Professional Documents
Culture Documents
DOI 10.1007/s11604-015-0420-7
ORIGINAL ARTICLE
13
Jpn J Radiol
order to determine the appropriate antibiotic therapy for age SD: 66.018.5years) and the control subjects
CAP patients [47]. were 62 males and 43 females aged 2988years (mean
Despite its superiority, CT should be used judiciously age SD: 5818.5years). There were 63 male and 44
because of the higher cost and the high radiation dose to female patients with bacterial pneumonia aged 2894years
the thorax. The purpose of this study was to evaluate the (mean ageSD: 70.513.9years) and 15 males and 7
diagnostic capability of chest radiographs compared with females with atypical pneumonia aged 1285years (mean
HRCT when used as a gold standard for CAP, which is ageSD: 43.922.2years). There were no patients in an
divided into bacterial and atypical pneumonia, and to deter- immunosuppressive state in either group. The institutional
mine what kinds of lesions tend to be overlooked or misin- review board at this hospital did not require approval for a
terpreted and why they were overlooked or misinterpreted retrospective study of routinely obtained clinical data and
on chest radiographs. This will make it possible to deter- radiological images.
mine the capability and limitations of chest radiographs, as
well as the appropriate indication for using HRCT in the Chest radiograph andCT scanning
diagnosis of CAP.
The CT scans were obtained with an X-vision instrument
(Toshiba Medical Systems, Tokyo, Japan) at 120kV and
Materials andmethods 140mA. After contiguous 10-mm section CT was per-
formed through the chest, additional HRCT images con-
Patients sisting of 2-mm section images were obtained at 2-, 5- or
10-mm intervals through the abnormal lung parenchyma
Chest radiographs and HRCT scans obtained in 129 depicted on 10-mm section CT images. HRCT images were
consecutive patients with CAP in one affiliated institu- reconstructed with a high-spatial-frequency algorithm.
tion between April of 2002 and December of 2010 were CT scans were obtained at suspended end-inspiration, in
included in our study. Patients were identified through the supine position, without intravenous contrast material.
a review of a CT database of outpatients who had under- All images were viewed at window levels appropriate for
gone chest CT for suspected CAP. There were no criteria lung parenchyma (window width: 1500HU, window level:
as to whether CT should be done or not in that hospital. CT 600 HU) and mediastinum (window width: 300HU,
examination was performed after a judgement and request window level: 35HU). All 129 patients with pneumonia
by the attending physician. Patients with CAP were further underwent HRCT scans. The postero-anterior computed
divided into bacterial and atypical pneumonia cases. The radiograph images were obtained with an FCR XU-D1
diagnosis of bacterial pneumonia was made by two pat- instrument (Fujifilm Corporation, Tokyo, Japan) using
terns; (1) a combination of laboratory examinations, includ- 130kVp and 100mA with a photo-timer and a 12:1 grid.
ing an elevation of the white blood cell (WBC) count and The focus-object distance was 200cm. The mean interval
C-reactive protein (CRP), and positive results of sputum between chest radiograph and CT was 0.3days (range,
culture, and (2) a combination of laboratory examinations, 01days). All patients underwent a chest radiograph before
including an elevation of the WBC count and CRP, and the chest CT examination.
immediate response using only penicillin or cephem anti-
biotics. A diagnosis of atypical pneumonia was made based Interpretation ofimages
on a combination of laboratory examination, including an
elevation of CRP, and an elevation of the serum antibody The chest radiographs of all cases and HRCT scans of
for Mycoplasma, Chlamydophila, or viruses. As a result, patients with CAP were separately reviewed at 1-month
107 patients with bacterial pneumonia and 22 patients intervals by two chest radiologists and a final decision was
with atypical pneumonia (18 patients with Mycoplasma reached by consensus. The two observers had no knowl-
pneumonia, 3 patients with Influenza viral pneumonia, and edge of the clinical information except that the subjects
one patient with Chlamydophila pneumonia) were identi- were either patients with CAP or healthy subjects, so that
fied. The cases with mixed infection of both bacterial and they could pay attention only to the presence or absence
atypical agents were excluded from this study. As a control of pneumonia shadows instead of making differential
group, 105 consecutive healthy subjects were also included diagnoses.
