differences in the treatment volume.This could potentially undermine the ef-fectiveness of PBRT and compromise the validity of trials’results.Studies evaluating contouring practices in breast cancer (20–22)and other tumor sites(23–27)have shown that var-iability in target volume delineation occurs frequently in radiation treatment planning.Contouring guidelines and computer-based educational tools have been demonstrated to be practical measures that can improve consistency(20, 23–26).Another potential measure to improve consistency is to evaluate geometric parameters and clinical features that may be associated with greater observer variability. Awareness of these characteristics may then be applied to re-fining contouring protocols and in training radiation oncolo-gists and radiation therapy staff who participate in trials of PBRT.The present analysis aimed to evaluate variability in CT-based delineation of the seroma volume for PBRT planning and to describe clinical characteristics associated with low interobserver concordance.METHODS AND MATERIALSPatient cohortThirty patients with early-stage breast cancer treated with breast-conserving surgery who underwent planning CT for adjuvant breast RT at the British Columbia(BC)Cancer Agency,Vancouver Island Centre,between June2002and April2005formed the study cohort. The median age at diagnosis was67years(range,47–80years).All patients had invasive or in situ ductal carcinoma,pathologic tumor size<3cm(median,1.15cm;range,0.2–2.2cm),negative sentinel/ axillary nodes,and clear surgical margins,defined as tumor>2mm from inked margins,after partial mastectomy.Radiotherapy planningAll patients underwent planning CT using a CT scan unit(Hi-Speed FX/I;GE Medical Systems,Milwaukee,WI).The median time from surgery to CT was42.5days(range,20–77days).Patients were imaged in the supine treatment position with the ipsilateral arm abducted using a breast board and arm rest immobilization system. Before CT scanning,the ipsilateral breast tissue and the breast sur-gical scar were marked with lead wire by trained radiotherapy staff. Computed tomography images were obtained using a5-mm inter-slice thickness from above the suprasternal notch to the mid-abdomen.Images were transferred to a treatment-planning system (Varian Eclipse,version7.3.10;Varian Medical Systems,Palo Alto,CA)for contouring.The ipsilateral breast,chest wall,pector-alis major muscle,lungs,and heart were contoured on each slice by an experienced dosimetrist.Seroma clarityA Seroma Clarity Scale was developed to standardize grading of the clarity and ease of delineation of the seroma volume(19).This numeric scale ranged from0to5,with0representing no visible se-roma and5representing a clear,homogenous seroma with sharp boundaries(Fig.1).Three radiation oncologists,each with at least 10years of experience treating breast cancer,served as observer subjects for this analysis.The radiation oncologists independently assigned a seroma clarity score for each case.Means and standard deviations(SDs)in seroma clarity scores were calculated for the study cohort.Seroma volume delineationThe consensus guidelines used in this study to contour the seroma volume have been previously published(20).These guidelines were derived from the PBRT literature and from consensus discussions among BC Cancer Agency radiation oncologists participating in PBRT trials(20).Using these guidelines,each of the three radiation oncologists independently delineated the seroma volume for each case using Varian Eclipse software.The observers were permitted to adjust the CT window settings and magnification views at their discretion.Clinical records,including surgical and pathology re-ports,mammography,and other imaging modalities,were available to each observer.Variations from the means and SDsThe seroma contours were analyzed for general trends by exam-ining seroma volumes,geometric center coordinates,and maximum dimensions in the anterior–posterior,medial–lateral,and superior–inferior axes using the statistics function of the workstation.Seven variables were examined:seroma volume,seroma center coordi-nates in the three axes,and maximum seroma dimensions in the three axes.These measurements were performed for30cases, each contoured by three observers,yielding630measurements for analysis.The means and SDs of each variable were calculated for each patient.The average variation from the mean was calculated for each observer with the corresponding SDs.Individual observer’s seroma volumes were plotted as a function of the mean seroma volumes of the group.Conformity indexTo examine case-specific interobserver variability,we used the Conformity Index(CI),defined by Struikmans et al.