validation of automated quantitation of myocardial

Circulation Journal Vol.75, September 2011

Circulation JournalOfficial Journal of the Japanese Circulation Societyhttp://www.j-circ.or.jp

yocardial perfusion imaging (MPI) plays an impor-tant role in the assessment of cardiovascular event risk among patients with coronary artery disease

(CAD).1–8

The roles of systematic scoring systems such as the summed stress score (SSS) generated using MPI in predicting car-diovascular events in patients with known or suspected CAD have been validated.9 Summed scores are also calculated from 123I-labelled β-methyl iodophenyl pentadecanoic acid (BMIPP) single emission computed tomography (SPECT) images, which are also useful for the diagnosis and risk strati-fication of patients with CAD.10–13Assessments of SPECT images are more objective using this scoring system than the standard visual analysis.14

Berman et al reported a close and linear correlation be-tween total perfusion deficit, which is an automated quanti-tative MPI SPECT value based upon normal data files and expert visual analysis. Their automated program was also more reproducible than visual scoring.15 However, these pro-grams have not been validated by standard semi-quantitative visual analysis. Because Asian and North American patients have a different physique,16 automated programs should be specifically validated and developed for these populations.

Nakajima et al developed a database from a Japanese normal population16,17 and Matsuo et al developed a normal BMIPP database.18 We recently developed a new automated program for myocardial SPECT based on this normal data-base called Heart Score View. The present study validates

Received December 20, 2010; revised manuscript received April 18, 2011; accepted May 10, 2011; released online July 12, 2011 Time for primary review: 43 days

Department of Photobiology, Division of Molecular, Cellular Imaging, Hokkaido University Graduate School of Medicine, Sapporo (K.Y.); Department of Cardiology, Shin-Nittetsu Muroran General Hospital, Muroran (T.M.); Second Department of Internal Medicine, Sapporo Medical University, Sapporo (A.H., T.N.); Nihon Medi-Physics Co Ltd, Tokyo (K.T.); Department of Cardiology, Hokkaido Prefectural Esashi Hospital, Esashi (T.N.); Department of Nuclear Medicine, Hokkaido University Graduate School of Medicine, Sapporo (N.T.), Japan

Mailing address: Keiichiro Yoshinaga, MD, PhD, FACC, Associate Professor, Department of Photobiology, Division of Molecular Cellu-lar Imaging, Hokkaido University Graduate School of Medicine, Kita 15 Nishi7, Kita-ku, Sapporo 060-8638, Japan. E-mail: kyoshi@ med.hokudai.ac.jp

ISSN-1346-9843 doi: 10.1253/circj.CJ-10-1272All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: [email protected]

Validation of Automated Quantitation of Myocardial Perfusion and Fatty Acid Metabolism Abnormalities

on SPECT ImagesKeiichiro Yoshinaga, MD, PhD; Takayuki Matsuki, MD, PhD; Akiyoshi Hashimoto, MD, PhD;

Kazumasa Tsukamoto, PhD; Tomoaki Nakata, MD, PhD; Nagara Tamaki, MD, PhD

Background: Myocardial perfusion and fatty acid imaging have played important roles in the risk stratification of patients with coronary artery disease (CAD). However, visual image assessment requires considerable experi-ence and training. Therefore, an automated program has been developed that can quantify perfusion and fatty acid uptake on myocardial single emission computed tomography (SPECT). The present study aimed to validate the automated quantitative program.

Methods and Results: A total of 50 patients were studied with known or suspected CAD who underwent stress 201Thallium (201Tl) and resting 123I-labelled β-methyl iodophenyl pentadecanoic acid (BMIPP) SPECT. The SPECT images were quantified in 17 segments visually and using our Heart Score View software. Values were compared with those in a normal Japanese database and calculated summed stress (SSS), summed rest (SRS), summed difference (SDS), and summed BMIPP scores for each modality. Summed scores obtained using standard visual analysis and Heart Score View significantly correlated (201Tl: SSS: r=0.934; SRS: r=0.827; SDS: r=0.743 summed BMIPP score: r=0.913) (each P<0.001) and Bland-Altman analysis revealed good agreement between the 2 ap-proaches.