who underwent chest radiograph and CT examination due Each of the CT and radiographic findings was separately
to a suspicion of lung cancer or tuberculosis after lung coded as present or absent for the following: (1) airspace
cancer or tuberculosis screening tests, in which no abnor- consolidation, (2) ground-glass opacity (GGO), (3) nodules
malities were found on CT. The patients with pneumonia and (4) thickening of the bronchial wall. Concerning HRCT
included 78 males and 51 females aged 1294years (mean findings, consolidation was defined as marked increase in
13
Jpn J Radiol
lung attenuation with obscuration of underlying pulmo- include predominant airspace consolidation pattern, and
nary vessels. GGO was defined as a hazy increase in lung segmental, lobar or lobular distribution [913]; and those
attenuation without obscuration of underlying pulmonary for atypical pneumonia include bronchopneumonia pat-
vessels. Nodule was defined as a rounded or irregular opac- tern or extensive or multifocal ground-glass opacification,
ity, well defined or poorly defined, measuring up to 3cm sometimes accompanied by bronchial wall thickening or
in diameter. Concerning chest radiographic findings, con- reticular opacities [1017].The distribution of pulmonary
solidation was defined as marked homogeneous increased disease was evaluated in six areas of the bilateral lungs as
lung opacification which completely obscures the margins follows. The upper zones were defined as the areas above
of vessels and airway walls and often contains air broncho- the level of the carina; the middle zones, between the level
gram within it. Ground-glass opacity was defined as hazy of the carina and the level of the inferior pulmonary veins;
increased lung opacification without air bronchogram the lower zones, under the level of the inferior pulmonary
within which the margins of pulmonary vessels become veins. All chest radiographic and HRCT findings were sep-
slightly indistinct. Nodules were defined as a rounded arately coded as present or absent in the six areas. When
opacity, well or poorly defined, measuring up to 3cm in more than two HRCT findings were present within one
diameter [8]. The distribution of nodules in the interpreta- lung area, all findings were recorded as present.
tion of HRCT findings was recorded as centrilobular, peri- We hypothesized that the chest radiographic interpre-
lymphatic or random. Each HRCT finding was compared tation of lesions, especially that concerning airspace con-
between bacterial pneumonia and atypical pneumonia. solidation and GGO, can be influenced by the thickness of
The most likely chest radiographic diagnosis of normal, lesions detected and measured in the antero-posterior direc-
bacterial or atypical pneumonia was made on the basis of tion on HRCT images; therefore, we also measured the
previous chest radiographic descriptions of these diseases maximum antero-posterior distance (thickness) of lesions
in the literature [918], and the final decision was also in order to correlate the interpretation of consolidation or
reached by consensus of the same radiologists. The diag- GGO with the thickness of the lesions. The interpretation
nostic criteria of chest radiographs for bacterial pneumonia of lesions between the two modalities (patterns of HRCT/
Table1The HRCT findings and the accuracy of chest radiographic interpretation compared with each HRCT finding in all cases
HRCT findings Accuracy of chest radiographic interpretation
b