(21)as the ratio of overlapping volume to encompassing delineated volume(Figs.2a and2b).A CI of1indicates100%concordance,a CI of0.50indi-cates that the observers agreed on50%of the encompassing delin-eated volume,and a CI of0indicates no concordance(Fig.2a). The CI of the three observers was computed for each case(illustra-tive case,Fig.2b).In cases with CI#0.50,an investigator indepen-dent of the contouring radiation oncologists reviewed each case to identify clinical features that may have contributed to the observed low concordance.Associations between CI and patient age,pathologic tumor size, time interval from surgery to CT,mean seroma clarity score,and mean seroma volume were tested using Pearson correlation statis-tics.Statistical significance was defined as a p value of<0.05.All statistical analyses were performed using commercial software (SPSS11.0,Chicago,Illinois).RESULTSSeroma claritySeroma clarity grading using the Seroma Clarity Scale was reproducible among the observers.The mean(SD)seroma clarity score for the patient cohort was3.74(0.68).In8of 30cases,the scores were the same among all three observers; 16of30cases differed by1point;and the remaining6cases differed by2points.No deviation greater than2points was observed.42I.J.Radiation Oncology d Biology d Physics Volume69,Number1,2007Seroma volumesThe mean seroma volume was 48.7cm 3(range,10.3–189.3cm 3).Figure 3presents the plot of each observer’s con-toured volume relative to the mean volume of the group.No trends were identified in terms of any observer characteristi-cally contouring smaller or larger volumes relative to the mean of the group.Seroma geometric center coordinates and maximum dimensionsTable 1summarizes the average variations from the mean and the SDs for each observer’s identification of the seroma geometric center coordinate.Variations from the mean and SDs were <2mm in all dimensions.The largest SDs in geo-metric center coordinates were noted in the medial–lateral axis (range,1.2–1.8mm).Table 2summarizes the average variations from the mean and standard deviations for the maximum seroma diameters in three axes.Variations from the mean were <5mm in all di-mensions.The SDs in maximum seroma diameters were larg-est in the medial–lateral axis (range,4.3–7.7mm),compared with the superior–inferior axis (range,2.4–3.4mm)or the anterior–posterior axis (range,1.8–3.3mm).CI and characteristics associated with reduced concordanceFigure 4presents the distribution of CIs for the patient cohort.The mean CI was 0.61(range,0.27–0.84).FivecasesFig.1.British Columbia Cancer Agency Seroma Clarity Scale.0=no visible seroma,1=scar/shadow,2=seroma identifi-able but with significant uncertainties,3=seroma identifiable with minor uncertainties,4=seroma easily identifiable,gen-erally homogenous with slightly blurred margin,5=seroma easily identifiable,homogenous with sharp boundaries.Partial breast contouring variability d R.P.P ETERSEN et al .43had CI #0.50.Screen captures of illustrative cases with low CI are presented in Figs.5a–5c.Clinical features associated with reduced concordance in these five cases included the fol-lowing:low seroma clarity score,tissue stranding from the surgical cavity,proximity to the pectoralis muscle,dense breast parenchyma,and presence of benign calcifications that were mistaken for surgical clips.In the Pearson correlation analysis,CI was significantly as-sociated with seroma clarity score (r =0.78,p <0.001)and seroma volume (r =0.55,p <0.002).No significant correla-tion was found between CI and patient age at diagnosis,pathologic tumor size,or time interval from surgery to CT (all p >0.05).DISCUSSIONDespite the proliferation of innovative PBRT techniques and increased efforts to translate technological advances into patient benefits,data addressing the pitfalls of variability in PBRT target volume definition remain sparse.With the im-plementation of any new radiotherapeutic approach,all clini-cians will require training and will undergo a learning curve.This study demonstrated that in PBRT target volume defini-tion,the use of consensus guidelines resulted in good inter-observer concordance for the majority of cases,but there remained specific situations in which higher variability oc-curred.The recognition of clinical features associated with low concordance can serve to raise awareness of potential caveats in PBRT volume contouring.This knowledge can be used to train staff and to refine contouring protocols to improve the consistency of seromacontouring.Fig.2.