Conclusions: Correlations between scores determined using Heart Score View software and standard visual interpretation were linear for both perfusion and fatty acid images. Thus, our new automated program might be useful for the risk stratification of patients with CAD in the clinical setting. (Circ J 2011; 75: 2187 – 2195)

Key Words: Coronary artery disease; Quantitative assessment; Risk stratification

M

ORIGINAL ARTICLEImaging


2188 YOSHINAGA K et al.

Figure 1. (A) Concept of Heart Score View software. (B) Threshold settings based on normal database established by Japanese Society of Nuclear Medicine. Mid A, middle anterior; Mid AL, middle antero-lateral.


2189Quantitative Assessment of Myocardial SPECT Images

the automated program by comparison with standard expert visual interpretations of both 201Thallium (201Tl) and BMIPP myocardial SPECT images.

MethodsStudy PopulationThis retrospective study included 196 consecutive patients with known or suspected CAD who underwent stress 201Tl and rest BMIPP myocardial SPECT at Shin-Nittetsu Muroran General Hospital between January 1995 and December 2000.10 Patients with known myocardial infarction were excluded. We randomly assessed 50 (25 males and 25 females) of these 196 patients. The Ethics Committee of Shin-Nittetsu Muroran General Hospital approved the study protocol.

Myocardial Perfusion and Fatty Acid ImagingThe data acquisition protocols have been described else-where.10 Briefly, patients fasted for >6 h before the studies. Forty-two and 8 patients underwent exercise and pharmaco-logical stress-testing, respectively. The patients who under-went exercise symptom-limited treadmill stress testing were injected with 74 to 111 MBq of 201Tl at peak exercise and then continued the exercise for 1 min. The patients who underwent dipyridamole stress testing were administered with dipyri-damole (0.14 mg · kg–1 · min–1) and then injected with 201Tl at 3 min later.10,19,20 SPECT images were acquired starting from 15 min after tracer injection using a digital gamma camera (Shimazu 510R, Shimadzu CO, Kyoto, Japan). Rest images were acquired 4 h later. Patients were intravenously adminis-tered under both fasting and resting conditions with 111 MBq of BMIPP (Nihon Medi-Physics Co Ltd, Tokyo, Japan).10,12,21 SPECT images were acquired starting from 30 min after tracer injection.

Semi-Quantitative Visual InterpretationSeventeen segments were semi-quantified using the 5-point visual scoring model of tracer uptake (0, normal; 4, absent) recommended by the Cardiac Imaging Committee of the American Heart Association and the American Society of Nuclear Cardiology (ASNC).22 Tracer uptake in each segment was scored by 2 expert independent teams of assessors that were provided with gender information but not with clinical data (Team 1, K.Y. and N.T.; Team 2, T.M., Y.H. and T.N.). SSS, summed rest (SRS), and summed difference (SDS) scores were obtained from 201Tl SPECT myocardial perfu-sion images. A similar scoring system was applied to BMIPP images to generate summed rest BMIPP scores.

We randomly selected SPECT images from 12 patients to evaluate the concordance of scores between the 2 assessment teams.

Automated Quantitation of Myocardial SPECT Images Using Heart Score View SoftwareWe developed an automated program for myocardial SPECT called Heart Score View software (Nihon Medi-Physics Co Ltd, Tokyo, Japan) that was designed for use with the Windows operating system. It has multiple analytical tools including SSS and 201Tl washout, as well as perfusion and fatty acid mismatch. The algorithm for generating a polar map was a hybrid 2-part sampling method23,24 that operates in 3-dimen-sional space and uses short axis images to generate count profiles from a 3-dimensional sampling scheme of short-axis slices. Images were co-registered and fused using a fully automated and mutual information algorithm.25

The software requires manual setting of the left ventricu-lar cavity and then it generates polar maps from myocardial SPECT data that are divided into 17 segments based on ASNC guidelines to calculate the mean count in each segment.22 These mean counts were compared with the normal stress/rest 201Tl, 99mTc MPI, and rest BMIPP database developed for Japanese patients by the Japanese Society of Nuclear Medicine working group.16,17 The mean % uptake in each seg-ment was derived, converted to scores using a 5-point scale ranging from normal to absent and then described on polar maps (Figures 1A,B). The % uptake was usually lower in the basal septum due to the membranous septum. Therefore, we excluded the 6 basal segments when determining thresh-old values.

We based the following definitions for scoring parameters based on various degrees of abnormal perfusion.