Positive (%) Negative (%) p value False negative (%) True positive (%) False positive (%) p valueb
13
Jpn J Radiol
Table2The sensitivity, specificity, and accuracy of chest radiographic interpretation compared with each HRCT finding in the six separate
areas
Right middle (%) Right middle (%) Right lower (%) Left upper (%) Left middle (%) Left lower (%)
Sensitivity
Consolidation
Bacterial pneumonia 73.3 50.0 71.2 77.8 62.5 71.4
Atypical pneumonia 100 60.0 60.0 100 100 83.3
All cases 75.0 51.2 70.2 80.0 67.9 73.2
GGO
Bacterial pneumonia 64.7 72.2 76.7 75.0 69.2 71.0
Atypical pneumonia 71.4 63.6 50.0 40.0 81.8 62.5
All cases 65.9 70.8 72.9 64.7 73.0 69.2
Nodule
Bacterial pneumonia 28.6 28.6 30.6 20.0 28.6 38.1
Atypical pneumonia 33.3 25.0 22.2 0 0 0
All cases 29.4 27.9 28. 16.7 23.1 26.7
Bronchial wall thickening
Bacterial pneumonia 20.0 32.5 51.3 37.5 35.3 54.5
Atypical pneumonia 16.7 12.5 40.0 50.0 20.0 25.0
All cases 19.4 29.2 50.0 40.0 31.8 50.0
Specificity
Consolidation
Healthy subject 100 98.1 98.1 100 100 99.0
Bacterial pneumonia 95.7 89.9 98.2 99.0 94.0 93.1
Atypical pneumonia 85.7 82.4 88.2 100 83.3 87.5
All cases 96.8 93.7 97.2 99.6 96.1 95.9
GGO
Healthy subject 100 99.0 97.1 99.0 98.1 97.1
Bacterial pneumonia 95.9 86.8 78.7 95.8 87.7 84.2
Atypical pneumonia 93.3 90.9 91.7 100 100 78.6
All cases 97.9 94.7 91.5 97.7 93.9 90.8
Nodule
Healthy subject 100 100 99.0 99.0 99.0 98.1
Bacterial pneumonia 96.2 88.9 88.7 97.9 93.0 88.4
Atypical pneumonia 100 92.9 100 100 100 100
All cases 98.5 95.3 95.2 98.6 96.6 94.1
Bronchial wall thickening
Healthy subject 100 100 99.0 100 100 99.0
Bacterial pneumonia 95.1 92.5 88.2 99.0 93.3 94.1
Atypical pneumonia 100 100 94.1 100 100 94.4
All cases 98.0 97.3 94.7 99.6 97.2 96.6
Accuracy
Consolidation
Healthy subject 100 98.1 98.1 100 100 99.0
Bacterial pneumonia 92.5 75.7 85.0 97.2 86.9 86.0
Atypical pneumonia 86.4 77.3 81.8 100 86.4 86.4
All cases 95.3 85.9 90.6 98.7 92.7 91.9
GGO
Healthy subject 100 99.0 97.1 99.0 98.1 97.1
Bacterial pneumonia 86.0 79.4 77.6 93.5 83.2 80.4
Atypical pneumonia 86.4 77.3 72.7 86.4 90.9 72.7
13
Jpn J Radiol
Table2continued
Right middle (%) Right middle (%) Right lower (%) Left upper (%) Left middle (%) Left lower (%)
All cases 92.3 88.0 85.9 95.3 90.6 87.2
Nodule
Healthy subject 100 100 99.0 99.0 99.0 98.1
Bacterial pneumonia 78.5 69.2 69.2 90.7 80.4 78.5
Atypical pneumonia 81.8 68.2 68.2 90.9 77.3 59.1
All cases 88.5 82.9 82.5 94.4 88.5 85.5
Bronchial wall thickening
Healthy subject 100 100 99.0 100 100 99.0
Bacterial pneumonia 77.6 70.1 74.8 94.4 84.1 86.0
Atypical pneumonia 77.3 68.2 81.8 95.5 81.8 81.8
All cases 87.6 83.3 86.3 97.0 91.0 91.5
radiographic interpretation) was coded as follows: the Diagnostic capability ofchest radiograph inthe
interpretation of airspace consolidation on HRCT seen diagnosis ofCAP
as normal, GGO, and consolidation on chest radiographs
was coded as Cons/Normal, Cons/GGO, and Cons/Cons, We calculated the sensitivity, specificity, accuracy, positive
respectively. In the same way, the interpretation of GGO on predictive value, and negative predictive value for the chest
HRCT seen as normal, GGO, and consolidation on chest radiographic diagnosis, including healthy subjects, bacterial
radiographs was coded as GGO/Normal, GGO/GGO, and pneumonia, or atypical pneumonia, by comparing with the
GGO/Cons, respectively. Some patients had consolidation HRCT diagnosis as the gold standard. For each incorrect diag-
or GGO in more than two areas on HRCT; therefore, in nosis, possible reasons were suggested by the two radiologists.
