Arteries: How Far in the Lung Does Multi–Detector Row

(1)

Benoit Ghaye, MD David Szapiro, MD Ioana Mastora, MD Vale´rie Delannoy, MS Alain Duhamel, PhD Jacques Remy, MD Martine Remy-Jardin,

MD, PhD

Index terms:

Computed tomography (CT), angiography, 944.12916 Computed tomography (CT),

technology, 944.12915, 944.12916, 944.12919 Embolism, pulmonary, 944.77 Lung, anatomy, 60.12115, 60.92 Pulmonary arteries, CT, 944.12915,

944.12916, 944.12919

Radiology 2001;219:629 – 636

1From the Department of Radiology, University Center Hospital Calmette, Blvd Jules Leclerc, 59037 Lille Cedex, France (B.G., D.S., I.M., J.R., M.R.J.);

the Department of Medical Statistics, University of Lille, France (V.D., A.D.);

and the Medical Research Group, Lille, France (I.M., J.R., M.R.J.). From the 2000 RSNA scientific assembly. Re- ceived July 31, 2000; revision requested September 9; final revision received De- cember 27; accepted January 11, 2001.

Address correspondence to M.R.J.

(e-mail:mremy-jardin@chru-lille.fr).

©RSNA, 2001

Author contributions:

Guarantors of integrity of entire study, M.R.J., J.R.; study concepts and design, M.R.J., B.G.; literature research, M.R.J., J.R.; clinical studies, M.R.J., J.R., I.M.; data acquisition, M.R.J., D.S., B.G.; data analysis/interpretation, D.S., B.G.; statistical analysis, V.D., A.D.;

manuscript preparation, M.R.J.; manuscript definition of intellectual content, M.R.J., J.R.; manuscript editing, M.R.J.;

manuscript revision/review, M.R.J., B.G.;

manuscript final version approval, J.R.

Peripheral Pulmonary

Arteries: How Far in the Lung Does Multi–Detector Row

Spiral CT Allow Analysis?

¹

PURPOSE:To analyze the influence of multi– detector row spiral computed tomography (CT) on identification of peripheral pulmonary arteries.

MATERIALS AND METHODS: Peripheral pulmonary arteries were analyzed on optimally opacified contrast material– enhanced spiral CT angiograms in 30 patients devoid of pleuroparenchymal disease who underwent scanning with multi– detector row CT (collimation, 4⫻1 mm; pitch, 1.7–2.0; scanning time, 0.5 second). Two series of scans were systematically generated from each data set, 1.25-mm-thick (group 1) and 3-mm-thick (group 2) sections, leading to the analysis of 600 segmental (20 arteries per patient), 1,200 subsegmental (40 arteries per patient), 2,400 fifth-order (80 arteries per patient), and 4,800 sixth-order (160 arteries per patient) pulmonary arteries in each group.

RESULTS:Multi– detector row CT with reconstructed scans of 1.25-mm-thick sections (group 1) allowed (a) analysis of a significantly higher percentage of subsegmental arteries (94% in group 1 vs 82% in group 2;P⬍.001) and (b) a significantly higher percentage of fifth- and sixth-order arteries, respectively, identified in 74%

and 35% of cases in group 1 and 47% and 16% in group 2 (P⬍.001). The causes for inadequate depiction of subsegmental branches in group 1 were partial volume effect (43%), anatomic variants (39%), and cardiac (17%) and respiratory (1%) motion artifacts.

CONCLUSION:Multi– detector row CT with reconstructed scans of 1.25-mm-thick sections enables accurate analysis of peripheral pulmonary arteries down to the fifth order on spiral CT angiograms.

In the early 1990s, the introduction of spiral computed tomographic (CT) technology dramatically modified the evaluation of pulmonary arteries in routine clinical practice, which was previously based either on a noninvasive but indirect method, that is, ventilation-perfusion scintigraphy, or on an invasive and underused study, that is, pulmonary angiography. As a minimally invasive examination, spiral CT angiography rapidly emerged as a potentially useful diagnostic method, enabling a direct insight into endo- vascular abnormalities and thus a direct depiction of endoluminal clots (1–9).