(a)Conformity index (CI)=ratio of overlapping volume to encompassing delineated volume.Diagrammatic representation of CI 0,0.5,and 1.(b)Illustrative case of seroma contouring performed by threeobservers.Fig.3.Plot of individual observer’s seroma volumes as a function of the mean seroma volumes.The solid line indicates the linear relationship between the mean and the observer’s volumes.Table 1.Seroma geometric center coordinates:average variations from the mean and standard deviations,inmillimeters Observer Medial–lateral Superior–inferiorAnterior–posterior1 1.0(1.8)0.6(0.8)0.5(0.6)2 1.1(1.2)0.5(0.6)0.5(0.5)30.9(1.2)0.5(0.5)0.5(0.5)Values in parentheses are standard deviation.44I.J.Radiation Oncology d Biology d Physics Volume 69,Number 1,2007In the present study’s examination of seroma geometric center coordinates and maximum diameters,the largest SDs were noted in the medial–lateral axis.Potential sources of contouring discrepancies in this axis include seroma vol-umes with indistinct borders and low clarity scores,tissue ex-tensions from the tumor bed,and dense breast parenchyma (Figs.5a–5c).In the examination of cases with low CI,although dense extensions from the tumor bed may occur in any dimension,they were prominent in the medial–lateral di-rection (Fig.5b).The consensus adopted by our institution was that such extensions or stranding likely represented sur-gical disturbance rather than the tumor bed and hence were not to be included in contouring the seroma.However,we acknowledge that uncertainty can arise,and the question of whether all tissue extensions should be included as part of the target volume remains controversial.Another clinical characteristic noted to contribute to greater contouring varia-tions in the medial–lateral axis was the presence of dense breast parenchyma.Dense parenchyma,especially when ad-jacent to the tumor bed,may often appear on CT as lateral densities merging with the tumor bed (Fig.5c).Determina-tion of the boundaries between normal dense breast tissue and the tumor bed can thus be difficult.Objective evaluation of tissue radiodensity using the Hounsfield Unit measure-ment tool in RT planning systems may be helpful in such a situation.Our data demonstrated that although the largest SDs were in the medial–lateral axis,variability in the other axes alsooccurred.A relatively large deviation from the mean of 4mm was noted in the superior–inferior axis for one observer.With a CT interslice thickness of 5mm,the observer may have experienced uncertainty regarding whether an addi-tional contour was warranted in the superior and inferior limits.Studies of various tumor sites,including prostate and lung cancer (27,28),have suggested that the selection of CT slice thickness can impact volume definitions.In a se-ries reporting reproducibility among three radiation oncolo-gists performing CT-based prostate organ contouring in 40cases,the use of a 5-mm CT slice thickness was implicated as a factor contributing to difficulties in localizing the pros-tate apex and poor demarcation between the prostate base and seminal vesicles (27).Another series,which compared the use of 1.25-mm and 5-mm CT slice thicknesses in sub-jects with pulmonary nodules,reported that reduced slice thickness improved detection of small nodules and inter-observer agreement (28).These data support the suggestion that reducing interslice thickness can correspondingly reduce interobserver variability,a strategy that warrants investiga-tion in the setting of PBRT planning.In the present analysis,the anterior–posterior axis was associated with the smallest degree of contouring variability.A potential explanation for this observation may be that the skin and the chest wall provided clear boundaries in this dimen-sion for the majority of cases and that the anterior–posterior dimension is generally the smallest breast dimension on a supine CT scan.Variability in the anterior–posterior axis arose when the seroma abutted the pectoralis major muscle (Fig.5a).There is no need to include the pectoralis muscle in the seroma definition if the surgical margins are clear and cancer has not invaded the muscle.Difficulty in demar-cating the interface between seroma and muscle can lead to a failure to exclude muscle from the target volume and can contribute to greater interobserver discrepancies.Review of the operative and pathology reports may be helpful in these situations to determine the extent of surgery and to verify whether fascia