some cases there were more than two lesions of which the
thickness was measured. Statistical analysis
Correlation betweeneach chest radiographic andeach The inter-observer agreement between the first two radiolo-
HRCT finding gists in the interpretation of each chest radiographic finding
was assessed by calculating the kappa value (-value). The
We evaluated the diagnostic accuracy of chest radiographs comparisons between the HRCT findings in the two types of
of patients with pneumonia by defining the HRCT finding pneumonia were made using a chi-square test for independ-
as the gold standard. The interpretation of each chest radio- ence. The comparisons of FP, TD, and FN of the chest radio-
graphic finding was divided into false negative (FN), true graphic interpretations between the two types of pneumonia
diagnosis (TD), and false positive (FP) diagnosis based on were also made using a chi-square test for independence.
the HRCT findings as follows: (1) FN, abnormal lesions When the sample size was small, for example, a cell value
depicted on HRCT images could not be detected as such in the data table was less than 5, we used Fishers exact test
on chest radiograph, (2) TD, including true positive (TP), instead of a chi-square test. A p value <0.05 was considered to
all abnormal lesions on HRCT could be detected as such on indicate a significant difference. If a significant difference was
chest radiographs; and true negative (TN), all normal areas found between the groups using the chi-square test, adjusted
on HRCT were interpreted as normal on chest radiograph, standardized residuals were calculated in order to clarify
(3) FP, normal areas on HRCT images were misinterpreted which groups chest radiographic interpretation contributed to
as abnormal lesions. the significant difference. An adjusted standardized residual of
We also evaluated chest radiographic interpretation >1.96 or <1.96 was considered to indicate significantly more
as TD, FP, and FN separately in the above-mentioned six frequent or less frequent results, respectively. The correlation
areas. In each radiographic finding of FN and FP, the rea- between the interpretation of lesions (airspace consolidation
sons for such misinterpretations were suggested by the two and GGO) and the thickness of lesions was assessed by Fish-
radiologists based on their consensus. The reasons for chest ers protected least significant difference (PLSD) test. A sig-
radiographic misinterpretation include lesional thickness. nificant correlation was considered present when the p value
In this respect, we defined thin or thick lesions as hav- was less than 0.05. These tests were performed using a com-
ing maximum antero-posterior distances less than 5cm or mercially available software program (SPSS Statistics, version
more than 10cm, respectively. 21 for Windows, IBM, Japan).
13
Jpn J Radiol
Table3The suggested reasons for false negative and positive interpretations each finding in CAP cases
Pneumonia HRCT findings The suggested reasons
13
Jpn J Radiol
13
Jpn J Radiol
most frequent reason. However, there were no cases in the counterpart of either pneumonia. As to the reason for
either type of pneumonia which were interpreted as nor- the misdiagnosis of one pneumonia as the other, misinter-
mal on chest radiographs due to overlooked airspace con- pretation of consolidation or GGO as the other due to the
solidation. Twenty cases with bacterial pneumonia and 13 thickness of the lesion was frequent (Fig.2). Furthermore,
with atypical pneumonia were misinterpreted as atypical the most frequent reason for the misdiagnosis of atypical
and bacterial pneumonia, respectively. For about half of pneumonia as bacterial pneumonia was overlapping of
the patients with either pneumonia, the reason for misdi- nodules or bronchial wall thickening with consolidation or
agnosis was the resemblance of HRCT findings to those of GGO (Fig.4).