Table 1. Clinical Characteristics of the Study Population

Age (years) 66.0±10.3

Gender (male) 25/50 (50.0%)

Unstable angina pectoris 5/50 (10.0%)

LVEF (%) 67.9±9.2　　

Coronary risk factors

Current smoker 15/50 (30.0%)

Diabetes mellitus 10/50 (20.0%)

Hypertension 30/50 (60.0%)

Dyslipidemia 17/50 (34.0%)

LVEF, left ventricular ejection fraction.

Table 2. Profiles of Regression Coefficients of Summed Scores Between Visual Interpretation and Heart Score View Analysis

ScoreDefinition 1 Definition 2 Definition 3

R Equation R Equation R Equation

SSS 0.953　 Y=1.554x+0.415 0.934 Y=1.065x–0.447 0.852　 Y=0.691x–1.712

SRS 0.906* Y=1.448x–0.001 0.827 Y=0.915x–0.236 0.716* Y=0.537x–0.769

SDS 0.850* Y=1.388x+0.569 0.743 Y=0.903x+0.187 0.593* Y=0.437x+0.175

Summed BMIPP score 0.953　 Y=0.660x–0.227 0.913 Y=0.916x+0.016 0.882　 Y=0.673x–0.860

*Significant difference in regression coefficients for SRS between thresholds 1 and 3, P<0.01 as well as in SDS between definitions 1 and 3, P<0.01. NS between thresholds 1 and 2.Equations: y, scored by visual interpretation; x, scored by Heart Score View calculation (SSS, SRS, and SDS obtained from 201Tl SPECT images).R, regression coefficient; SSS, summed stress score; SRS, summed rest score; SDS, summed difference score; BMIPP, 123I-labelled β-methyl iodophenyl pentadecanoic acid; 201Tl, 201Thallium; SPECT, single photon emission computed tomography.



A

B

C

D

Figure 2. Correlations and Bland-Altman plot of data generated from visual interpretation (VI) and Heart Score View software using 201Thallium (201Tl) MPI. (A) Summed stress scores (SSS). (B) Summed rest scores (SRS). (C) Summed difference scores (SDS). (D) Representative images of 66-year-old diabetic woman with effort angina pectoris. 201Tl images show moderately sized, moderate intensity of perfusion defect in inferolateral wall that improved at rest, indicating moderate ischemia in LCX territory. Visual and automated SSS were 9 and 8, respectively.



Definition 1 All segments with a mean uptake of <70% on images were defined as abnormal.26 Segments with mild, moderate, or severe reduction were defined as 60 to 70%, 50 to 60%, and 40 to 50% uptake, and scored as 1, 2, and 3, respectively. Segments with absent uptake were defined as <40% uptake and scored as 4. Normal segments were scored as 0.

Definition 2 All segments with a mean uptake of <75% on images were defined as abnormal.27 Mild, moderate, and severe reduction was defined using the same approach. Ab-sent uptake was defined as <45% uptake.

Definition 3 Segments with a mean uptake of <80% on images were defined as abnormal.28 Mild, moderate, and severe reduction was defined using the same approach. Ab-sent uptake was defined as <50% uptake.

Summed BMIPP Scores The same 3 definitions were also applied to BMIPP images and summed BMIPP scores were obtained using the automated program.

Regional AnalysisThe LV wall was divided into 17 segments that were subdi-vided into 3 coronary artery territories according to the ASNC imaging guidelines, and then SSS and SDS were evaluated.9

Inter- and Intra-Operator Variability of Heart Score View SoftwareInter- and intra-operator variability in stress 201Tl MPI and rest BMIPP values was analyzed using the datasets. Two data-sets analyzed by 1 operator were compared to determine intra-operator agreement and datasets analyzed by different opera-tors were compared to assess inter-operator agreement.

Coronary Angiography (CAG)Among the 50 patients, 7 underwent CAG using a standard technique within 2 months of the index 201Tl stress/rest SPECT study. The presence or absence of CAD was determined from clinical CAG reports. The presence of significant CAD was a visually determined diameter stenosis of ≥75% for the 3 main coronary arteries or their major branches.29 Reversible myocardial ischemia was defined as SDS ≥2.30

Statistical AnalysisContinuous values are expressed as means ± SD. Categori-cal measures are presented as frequencies with percentages. Pearson’s correlation coefficients were calculated. Agreement among summed scores between visual interpretations and the automated program was determined using the Bland-Altman method. The optimal cut-off value for 201Tl SDS was deter-mined using receiver operating characteristics (ROC) curve analysis with the automated software. Concordance between the SDS and CAG was evaluated using the κ statistic. All data were statistically analyzed using MedCalc Software

(Version 9.4.2.0, Medcalc Software, Mariakerke, Belgium). A P-value of <0.05 was considered significant.