and/or muscle were removed at the time ofTable 2.Maximum seroma dimensions:average variation from the mean and standard deviations,in millimeters Observer Medial–lateral Superior–inferiorAnterior–posterior1 2.7(4.5) 2.6(2.9) 1.8(2.2)2 3.2(4.3) 4.1(3.4) 3.2(3.3)33.4(7.7)3.2(2.4)2.4(1.9)Values in parentheses are standarddeviation.24681012140-0.100.11-0.200.21-0.300.31-0.400.41-0.500.51-0.600.61-0.700.71-0.800.81-0.900.91-1.00Conformity Index# o f P a t i e n t sFig.4.Distribution of conformity indices for the study cohort.Partial breast contouring variability d R.P.P ETERSEN et al .45surgery and to determine margin status.Seromas located more anteriorly can also present challenges in contouring in the anterior–posterior dimension (Fig.5c).For example,if the seroma was very close to the skin surface or was blurred by dense tissue behind the nipple–areolar complex,clarity can be poor,resulting in greater contouring variation.Benign calcifications in the breast were identified as a source of discrepancy in PBRT contouring.In one case,two observers misidentified a benign breast calcification for a surgical clip and included this in their contours,resulting in larger volumes and reduced concordance (Fig.5a).Mea-sures that can reduce this pitfall include careful review of mammographic and CT images of both breasts to look for other benign calcifications and for artifacts associated with metallic markers,verification of operative reports to confirm surgical clip placement,and applying tools measuring radio-density to distinguish calcifications from clips.Radio-opaque surgical clips have been demonstrated to be useful inguidingFig.5.Illustrative cases with low conformity indices.(a)Computed tomography transverse plane showing a seroma abut-ting the pectoralis major muscle (yellow contour)and the presence of a benign breast calcification (arrow).Failure to exclude muscle when contouring the seroma and misidentification of a benign breast calcification as a surgical clip contributed to a low conformity index of 0.46.(b)Computed tomography transverse plane showing a seroma with tissue extension from the core volume.The inclusion of this tissue extension in the seroma definition by one observer reduced the conformity index to 0.50.(c)Computed tomography transverse plane showing a seroma located near the skin surface,with indistinct borders and dense surrounding breast parenchyma.The conformity index was 0.38.46I.J.Radiation Oncology d Biology d Physics Volume 69,Number 1,2007PBRT volume definition(29)but were not used routinely by surgeons referring patients with breast cancer to our center during the era of this study.The pitfalls in seroma delineation identified in our study support the contention that clip place-ment may be of added value in PBRT volume localization. The CI was found to be a useful method to quantify inter-observer variability and to highlight specific cases with low concordance for further analysis.The mean CI of0.61(range, 0.27–0.84)found in the current study of three observers,each contouring30partial breast volumes,was comparable to the mean CI of0.56(range,0.39–0.74),reported by Struikmans et al.(21)in their study offive observers,each contouring18 breast boost volumes.However,distinct from that study’s identification of one observer who consistently contoured smaller volumes with respect to the mean,no characteristic tendency to contour smaller or larger volumes relative to the mean was demonstrated among our subjects.The use of consensus-based contouring guidelines may have contrib-uted to the concordance of the individual’s volumes in the majority of cases.The ongoing randomized trials of PBRT,including the Na-tional Surgical Adjuvant Breast and Bowel Project B39trial, the Canadian Randomized Trial of Accelerated Partial Breast Irradiation,and the European Society of Therapeutic Radiol-ogy and Oncology trial,require that subjects have clearly identifiable excision cavities.The correlations between con-touring concordance and seroma clarity score and seroma volume demonstrated in the present analysis have clinical implications in identifying appropriate PBRT candidates for these trials.A seroma clarity score of3or less may serve to flag cases that require consensus review of applied contours before trial enrolment and randomization.When the seroma volume is small,external review of applied contours may also be helpful to avert potentially large shifts in the planning target volume.The present study did not demonstrate significant associa-tions between CI and patient age,tumor size,or time interval from surgery to CT.The vast majority of subjects in our study cohort were aged>50years with small tumors,and all under-went planning CT within11weeks from surgery.The relative homogeneity in these characteristics and small sample size may have limited the power to detect statistically significant associations between CI and these characteristics.In a study evaluating seroma clarity and volume as a function of time from surgery,Kader et al.