13
Jpn J Radiol
Fig.3A 43-year-old female patient with bacterial pneumonia. The Fig.4A 31-year-old female patient with atypical (mycoplasma)
chest radiograph of this patient was interpreted as bacterial pneumo- pneumonia. The chest radiograph of this patient was interpreted
nia. a HRCT scan in the right lower lobe shows focal segmental air- as bacterial pneumonia. a HRCT scan in the left upper lobe shows
space consolidation, around which tiny GGO nodules (arrows) and segmental GGO, within which tiny GGO nodules (arrowheads) and
thickening of the bronchial walls (arrow heads) are noted. b Chest thickening of the bronchial walls (arrows) are seen. b Chest radio-
radiograph shows airspace consolidation in the right lower lung; how- graph shows GGO in the left upper lung; however, it fails to show
ever, it fails to show nodules or thickening of the bronchial walls nodules or thickening of the bronchial walls
13
Jpn J Radiol
Fig.5The correlation of the chest radiographic interpretation with and chest radiographic interpretation except for between Cons/GGO
the thickness of the lesions. The measurement of the thickness of and Cons/Normal by using Fishers PLSD test
pneumonic lesions was different between all the patterns of HRCT
Table4The accuracy of chest radiographic diagnosis in CAP and prominence of centrilobular nodules or the thickening of
normal cases the bronchial wall are the most important key HRCT find-
Chest radiographic diagnosis ings which discriminate atypical pneumonia from bacterial
CAP Healthy subjects
pneumonia [47, 16, 17, 20]. Also, in the current study,
nodules (centrilobular in distribution in all cases) were sig-
Bacterial Atypical nificantly more frequent in atypical pneumonia than bacte-
CAP (n=129) 119 10 rial pneumonia. However, there was no significant differ-
Bacterial (n=107) 80 20 7 ence in the frequency of bronchial wall thickening between
Atypical (n=22) 13 6 3 the two groups, the reason for which remains unknown. In
Healthy subjects (n=105) 9 96 the interpretations of radiographs, nodules and thickening
5 4 of the bronchial walls tended to be overlooked, especially
in patients with atypical pneumonia in almost all lung
Sensitivity: 92.2% (119/129), Specificity: 91.4% (96/105), accuracy: areas, although there was no statistically significant differ-
91.9% (215/234), positive predictive value (PPV): 93.0% (119/128),
negative predictive value (NPV): 90.6% (96/106)
ence between the interpretations. Reittner also mentioned
the difficulty in detecting lobular distribution, centrilobular
involvement, and interstitial abnormalities on chest radio-
opacities, except for 17 false positive interpretations we graphs, which could usually be detected on HRCT [17].
classified as unidentified. Concerning the lower sensitivity of nodules and bronchial
In clinical practice, it is essential to differentiate between wall thickening especially, the sensitivity of nodules in
bacterial and atypical pneumonia, and such classification is atypical pneumonia was 0% in all the left lung areas. The
one of the most important aspects described in the manage- reason for this remains uncertain because the influence of
ment guidelines for CAP published by the Japanese Res- cardiac silhouette makes it difficult to identify lower sensi-
piratory Society [19]. tivity in all the left lung areas.
Therefore, the differential diagnosis based on radio- Furthermore, the overlooking of nodules or thickening
logic findings is essential. Although there have been few of the bronchial wall was indicated as a possible explana-
reports of HRCT findings in patients with atypical pneumo- tion for the misdiagnosis of both types of pneumonia as
nia other than mycoplasma pneumonia, it is likely that the normal. On the other hand, the overlooking of airspace
13
Jpn J Radiol
Bacterial pneumonia (Total 107 cases) Atypical pneumonia (n=20) Resemblance of HRCT findings to atypical pneumonia 11
Misinterpretation of consolidation as GGO due to the thin 9
consolidation
Normal (n=7) Overlooking of nodules or bronchial wall thickening due to small 7
abnormalities
Overlooking of GGO due to the thin lesion 3
Overlooking of GGO hidden behind the diaphragm 1
Atypical pneumonia (Total 22 cases) Bacterial pneumonia (n=13) Overlapping of nodules or bronchial wall thickening with 7
consolidation or GGO
Resemblance of HRCT findings to bacterial pneumonia 6
Misinterpretation of GGO as consolidation due to the thick GGO 2
Normal (n=3) Overlooking of nodules or bronchial wall thickening due to small 2
abnormalities
Overlooking of GGO due to the thin lesion 1
Overlooking of GGO hidden behind the diaphragm 1