ResultsPatients’ CharacteristicsTable 1 shows the patients’ characteristics. Five of them had unstable angina pectoris but this stabilized by the time of the SPECT studies.

Correlations Among Summed Scores From MPI Between Visual Interpretation and Automated SoftwareTable 2 shows the correlation coefficients and the results of the Bland-Altman analysis of results using visual and auto-mated modalities.

Definition 1 (Abnormal Uptake Defined as <70% Uptake) Values closely correlated (SSS: r=0.953, P<0.001; SRS: r= 0.906, P<0.001; SDS: r=0.850, P<0.001) and Bland-Altman analysis revealed good agreement in the comparison of SSS, SRS and SDS between the 2 modalities.

Definition 2 (Abnormal Uptake Defined as <75% Uptake) Values closely correlated (SSS: r=0.934, P<0.001; SRS: r= 0.827, P<0.001; SDS: r=0.743, P<0.001) (Figures 2A–D) and Bland-Altman analysis revealed good agreement in the comparison of SSS, SRS, and SDS between the 2 modalities with a mean difference near 0 (Figures 2A–C).

Definition 3 (Abnormal Uptake Defined as <80% Uptake) Values closely correlated (SSS: r=0.852, P<0.001; SRS: r= 0.716, P<0.001; SDS: r=0.593, P<0.001) and Bland-Altman analysis revealed good agreement for all summed scores between the 2 modalities.

Correlations Among Regional Summed Scores From MPI Between Visual and Automated InterpretationTable 3 shows the correlation coefficients of the visual and automated modalities with regional analysis. Correlations between the 2 approaches were good for stress perfusion de-fects. Summed difference scores also correlated between the 2 methods.

Assessment of Anatomical CAD Using SDS Value of Heart Score View SoftwareUsing definition 2, the exact agreement between myocardial ischemia, SDS ≥2 by 201Tl SPECT, and CAG in assessing the anatomical extent of CAD was 85.7% (Table 4). The κ value was 0.695, indicating substantial agreement. The sensitivity and specificity were 80% and 100%, respectively, using 201Tl SDS with definition 2.

Among the 3 definitions, diagnostic accuracy was best for definition 2. However, the area under the curve did not signifi-cantly differ among the 3 definitions of an abnormal threshold in 201Tl SDS based on ROC analysis (Figure 3).

Table 3. Profiles of Correlations of Regional Summed Scores Between Visual Interpretation and Heart Score View Calculation

Score LAD territory LCX territory RCA territory

SSS r=0.903(P<0.001)

r=0.911(P<0.001)

r=0.778(P<0.001)

SDS r=0.793(P<0.001)

r=0.501(P<0.001)

r=0.530(P<0.001)

LAD, left anterior descending coronary artery; LCX, left circumflex coronary artery; RCA, right coronary artery. Other abbreviations see in Table 2.

Table 4. Detection of CAD Using the Automated Software in Patients Who Underwent CAG

Ischemia by 201Tl SPECTCAD by CAG

TotalNo Yes

No ischemia 2 1 3

Ischemia 0 4 4

Total 2 5 7

CAD, coronary artery disease; CAG, coronary angiography. Other abbreviation see in Table 2.



Correlations for Summed Scores From BMIPP Between Visual and Automated Interpretation

Definition 1 (Abnormal Uptake Defined as <70% Uptake) Values closely correlated (Summed BMIPP score: r=0.953, P<0.001) and Bland-Altman analysis revealed good agree-ment for summed BMIPP scores between the 2 modalities.

Definition 2 (Abnormal Uptake Defined as <75% Uptake) Values closely correlated (summed BMIPP score: r=0.913, P<0.001) (Figures 4A,B) and Bland-Altman analysis re-vealed good agreement for summed BMIPP scores between the 2 modalities with a mean difference near 0 (Figure 4A).

Definition 3 (Abnormal Uptake Defined as <80% Uptake) Values closely correlated (summed BMIPP score: r=0.882, P<0.001) and Bland-Altman analysis revealed good agree-ment for summed BMIPP scores between the 2 modalities.