(19)applied the same Seroma Clar-ity Scale used in the present study to a series of206women undergoing planning CT after breast-conserving surgery.The mean seroma clarity score was3.4and most stable from3to8 weeks after surgery.Thereafter,the mean seroma clarity score decreased to2.5during weeks9to14,and further declined to1.7beyond14weeks from surgery.Seroma vol-umes also progressively decreased with longer time from sur-gery(19),afinding consistent with a study by Oh et al.(30). These data and the present study’s demonstration of reduced concordance with low seroma clarity and volume support the suggestion that the optimal time to perform planning CT for PBRT candidates is within8weeks from surgery to reduce the likelihood of encountering difficulties in target volume delineation associated with low seroma clarity and small volume.With continuing advances in imaging technology,the application of modalities such as breast ultrasound(31–33) and breast magnetic resonance imaging(33,34)in PBRT planning warrant investigation.Correlative studies to assess relationships between the postoperative seroma and the preoperative breast tumor as defined by mammography, CT,or ultrasound are also warranted because they have the potential to further improve accuracy in localizing the PBRT target volume.CONCLUSIONVariability in seroma contouring can occur in three dimen-sions but was most prominent in the medial–lateral axis. Clinical features associated with reduced interobserver con-cordance included low seroma clarity score,small volume, tissue extension from the surgical cavity,proximity to the pectoralis muscle,dense breast parenchyma,and the pres-ence of benign calcifications.Knowledge of these features may be applied to train staff participating in PBRT trials and to refine contouring guidelines and quality assurance processes for PBRT protocols.REFERENCES1.Clark RM,McCullock PB,Levine MN,et al.Randomized clin-ical trial to assess the effectiveness of breast irradiation follow-ing lumpectomy and axillary dissection for node-negative breast cancer.J Natl Cancer Inst1992;84:683–689.2.Liljgren G,Homberg L,Bergh J,et al.10year results after sec-tor resection with or without postoperative radiotherapy for stage I breast cancer:A randomized trial.J Clin Oncol1999;17:2326–2333.3.Veronesi U,Marubini E,Mariani L,et al.Radiotherapy afterbreast-conserving surgery in small breast carcinoma:Long term results of a randomized trial.Ann Oncol2001;12: 997–1003.4.Smith TE,Lee D,Turner BC,et al.True recurrence vs.new pri-mary ipsilateral breast tumor relapse:An analysis of clinical and pathologic differences and their implications in natural history,prognoses,and therapeutic management.Int J Radiat Oncol Biol Phys2000;48:1281–1289.5.Baglan KL,Martinez AA,Frazier RC,et al.The use of high-dose-rate brachytherapy alone after lumpectomy in patients with early-stage breast cancer treated with breast-conserving therapy.Int J Radiat Oncol Biol Phys2001;50:1003–1011. 6.Arthur DW,Koo D,Zwicker RD,et al.Partial breast brachy-therapy after lumpectomy:Low dose rate and high dose rate experience.Int J Radiat Oncol Biol Phys2003;56:681–689. 7.Vicini FA,Arthur DW.Breast brachytherapy:North Americanexperience.Semin Radiat Oncol2005;15:108–115.8.Vicini F,Winter K,Straube W,et al.A phase I/II trial to eval-uate three-dimensional conformal radiation therapy confined to the region of the lumpectomy cavity for stage I/II breast carci-noma:Initial report of feasibility and reproducibility of radiationPartial breast contouring variability d R.P.P ETERSEN et al.47therapy oncology group(RTOG)study0319.Int J Radiat Oncology Biol Phys2005;63:1531–1537.9.Kuske RR,Winter K,Arthur DW,et al.Phase II trial of brachy-therapy alone after lumpectomy for select breast cancer:Toxic-ity analysis of RTOG95-17.Int J Radiat Oncol Biol Phys2006;65:45–51.10.Keisch M,Vicini F,Kuske RR,et al.Initial clinical experiencewith the MammoSite breast brachytherapy applicator in women with early-stage breast cancer treated with breast-conserving therapy.Int J Radiat Oncol Biol Phys2003;55:289–293. 