Overlooking of nodules or bronchial wall thickening hidden behind 1
the diaphragm
Healthy subjects (Total 105 cases) Bacterial pneumonia (n=5) Misinterpretation of normal structure as consolidation 3
Misinterpretation of normal structure as GGO 2
Misinterpretation of normal structure as nodules 1
Atypical pneumonia (n=4) Misinterpretation of normal structure as GGO 2
Misinterpretation of normal structure as nodules or bronchial wall 2
thickening
There were some cases with multiple reasons for false negative diagnosis
consolidation did not contribute to the reasons for a missed NPV were generally good, in spite of the difficulty in dif-
diagnosis of pneumonia. This result shows that lobar pneu- ferentiating between bacterial and atypical pneumonia. The
monia, which is supposed to be more frequent in bacterial inter-observer agreement of chest radiographic interpreta-
pneumonia rather than in atypical pneumonia [4], could tion between the two radiologists was also good. The rea-
confidently be detected by chest radiograph, since this type son for these two good results is partially because the two
of pneumonia usually shows extensive airspace consolida- chest radiologists had known that patients were all either
tion by the time patients show clinical symptoms. It also normal subjects or patients with pneumonia. Because other
indicates that a correct radiographic diagnosis of atypical pathologies or circumstances would likely complicate the
pneumonia may sometimes be impossible due to the dif- diagnosis in actual clinical practice, the overall diagnostic
ficulty in detecting nodules or thickening of the bronchial accuracy of a chest radiograph would likely be lower.
wall when consolidation or GGO is predominant. This is There were several limitations in this study. First, the
the most frequent reason for the misdiagnosis of atypical number of atypical pneumonia cases is smaller than that
pneumonia as bacterial pneumonia. Although the misin- of bacterial pneumonia, which may be associated with
terpretation of thick GGO as airspace consolidation was the larger number of elderly patients. Therefore, it might
observed in 5 patients with atypical pneumonia, only 2 not be possible to draw solid conclusions about the differ-
patients were misdiagnosed as having bacterial pneumonia. ences in the detectability of pneumonia between the two
In this study, the accurate diagnosis of atypical pneumonia types of pneumonia. Another weakness of this study is its
was made in only 6 of the 22 cases. It might be related to retrospective design. It is difficult to perform a prospective
the technical disadvantages derived from chest radiographs study of pneumonia since it is difficult to perform CT in all
due to their inability to enable differentiation of superim- cases which are suspected of having pneumonia from the
posed opacities. Therefore, HRCT could be indicated if the viewpoint of radiation exposure. Furthermore, concerning
radiographic features do not match those of atypical pneu- the chest radiographic interpretation of consolidation and
monia although atypical pneumonia is clinically suspected. GGO, the differentiation between the two findings is often
As to the diagnostic capability of chest radiographs difficult, even though two chest radiographic findings were
for CAP, by combining bacterial and atypical pneumonia defined in this study. This might affect the results of corre-
together, the sensitivity, specificity, accuracy, PPV, and lation between chest radiographic and HRCT findings, and
13
Jpn J Radiol
in addition, the diagnostic capability of chest radiographs. 5. Nambu A, Saito A, Araki T, Ozawa K, Hiejima Y, Akao M, etal.
Finally, there might be a selection bias in the process of Chlamydia pneumoniae: comparison with findings of Myco-
plasma pneumoniae and Streptococcus pneumoniae at thin-sec-
choosing pneumonia patients because only those patients tion CT. Radiology. 2006;238:3308.
who had uncertain radiographic findings underwent chest 6. Okada F, Ando Y, Wakisaka M, Matsumoto S, Mori H. Chla-
CT. Therefore, it might be difficult to generalize the diag- mydia pneumoniae pneumonia and Mycoplasma pneumoniae
nostic capability of chest radiographic diagnosis for CAP pneumonia: comparison of clinical findings and CT findings. J
Comput Assist Tomogr. 2005;29:62632.
in this study. 7. Reittner P, Ward S, Heyneman L, Johkoh T, Mller NL. Pneu-
monia: high-resolution CT findings in 114 patients. Eur Radiol.
2003;13:51521.
Conclusion 8. Hansell DM, Bankier AA, MacMahon H, McLoud TC, Mller
NL, Remy J. Fleischner Society: glossary of terms for thoracic
imaging. Radiology. 2008;246:697722.