Since its introduction, spiral CT technology has progressively improved and subse- quently influenced the overall accuracy of spiral CT angiography in the diagnostic work-up of pulmonary embolism. Initially performed with 5-mm collimation and 1-second rotation time, spiral CT angiography limited the detection of endoluminal clots to the segmental arteries (1–9). The availability of subsecond scanning then offered the possibil- ity to improve longitudinal spatial resolution, previously not accessible in practical scanning times.

In an anatomic study, Remy-Jardin et al (10,11) demonstrated that scanning with 2-mm collimation at 0.75 second per revolution enabled marked improvement in the analysis of segmental and subsegmental arteries, results further confirmed in routine clinical practice.

In a similarly designed study, Schoepf et al (12) also recently reported that detailed visualization of peripheral pulmonary arteries could be attained with subsecond spiral CT.

(2)

tor row spiral CT offers further increase in performance, in particular the ability to scan larger anatomic volumes with high spatial resolution. The purpose of this study was to analyze the influence of multi– detector row spiral CT technology on the identification of peripheral pulmonary arteries.

MATERIALS AND METHODS Population

This study was based on a retrospective analysis of spiral CT scans obtained at routine clinical practice during a 3-month period from January to March 2000. Dur- ing this period, spiral CT angiograms of the pulmonary circulation were routinely obtained with a four– detector row CT scanner by using narrow collimation scanning. We selected scans in 30 patients from the 130 patients who underwent spiral CT angiography of the pulmonary circulation during this period.

To be included in the present study, spiral CT examinations had to meet the anatomic and technical criteria proposed by Remy-Jardin et al (10): (a) depiction of a complete nondilated pulmonary arterial bed in both lungs, which implied (i) the absence of a history of lung sur- gery, (ii) the absence of lung distortion and/or parenchymal infiltration, and (iii) the absence of patent or suspicion of pri- mary or secondary pulmonary hyperten- sion; (b) a technically acceptable spiral CT examination with contrast enhancement that comprised (i) acquisition during strict inspiratory apnea, (ii) a degree of arterial enhancement coded as “excel- lent” (high degree of vascular opacifica- tion) or “good” (degree of contrast enhancement not high but sufficient for the analysis of pulmonary arteries) from the top to the bottom of the area of interest, and (iii) absence of endoluminal and/or periluminal abnormality. CT scans in the first 30 patients who fulfilled these criteria were included in the study; these examinations were selected by the senior author (M.R.J.).

The study population comprised 18 men and 12 women (age range, 25– 63 years; mean age, 48.6 years ⫾15 [SD]).

The indications for CT were as follows:

(a) suspicion of acute pulmonary embolism (n⫽11); (b) assessment of lung nodules (n⫽11), including suspicion of ma- lignant lesions and pulmonary vascular malformations; (c) cause of mild hemoptysis (n⫽5), that is, hemoptysis of less than 50 mL/day; and (d) suspicion of hi-

lar and/or mediastinal adenopathy (n⫽ 3). In all cases, the diagnosis of pulmonary embolism was excluded on the basis of one or more of the following: a nega- tive spiral CT scan (n ⫽ 11), a normal ventilation-perfusion scan (n⫽7), or an alternative diagnosis (n ⫽4). In the re- maining 19 patients, the final diagnoses were (a) presence of a solitary pulmonary nodule (n ⫽5) and normal lung paren- chyma (n⫽6) in the 11 patients referred because they were suspected of having lung nodules; the small size (⬍5 mm) of the lung nodules and their subpleural lo-

cation did not affect the analysis of peripheral pulmonary vessels; (b) chronic obstructive lung disease and airway in- fection in the five patients examined for mild hemoptysis; and (c) absence of adenopathy in the three patients suspected of having a hilar and/or mediastinal abnormality.