Comparison of 3 Levels of Abnormal UptakeLinearity was the best among the 3 definitions for abnormal uptake defined as <75% uptake (Definition 2) with near y=x linearity between the 2 modalities (Table 2). Furthermore, Bland-Altman analysis showed that the mean of the differ-ence between the 2 methods was near 0 for this state.

Concordance of Visual Interpretations Between 2 Groups of AssessorsCorrelations coefficients and the results of Bland-Altman plot analysis of standard visual interpretations between 2 teams (n=12) were evaluated. Summed scores between assessor teams A and B closely correlated (SSS; r=0.895, P<0.001, SDS; r=0.950, P<0.001, summed BMIPP score; r=0.981, P< 0.001). Only SRS did not significantly correlate between the 2 teams (SRS; r=0.559, P=0.059). However, actual SRS did not significantly differ between them (1.1±1.2 vs. 0.3±0.7, NS).

Bland-Altman plot analysis revealed good agreement be-tween the 2 assessor teams for all variables (bias indicating mean difference between the 2 groups of assessors and 95% limits of agreement for SSS: 1.1 and –3.0 to 5.2; SRS: 1.1 and –2.0 to 4.1; SDS: 0.0 and –2.8 to 2.8; summed BMIPP score: –0.1 and –3.3 to 3.1, respectively).

Inter- and Intra-Operator Variability of Heart Score View SoftwareIntra-operator correlations were significantly higher in SSS

(r=0.995, P<0.001) and in BMIPP summed score (r=0.995, P<0.001), as was inter-operator agreement (r=0.976, P<0.001 and r=0.986, P<0.001, respectively).

DiscussionMyocardial perfusion and fatty acid abnormalities generated by the Heart Score View software and standard visual inter-pretations closely correlated and agreed between both stress/rest 201Tl and rest BMIPP myocardial imaging protocols. Thus, the program could objectively assess myocardial perfu-sion and metabolism that play important roles in the accurate diagnosis and risk assessment of patients with CAD.

Risk Assessment Using MPIStress MPI has excellent prognostic value. The ASNC guide-lines recommend semi-quantitative analysis of myocardial perfusion images and calculations of SSS, SRS, and SDS for risk stratification.19 Although the established scoring system is relatively simple,1 visual assessment of SPECT MPI is sub-jective and thus repeatability is lower than that of quantitative analysis.14 Quantitation of MPI has enhanced the reliability of interpretation.31 However, accuracy has been validated in very few automated programs15,32,33 that generate defect scores using a North American database. The North American phy-sique is quite different from that of Asians in terms of breast size, lateral fat pad, and diaphragm, all of which could atten-uate perfusion defects.16 Thus, automated programs should be designed considering target populations. Our automated Heart Score View software evaluates regional perfusion based on a Japanese database of normal values. Other studies have defined abnormal perfusion as relative % uptake <70%, 75%, and 80%.26–28 Therefore, we applied 80%, 75%, and 70% relative uptake as cut-off values in the present study and found good correlations between the scores generated visually and by the software. Thus, Heart Score View software was validated like the North American program. The closest cor-relation at 75% relative uptake among the 3 cut-off values agrees with previous findings.27 Quantitative perfusion SPECT and Heart View Software have similar functions,33 but the latter can also calculate SSS, SRS, SDS, and BMIPP summed score using cut-off values.

Only SRS did not significantly correlate between the 2 visual assessment teams. This was due to the small range of

Figure 3. Receiver operating curve analysis to detect coronary stenosis. Area under the curve was larger for definitions 1 and 2. However, the 3 definitions did not significantly differ. SDS, summed difference scores.



distribution in rest scores because none of the patients had prior myocardial infarction. In fact, SRS values did not sig-nificantly differ between the 2 teams.

Correlations between the 2 approaches were also good for regional SSS (Table 3) and SDS. Thus, Heart Score View software might be able to detect regional perfusion abnor-malities at stress and regional ischemia as well as the global approach. The correlation for SDS seems lower than that for SSS. The SDS is the result of subtracting the SSS from the SRS. Thus, the magnitude of the SDS is usually lower than that of the SSS, which would affect the results. However, the correlation remained significant and the present data agree with previous findings.15

We further applied the automated program to detect isch-emia location and culprit lesions. Although the number of study patients with CAG was limited, but when applied together with the ischemic parameter, SDS, the automated pro-gram substantially agreed with coronary stenosis and seemed to accurately detect culprit lesions because the diagnostic accuracy was comparable with previous reports.9,34,35 Thus, the automated program also has the potential to detect culprit lesions. However, this requires further investigation.