11.Harper JL,Jenrette JM,Vanek KN,et al.Acute complicationsof MammoSite brachytherapy:A single institution’s initial clin-ical experience.Int J Radiat Oncol Biol Phys2005;61:169–174.12.Dickler A,Kirk MC,Chu J,et al.The MammoSite breast bra-chytherapy applicator:A review of techniques and outcomes.Brachytherapy2005;4:130–136.13.Formenti SC,Truong MT,Goldberg JD,et al.Prone acceleratedpartial breast irradiation after breast-conserving surgery:Pre-liminary clinical results and dose-volume histogram analysis.Int J Radiat Oncol Biol Phys2004;60:493–504.14.Taghian AG,Kozak KR,Doppke KP,et al.Initial dosimetricexperience using simple three-dimensional conformal exter-nal-beam accelerated partial-breast irradiation.Int J Radiat Oncol Biol Phys2006;64:1092–1099.15.Baglan KL,Sharpe MB,Jaffray D,et al.Accelerated partialbreast irradiation using3D conformal radiation therapy(3D-CRT).Int J Radiat Oncol Biol Phys2003;55:302–311.16.Taghian AG,Kozak KR,Katz A,et al.Accelerated partialbreast irradiation using proton beam:Initial dosimetric experi-ence.Int J Radiat Oncol Biol Phys2006;65:1572–1578. 17.Veronesi U,Orecchia R,Luini A,et al.Full-dose intraoperativeradiotherapy with electrons during breast conserving surgery: Experience with590cases.Ann Surg2005;242:101–106. 18.Vicini FA,Kestin LL,Goldstein N.Defining the clinical targetvolume for patients with early-stage breast cancer treated with lumpectomy and accelerated partial breast irradiation:A patho-logic analysis.Int J Radiat Oncol Biol Phys2004;60:722–730.19.Kader HA,Truong PT,Panades M,et al.Is breast seroma theappropriate guide for PTV determination in accelerated partial breast irradiation planning?(Abstr.).Radiother Oncol2004;72(Suppl.1):S6.20.Wong EK,Truong PT,Kader HA,et al.Consistency in seromacontouring for partial breast radiotherapy:Impact of guidelines.Int J Radiat Oncol Biol Phys2006;66:372–376.21.Struikmans H,Warlam-Rodenhuis C,Stam T,et al.Interob-server variability of clinical target volume delineation of glan-dular breast tissue and of boost volume in tangential breast irradiation.Radiother Oncol2005;76:293–299.22.Hurkmans CW,Borger HJ,Pieters RB,et al.Variability in tar-get volume delineation on CT scans of the breast.Int J Radiat Oncol Biol Phys2001;50:1366–1372.23.Tai P,Van Dyk J,Battista JJ,et al.Improving the consistency incervical esophageal target volume definition by special training.Int J Radiat Oncol Biol Phys2002;53:766–774.24.Logue JP,Sharrock CL,Cowan RA,et al.Clinical variability oftarget volume description in conformal radiotherapy planning.Int J Radiat Oncol Biol Phys1998;41:929–931.25.van Sornsen de Koste JR,Senan S,Underberg RW,et e ofCD-ROM-based tool for analyzing contouring variations in involved-field radiotherapy for Stage III NSCLC.Int J Radiat Oncol Biol Phys2005;63:334–339.26.Steenbakkers RJ,Duppen JC,Fitton I,et al.Observer variationin target volume delineation of lung cancer related to radiation oncologist-computer interaction:A‘Big Brother’evaluation.Radiother Oncol2005;77:182–190.27.Berthelet E,Liu MCC,Agranovich A,et puted tomog-raphy determination of prostate volume and maximum dimen-sions:A study of interobserver variability.Radiother Oncol 2002;63:37–40.28.Fischbach F,Knollmann F,Griesshaber V,et al.Detection ofpulmonary nodules by multislice computed tomography:Im-proved detection rate with reduced slice thickness.Eur Radiol 2003;13:2378–2383.29.Weed DW,Yan D,Martinez AA,et al.The validity of surgicalclips as a radiographic surrogate for the lumpectomy cavity in image-guided accelerated partial breast irradiation.Int J Radiat Oncol Biol Phys2004;60:484–492.30.Oh KS,Kong FM,Griffith KA,et al.Planning the breast tumorbed:Changes in the excision cavity volume and surgical scar lo-cation after breast-conserving surgery and whole-breast irradia-tion.Int J Radiat Oncol Biol Phys2006;66:680–686.31.Thompson M,Henry-Tillman R,Margulies A,et al.Hema-toma-directed ultrasound-guided(HUG)breast lumpectomy.Ann Surg Oncol2007;14:148–156.32.Della Sala SW,Pellegrini M,Bernardi D,et al.Mammographicand ultrasonographic comparison between intraoperative radio-therapy(IORT)and conventional external radiotherapy(RT)in limited-stage breast cancer,conservatively treated.Eur J Radiol 2006;59:222–230.33.Li Y,Wang J,Holloway C,et al.Development of an MRI/x-ray/ultrasound compatible marker for pre-operative breast tumour localization.Phys Med Biol2005;50:3349–3360. 34.Tardivon AA,Athanasiou A,Thibault F,et al.Breast imagingand reporting data system(BIRADS):Magnetic resonance imaging.Eur J Radiol2007;61:212–215.48I.J.Radiation Oncology d Biology d Physics Volume69,Number1,2007。