CT Evaluation

Spiral CT evaluation of the pulmonary circulation was performed with a multi–

detector row spiral CT scanner (Volume TABLE 1

Nomenclature of Bronchopulmonary Anatomy

Jackson and Huber Nomenclature*

Boyden Nomenclature^†

Segments Segmental Arteries Subsegmental Arteries Right upper lobe

Apical S1 RA1 RA1a, posterior^‡

RA1b, anterior

Anterior S2 RA2 RA2a, lateral^‡

RA2b, anterior

Posterior S3 RA3 RA3a, lateral^‡

RA3b, posterior Right middle lobe

Lateral S4 RA4 RA4a, posterior

RA4b, anterior

Medial S5 RA5 RA5a, superior

RA5b, inferior Right lower lobe

Superior (apical) S6 RA6 RA6a⫹b, superomedial

RA6c, lateral

Medial basal (paracardiac) S7 RA7 RA7a, anterolateral

RA7b, anteromedial

Anterior basal S8 RA8 RA8a, lateral

RA8b, basal

Lateral basal S9 RA9 RA9a, lateral

RA9b, basal

Posterior basal S10 RA10 RA10a, laterobasal

RA10b, mediobasal Left upper lobe

Upper division

Apicoposterior S1⫹3 LA1 LA1a, posterior^‡

LA1b, anterior^‡

LA3 LA3a, lateral^‡

LA3b, posterior^‡

Anterior S2 LA2 LA2a, lateral^‡

LA2b, anterior^‡ Lower lingular division

Superior S4 LA4 LA4a, posterior

LA4b, anterior

Inferior S5 LA5 LA5a, superior

LA5b, inferior Left lower lobe

Superior (apical) S6 LA6 LA6a⫹b, superomedial

LA6c, lateral

Anteromedial basal S7⫹8 LA7⫹8 LA7a, anterior

LA7b, medial LA8a, lateral LA8b, basal

Lateral basal S9 LA9 LA9a, lateral

LA9b, basal

Posterior basal S10 LA10 LA10a, laterobasal

LA10b, mediobasal

* Reference 13.

†Reference 14.

‡Nomenclature modified slightly for the purpose of CT interpretation.

(3)

Zoom; Siemens, Forcheim, Germany) by using 4 ⫻ 1-mm collimation, a table speed of 7 mm (n⫽19) per rotation (ie, pitch, 1.7) or 8 mm (n⫽11) per rotation (ie, pitch, 2.0), and a 0.5-second rotation time (140 kV and 20 –100 mAs per image, according to the indication and patient body type). From each data set, two series of transverse CT scans were reconstructed as further detailed. The scanning protocol included the survey of the entire tho-

rax with simultaneous administration of a 24%–30% contrast material. The mean duration of data acquisition was 16 seconds (range, 9 –22 seconds), and the mean z-axis coverage was 198 mm (range, 130 – 308 mm). Data were systematically obtained in the craniocaudal direction, from the lung apices to the level of the posterior costophrenic angles.

The patients received an injection of 120 (n⫽15) or 140 mL (n⫽15) of 24%

ioversol (Optiray 240; Guerbet, Roissy, France) or iohexol (Omnipaque 240; Ny- comed Ingenor, Amersham, United King- dom) (n⫽26) or 30% iohexol (Omnipaque 300; Nycomed Ingenor) (n⫽4). The iodinated contrast material was injected by way of peripheral venous access at a rate of 4 mL/sec (n⫽30); the mean start delay was 18 seconds (range, 15–28 seconds).

The bolus injection technique was used to administer contrast material with an automated injector (CT 9000; Liebel- Flarsheim, Cincinnati, Ohio) in every case.

The injection of contrast material was carefully monitored by a physician. Con- trast material was administered with the patient’s arm alongside the thorax, which thus allowed physician control of the venous access during the injection and avoided venous compression at the tho- racobrachial junction. All patients underwent scanning in the supine position.