Experts in visual analysis usually consider the characteris-tics of given tracers and soft tissue attenuation. Mild defects related to soft tissue attenuation are usually considered nor-mal. Thus, summed stress and rest scores might be lower

A

B

Figure 4. Correlations and Bland-Altman plot of data generated from visual interpretation (VI) and Heart Score View software using β-methyl iodophenyl pentadecanoic acid (BMIPP) imaging. (A) Summed BMIPP scores. (B) Representative images of 66-year-old diabetic woman with effort angina pectoris. Rest BMIPP images show moderately sized, moderate intensity of per-fusion defect in inferolateral wall that improved at rest, indicating moderate ischemia in LCX territory. Visual and automated summed stress scores were both 8.



than those generated by the automatic program. Correlations between the 2 approaches were good in the present study, but the SDS and SRS were slightly higher in the automatic approach. However, differences between them were not sig-nificant. One solution for reducing this discrepancy might be to use an appropriate normal database as in the present study. However, further technical improvements are required to overcome this issue.

Risk Assessment Using Myocardial Fatty Acid ImagingBecause abnormal fatty acid metabolism can persist after myocardial ischemia, fatty acid imaging at rest might iden-tify ischemic history.36–38 We, Matsuki et al and others have reported that summed scores obtained from BMIPP images have prognostic value for cardiac events.10–12 However, auto-mated quantitative BMIPP imaging has not been validated. The approach used in the present study was the same that used for SPECT MPI. Heart Score View closely correlated with visual interpretations. Thus, this automated method of quantitative measurement might enhance the reliability of BMIPP imaging as well as of SPECT MPI.

Study LimitationsAlthough the current study population was small, the sample size was similar to that of a previous validation study.15 Our sample data represented a relatively larger CAD cohort of 196 patients and thus we consider it to have sufficient power to validate the new program.

Although we found close correlations between the visual and automated modalities, we did not have outcome data for the patients. Outcomes should be evaluated and a follow-up study is underway.

Only a few patients had CAG. The diagnostic accuracy of detecting the location of myocardial ischemic based on CAG data in the present study was relatively good, but further investigations are required with a larger population.

We validated the automated program using 201Tl, but we did not evaluate reliability using a technetium-labeled tracer of myocardial flow. This should also be addressed in further studies.

ConclusionMyocardial perfusion and fatty acid abnormalities closely cor-related and agreed between Heart Score View and standard visual interpretations both in stress/rest 201Tl and rest BMIPP myocardial imaging protocols.

This automated program might be useful for risk stratifica-tion in patients with CAD in the clinical setting.

AcknowledgmentsThe authors thank Masatsuyo Takano, RT, for technical expertise and Ms Norma Foster for critical reading of the manuscript.

DisclosureNone to disclose.

References 1. Hachamovitch R, Berman DS, Shaw LJ, Kiat H, Cohen I, Cabico

JA, et al. Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death: Differential stratification for risk of cardiac death and myocardial infarction. Circulation 1998; 97: 535 – 543.

2. Yoshinaga K, Chow BJ, Williams K, Chen L, deKemp RA, Garrard L, et al. What is the prognostic value of myocardial perfusion

imaging using rubidium-82 positron emission tomography? J Am Coll Cardiol 2006; 48: 1029 – 1039.

3. Nishimura T, Nakajima K, Kusuoka H, Yamashina A, Nishimura S. Prognostic study of risk stratification among Japanese patients with ischemic heart disease using gated myocardial perfusion SPECT: J-ACCESS study. Eur J Nucl Mol Imaging 2008; 35: 319 – 328.

4. Kato M, Matsumoto N, Nakano Y, Suzuki Y, Yoda S, Kasama S, et al. Combined assessment of myocardial perfusion and function by ECG-gated myocardial perfusion single-photon emission com-puted tomography for the prediction of future cardiac patients with type 2 diabetes mellitus. Circ J 2011; 75: 376 – 382.

5. Usui Y, Chikamori T, Nakajima K, Hida S, Yamashina A, Nishimura T; J-ACCESS Investigators. Prognostic value of post-ischemic stun-ning as assessed by gated myocardial perfusion single-photon emis-sion computed tomography: A subanalysis of the J-ACCESS study. Circ J 2010; 74: 1591 – 1599.