The chest radiographic interpretations of GGO and con- 9. British Thoracic Society Standards of Care Committee. BTS
solidation were influenced by the thickness of lesions, and guidelines for the management of community acquired pneumo-
those of nodules and thickening of bronchial walls were nia in adults. Thorax. 2001;56(Suppl 4):164.
10. Scanlon GT, Unger JD. The radiology of bacterial and viral
influenced by coexisting GGO and consolidation. As to the pneumonias. Radiol Clin North Am. 1973;11:31738.
causes of incorrect diagnosis of pneumonia by chest radio- 11. Vilar J, Domingo ML, Soto C, Cogollos J. Radiology of bacterial
graph, the overlooking of centrilobular nodules and the pneumonia. Eur J Radiol. 2004;51:10213.
thickening of bronchial walls occurred frequently. There- 12. Lvy M, Dromer F, Brion N, Leturdu F, Carbon C. Community-
acquired pneumonia. Importance of initial noninvasive bacterio-
fore, the diagnostic capability of chest radiographs seems a logic and radiographic investigations. Chest. 1988;93:438.
little inferior in atypical pneumonia to bacterial pneumonia, 13. Tew J, Calenoff L, Berlin BS. Bacterial or nonbacterial
although the sensitivity, specificity, and accuracy of chest pneumonia: accuracy of radiographic diagnosis. Radiology.
radiographic diagnosis were generally good. 1977;124:60712.
14. Lynch DA, Armstrong JD 2nd. A pattern-oriented approach to
chest radiographs in atypical pneumonia syndrome. Clin Chest
Acknowledgments This research received no specific grant from Med. 1991;12:20322.
any funding agency in the public, commercial, or not-for-profit sec- 15. Murray HW, Tuazon C. Atypical pneumonias. Med Clin North
tors. The authors are grateful to Steven Gardner for the preparation of Am. 1980;64:50727.
the English manuscript. 16. Kim EA, Lee KS, Primack SL, Yoon HK, Byun HS, Kim TS,
etal. Viral pneumonias in adults: radiologic and pathologic find-
Conflict of interest The authors declare that they have no conflict ings. Radiographics. 2002;22(Suppl 1):S13749.
of interest and there is no ethical problem. 17. Reittner P, Mller NL, Heyneman L, Johkoh T, Park JS, Lee KS,
etal. Mycoplasma pneumoniae pneumonia: radiographic and
high-resolution CT features in 28 patients. Am J Roentgenol.
2000;174:3741.
References 18. Syrjl H, Broas M, Suramo I, Ojala A, Lhde S. High-resolu-
tion computed tomography for the diagnosis of community-
1. Albaum MN, Hill LC, Murphy M, Li YH, Fuhrman CR, Brit- acquired pneumonia. Clin Infect Dis. 1998;27:35863.
ton CA, etal. Interobserver reliability of the chest radiograph 19. The committee for the JRS guidelines in management of respira-
in community-acquired pneumonia. PORT Investigators. Chest. tory infections. The JRS guidelines for the management of com-
1996;110:34350. munity acquired pneumonia in adults. Nihon Kokyuki Gakkai
2. McLoud TC, Carrington CB, Gaensler EA. Diffuse infiltra- Zasshi. 2007; Suppl:285.
tive lung disease: a new scheme for description. Radiology. 20. Miyashita N, Sugiu T, Kawai Y, Oda K, Yamaguchi T, Ouchi K,
1983;149:35363. etal. Radiographic features of Mycoplasma pneumoniae pneu-
3. Mathieson JR, Mayo JR, Staples CA, Mller NL. Chronic dif- monia: differential diagnosis and performance timing. BMC
fuse infiltrative lung disease: comparison of diagnostic accuracy Med Imaging. 2009;29:7.
of CT and chest radiography. Radiology. 1989;171:1116.
4. Tanaka N, Matsumoto T, Kuramitsu T, Nakaki H, Ito K, Uchi-
sako H, etal. High resolution CT findings in community-
acquired pneumonia. J Comput Assist Tomogr. 1996;20:6008.
13