To evaluate the influence of section thickness on the analysis of peripheral pulmonary arteries (arteries beyond the lobar level), two series of transverse CT scans were reconstructed from each data set, which led to the definition of two groups of images. Group 1 consisted of reconstructed scans of 1.25-mm-thick sections obtained at 1-mm intervals; group 2 consisted of reconstructed scans of 3-mm- thick sections obtained every 2 mm. In groups 1 and 2, two series of images were systematically considered: (a) mediastinal images, reconstructed with a soft re- construction kernel and viewed at mediastinal window settings (window width, 350 HU; window center, 50 HU); and (b) lung images, reconstructed with a high-spatial-frequency algorithm and viewed at lung window settings (window width, 1,600 HU; window center,⫺600 HU). The mean number of mediastinal images generated per patient was 198 (range, 130 –308) in group 1 and 99 (range, 65–154) in group 2.

Because our objective was to determine whether a 1.25-mm or a 3-mm section thickness is optimal for the evaluation of subsegmental pulmonary arteries, we systematically reconstructed two section widths from each data set. The section widths were chosen to allow comparison between the results of the present study with multi– detector row spiral CT and those of previous protocols based on single-section spiral CT. The section width of the reconstructed scans in group 2, that is, 3 mm, was close to the thinnest effective section thickness evaluated in the literature at the time this article was written, that is, 2.65 mm in scans with a 2-mm collimation and a pitch of 2.0 in the study by Remy-Jardin et al (10). The TABLE 2

Identification of Segmental Arteries according to Anatomic Region in 30 Patients at Multi–Detector Row Spiral CT Pulmonary Angiography A: Upper Lobes

Group

Right Left

RA1 Apical

RA2 Anterior

RA3 Posterior

LA1 Apical

LA2 Anterior

LA3 Posterior

1 26 (87) 25 (83) 22 (73) 14 (47) 27 (90) 23 (77)

2 26 (87) 25 (83) 22 (73) 14 (47) 27 (90) 22 (73)

B: Right Middle Lobe and Lingula

Group

Right Middle Lobe Lingula

RA4 Lateral

RA5 Medial

LA4 Superior

LA5 Inferior

1 25 (83) 28 (93) 25 (83) 29 (97)

2 24 (80) 28 (93) 25 (83) 28 (93)

C: Right Lower Lobe Group

RA6 Superior

RA7 Paracardiac

RA8 Anterior Basal

RA9 Lateral Basal

RA10 Posterior Basal

1 30 (100) 28 (93) 30 (100) 28 (93) 30 (100)

2 30 (100) 28 (93) 30 (100) 28 (93) 30 (100)

D: Left Lower Lobe Group

LA6 Superior

LA7 Anteromedial

LA8 Basal

LA9 Lateral Basal

LA10 Posterior Basal

1 27 (90) 29 (97) 30 (100) 26 (87) 29 (97)

2 27 (90) 29 (97) 30 (100) 26 (87) 29 (97)

Note.—Data are the number of patients in whom the artery was coded as analyzable. Numbers in parentheses are percentages. Group 1 consisted of 1.25-mm-thick reconstructed scans at 1-mm intervals; group 2 consisted of 3-mm-thick reconstructed scans at 2-mm intervals. No statistically significant difference in the identification of segmental arteries was found between groups 1 and 2.

TABLE 3

Causes for Inadequate Depiction of Segmental and Subsegmental Pulmonary Arteries

Arteries Coded as Nonanalyzable

Anatomic Variants

Partial Volume Effects

Cardiac Motion Artifacts

Respiratory Motion Artifacts Segmental

Group 1 (n⫽69) 69 (100) 0 0 0

Group 2 (n⫽72) 69 (96) 3 (4) 0 0

Subsegmental

Group 1 (n⫽75) 29 (39) 32 (43) 13 (17) 1 (1)

Group 2 (n⫽219) 29 (13) 165 (75) 23 (10) 2 (2)

Note.—Data are the number of arteries coded as nonanalyzable. Numbers in parentheses are percentages.

(4)

second section width, that is, 1.25 mm in group 1, represented a narrower thickness that was expected to minimize partial volume effects on peripheral pulmonary arteries.

Study Design

Consensus interpretation of the CT images was performed by two radiologists (B.G., D.S.), both experienced in the reading of helical CT angiograms of the pulmonary circulation (5 years of experi- ence), who read the images together. We

did not attempt to blind the readers to the scanning technique because of obvi- ous differences in the number of images in the two groups. The two readers analyzed group 2 images in random order.