6. Matsuo S, Nakajima K, Yamasaki Y, Kashiwagi A, Nishimura T. Prognostic value of normal stress myocardial perfusion imaging and ventricular function in Japanese asymptomatic patients with type 2 diabetes: A study based on the J-ACCESS-2 database. Circ J 2010; 74: 1916 – 1921.

7. Matsumoto N, Nagao K, Hirayama A, Sato Y. Non-invasive assess-ment and clinical strategy of stable coronary artery disease by mag-netic resonance imaging, multislice computed tomography and myo-cardial perfusion SPECT. Circ J 2010; 74: 34 – 40.

8. Hashimoto A, Nakata T, Wakabayashi T, Kusuoka H, Nishimura T. Incremental prognostic value of stress/rest gated perfusion SPECT in patients with coronary artery disease: Subanalysis of the J-ACCESS study. Circ J 2009; 73: 2288 – 2293.

9. Klocke FJ, Baird MG, Lorell BH, Bateman TM, Messer JV, Berman DS, et al. ACC/AHA/ASNC guidelines for the clinical use of car-diac radionuclide imaging-executive summary: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (ACC/AHA/ASNC Committee to Revise the 1995 Guidelines for the Clinical Use of Cardiac Radio-nuclide Imaging). J Am Coll Cardiol 2003; 42: 1318 – 1333.

10. Matsuki T, Tamaki N, Nakata T, Doi A, Takahashi H, Iwata M, et al. Prognostic value of fatty acid imaging in patients with angina pectoris without prior infarction: Comparison with stress thallium imaging. Eur J Nucl Mol Imaging 2004; 31: 1585 – 1591.

11. Chikamori T, Fujita H, Nanasato M, Toba M, Nishimura T. Prog-nostic value of 1-123-(p-iodophenyl)-3-(R,S) methylpentadecanoic acid myocardial imaging in patients with known or suspected coro-nary artery disease. J Nucl Cardiol 2005; 12: 172 – 178.

12. Nishimura M, Tsukamoto K, Hasebe N, Tamaki N, Kikuchi K, Ono T. Prediction of cardiac death in hemodialysis patients by myocar-dial fatty acid imaging. J Am Coll Cardiol 2008; 51: 139 – 145.

13. Kawai Y, Tsukamoto E, Nozaki Y, Morita K, Sakurai M, Tamaki N. Significance of reduced uptake of iodinated fatty acid analogue for the evaluation of patients with acute chest pain. J Am Coll Cardiol 2001; 38: 1888 – 1894.

14. Danias PG, Ahlberg AW, Travin MI, Mahr NC, Abreu JE, Marini D, et al. Visual assessment of left ventricular perfusion and function with electrocardiography-gated SPECT has high intraobserver and interobserver reproducibility among experienced nuclear cardiolo-gists and cardiology trainees. J Nucl Cardiol 2002; 9: 263 – 270.

15. Berman DS, Kang X, Gransar H, Gerlach J, Friedman JD, Hayes SW, et al. Quantitative assessment of myocardial perfusion abnor-mality on SPECT myocardial perfusion imaging is more reproduc-ible than expert visual analysis. J Nucl Cardiol 2009; 16: 45 – 53.

16. Nakajima K, Okuda K, Kawano M, Matsuo S, Slomka P, Germano G, et al. The importance of population-specific normal database for quantification of myocardial ischemia: Comparison between Japanese 360 and 180-degree databases and a US database. J Nucl Cardiol 2009; 16: 422 – 430.

17. Nakajima K, Kumita S, Ishida Y, Momose M, Hashimoto J, Morita K, et al. Creation and characterization of Japanese standards for myocardial perfusion SPECT: Database from the Japanese society of Nuclear Medicine Working Group. Ann Nucl Med 2007; 21: 505 – 511.

18. Matsuo S, Nakajima K, Yamashina S, Sakata K, Momose M, Hashimoto J, et al. Characterization of Japanese standards for myo-cardial sympathetic and metabolic imaging in comparison with per-fusion imaging. Ann Nucl Med 2009; 23: 517 – 522.

19. Holly TA, Abbott BG, Al-Mallah M, Calnon DA, Cohen MC, DiFilippo FP, et al. Single photon-emission computed tomography. J Nucl Cardiol 2010; 17: 941 – 973.

20. Yoshinaga K, Manabe O, Tamaki N. Physiological assessment of myocardial perfusion using nuclear cardiology would enhance coro-nary artery disease patient care: Which imaging modality is best for



evaluation of myocardial ischemia? (SPECT-Side) Circ J 2011; 75: 713 – 723.

21. Yoshinaga K, Tahara M, Torii H, Kihara K. Predicting the effects on patients with dilated cardiomyopathy of beta-blocker therapy, by using iodine-123 15-(p-iodophenyl)-3- R,S-methylpentadecanoic acid (BMIPP) myocardial scintigraphy. Ann Nucl Med 1998; 12: 341 – 347.

22. Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, et al. Standardized myocardial segmentation and nomen-clature for tomographic imaging of the heart: A statement for health-care professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation 2002; 105: 539 – 542.

23. Garcia EV. Quantification of myocardial perfusion and function. Korean J Nuclear Med 1998; 32: 443 – 449.

24. Garcia EV, Van Train KF, Maddahi J, Prigent F, Areeda J, Waxman A, et al. Quantification of rotational thallium-201 myocardial tomo-praphy. J Nucl Med 1985; 26: 17 – 26.

25. Maes F, Collignon A, Vandermeulen D, Marchal G, Suetens P. Mul-timodality image registration by maximization of mutual informa-tion. IEEE Trans Med Imaging 1997; 16: 187 – 198.

26. Cuocolo A, Acampa W, Nicolai E, Pace L, Petretta M, Salvatore M. Quantitative thallium-201 and technetium 99m sestamibi tomog-raphy at rest in detection of myocardial viability in patients with chronic ischemic left ventricular dysfunction. J Nucl Cardiol 2000; 7: 8 – 15.

27. Yoshinaga K, Beanlands RS, Dekemp RA, Lortie M, Morin J, Aung M, et al. Effects of exercise training on myocardial blood flow in patients with stable coronary artery disease. Am Heart J 2006; 151: 1324.e11 – e8.

28. Beanlands RS, Ruddy TD, deKemp RA, Iwanochko RM, Coates G, Freeman M, et al. Positron emission tomography and recovery following revascularization (PARR-1): The importance of scar and the development of a prediction rule for the degree of recovery of left ventricular function. J Am Coll Cardiol 2002; 40: 1735 – 1743.

29. Sampson UK, Dorbala S, Limaye A, Kwong R, Di Cari MF. Diag-nostic accuracy of rubidium-82 myocardial perfusion imaging with hybrid positron emission tomography/computed tomography in the detection of coronary artery disease. J Am Coll Cardiol 2007; 13: 1052 – 1058.

30. Imaging guidelines for nuclear cardiology procedures, part 2: American Society of Nuclear Cardiology. J Nucl Cardiol 1999; 6: G47 – G84.

31. Wackers FJ. Science, art, and artifacts: How important is quantifica-tions for the practicing physician interpreting myocardial perfusion studies? J Nucl Cardiol 1994; 1: S109 – S117.

32. Slomka PJ, Nishina H, Berman DS, Akincioglu C, Abidov A, Friedman JD, et al. Automated quantitation of myocardial perfu-sion SPECT using simplified normal limits. J Nucl Cardiol 2005; 12: 66 – 77.

33. Slomka PJ, Nishina H, Berman DS, Kang X, Friedman JD, Hayes SW, et al. Automatic quantification of myocardial perfusion stress-rest change: A new measure of ischemia. J Nucl Med 2004; 45: 183 – 191.

34. Iskandarian AE, Heo J, Nallamothu N. Detection of coronary artery disease in woman with use of stress thallium-201 single photon emis-sion computed tomography myocardial perfusion imaging. J Nucl Cardiol 1997; 4: 329 – 335.

35. Watanabe K, Sekiya M, Ikeda S, Miyagawa M, Kinoshita M, Kumano S. Comparison of adenosine triphosphate and dipyridamole in diagnosis by thallium-201 myocardial scintigraphy. J Nucl Med 1997; 38: 1089 – 1094.

36. Yoshinaga K, Tamaki N. Imaging myocardial metabolism. Curr Opin Biotechnol 2007; 18: 52 – 59.

37. Tamaki N, Morita K, Kawai Y. The Japanese experience with meta-bolic imaging in the clinical setting. J Nucl Cardiol 2007; 14: S145 – S152.

38. Tamaki N, Yoshinaga K. Novel iodinated tracers, MIBG and BMIPP for nuclear cardiology. J Nucl Cardiol 2011; 18: 135 – 143.

validation of automated quantitation of myocardial

Documents