Several weeks later, they analyzed group 1 images, also presented in random order.

To identify segmental and subsegmental arteries, we used the nomenclature outlined by Remy-Jardin et al (10). This nomenclature is based on the standard descriptions by Jackson and Huber (13) and Boyden (14), with slight modifica-

tions to account for anatomic variations and the orientation of vessels in the transverse plane on CT scans (Table 1).

Twenty segmental (ie, third-order) and 40 subsegmental (ie, fourth-order) arteries are described in this nomenclature.

The fifth-order pulmonary arteries (n ⫽ 80) were recognized as symmetric dichotomous divisions of the corresponding subsegmental branch. The sixth-order pulmonary arteries (n⫽160) were recognized as dichotomous divisions of the corresponding fifth-order pulmonary artery. To identify Figure 1. Multi– detector row spiral CT scans (4⫻1-mm collimation, pitch of 1.7, and administration of a 24% iodinated contrast material at a rate of 4 mL/sec) obtained at the level of the right middle lobe in a 63-year-old woman for comparison of reconstructed images of 1.25- and 3-mm-thick sections to depict subsegmental pulmonary arteries.(a– e)Reconstructed scans of 1.25-mm-thick sections obtained at 1-mm intervals, photographed at mediastinal window settings, and displayed in a cephalocaudal direction.(f–i) Reconstructed scans of 3-mm-thick sections obtained at 2-mm intervals in the same volume of interest as that ina– e(mediastinal images displayed in a cephalocaudal direction). The superior subsegmental ramus (RA5a; arrow) of the medial segmental artery of the right middle lobe (star) is adequately depicted on reconstructed images of 1.25-mm-thick sections, whereas it is coded as nonanalyzable on reconstructed images of 3-mm-thick sections owing to partial volume effect in its medial portion. Note the adequate depiction of the inferior subsegmental ramus (RA5b; arrowhead) on both series of images.

(5)

pulmonary arterial sections with confi- dence, we analyzed lung and mediastinal images simultaneously. Prior to image analysis, a training session (the 30 patient scans were not used for the training session) was held during which the readers were familiarized with the modifica- tions in nomenclature and agreed on the following strategy for the analysis.

Each artery was individually coded and was considered analyzable when depicted from the proximal to the distal portions on a single or successive transverse CT scans without partial volume effects. The reason for inadequate arterial depiction was systematically recorded, including respiratory and/or cardiac motion artifacts, partial volume effects due to the small size of the vessel, and absence of an ar-

tery because of anatomic variants. The absence of an artery was assessed by means of the nondetectability of a given artery on both mediastinal and lung images, interpreted in the light of a precise knowledge of the most frequent anatomic variants in each lobe. The z-axis coverage enabling depiction of pulmonary arteries down to the sixth order was systematically recorded. For practical pur- poses, the upper part of this volume of interest was recorded in relation to the top of the aortic arch, whereas the lower part was referenced in relation to the level of the right inferior pulmonary vein.

Statistical Analysis

The statistical analyses were performed with commercially available software (SAS;

SAS Institute, Cary, NC). Each artery was individually coded as analyzable or not, leading the readers to determine a number of arteries analyzable per patient in each anatomic zone of each lung. These results were also presented as rates of recognition of pulmonary arteries, ex- pressed as percentages and calculated by means of the following formula: The rate of recognition equalled the number of arteries coded as analyzable in a given anatomic region multiplied by 100 and divided by the maximum number of arteries anatomically present. For each cat- egory of pulmonary arteries, the distribu- tion of rates was compared between groups 1 and 2 by using the paired Wil- coxon rank sum test, which is adapted to ordinal variables.

TABLE 4

Identification of Subsegmental Arteries according to Anatomic Region in 30 Patients at Multi–Detector Row Spiral CT Pulmonary Angiography

A: Right Upper Lobe Group

RA1a Posterior

RA1b Anterior

RA2a Lateral

RA2b Anterior

RA3a Lateral

RA3b Posterior

1 29 (97) 28 (93) 18 (60) 30 (100) 29 (97) 26 (87)

2 29 (97) 28 (93) 15 (50) 29 (97) 26 (87) 25 (83)

B: Left Upper Lobe Group

LA1a Posterior

LA1b Anterior

LA2a Lateral

LA2b Anterior

LA3a Lateral

LA3b Posterior

1 29 (97) 28 (93) 27 (90) 29 (97) 27 (90)* 29 (97)

2 29 (97) 28 (93) 22 (73) 27 (90) 19 (63) 28 (93)

C: Right Middle Lobe

Group RA4a Posterior RA4b Anterior RA5a Superior RA5b Inferior

1 27 (90)^† 26 (87) 30 (100) 28 (93)

2 16 (53) 25 (83) 28 (93) 25 (83)

D: Lingula

Group LA4a Posterior LA4b Anterior LA5a Superior LA5b Inferior

1 28 (93)^† 30 (100) 29 (97)* 21 (70)

2 16 (53) 26 (87) 20 (67) 17 (57)

E: Right Lower Lobe Group

RA6a⫹b Superomedial

RA6c Lateral

RA7a Anterolateral

RA7b Anteromedial

RA8a Lateral

RA8b Basal

RA9a Lateral

RA9b Basal

RA10a Laterobasal

RA10b Mediobasal 1 30 (100) 30 (100) 29 (97) 30 (100) 28 (93)* 30 (100) 28 (93)* 30 (100) 29 (97)* 30 (100)

2 30 (100) 28 (93) 28 (93) 30 (100) 20 (67) 30 (100) 18 (60) 30 (100) 20 (67) 30 (100)

F: Left Lower Lobe Group

LA6a⫹b Superomedial

LA6c Lateral

LA7a Anterior

LA7b Medial

LA8a Lateral

LA8b Basal

LA9a Lateral

LA9b Basal

LA10a Laterobasal

LA10b Mediobasal 1 30 (100) 28 (93) 26 (87) 27 (90)* 28 (93)^† 30 (100) 26 (87)^† 30 (100) 28 (93)^‡ 30 (100)

2 30 (100) 28 (93) 22 (73) 19 (63) 15 (50) 30 (100) 13 (43) 30 (100) 22 (73) 30 (100)

Note.—Data are the number of patients in whom the artery was coded as analyzable. Numbers in parentheses are percentages.

* The difference between groups 1 and 2 was statistically significant:P⬍.01.

†The difference between groups 1 and 2 was statistically significant:P⬍.001.

‡The difference between groups 1 and 2 was statistically significant:P⬍.05.

(6)

Analysis of Segmental Arteries (Third-Order Arteries)

The two readers analyzed 600 segmental arteries in group 1 and 600 segmental arteries in group 2 (20 segmental arteries per patient). No statistically significant difference was found in the total number of analyzable segmental arteries in groups 1 and 2, regardless of whether the left and right lung were considered together or separately. The percentages of analyzable segmental arteries were 88.5% (531 of 600 segmental arteries) in group 1 and 88% (528 of 600 segmental arteries) in group 2. The percentages of analyzable segmental arteries according to anatomic region are summarized in Table 2. Ana- tomic variants accounted for inadequate depiction of segmental arteries in every nonanalyzable artery in group 1 (100%) and in 69 (96%) of the 72 nonanalyzable segmental arteries in group 2 (Table 3).

Analysis of Subsegmental Arteries (Fourth-Order Arteries)

A total of 1,200 subsegmental arteries in group 1 and 1,200 subsegmental arteries in group 2 were individually evaluated (40 subsegmental arteries per patient). The percentages of analyzable subsegmental arteries were 94% (1,125 of 1,200) in group 1 and 82% (981 of 1,200) in group 2 (P ⬍.001) (Fig 1). The percentages of analyzable subsegmental arteries according to anatomic region are summarized in Table 4.

No statistically significant difference was found between group 1 scans and group 2 scans in the identification of 29 of the 40 subsegmental arteries that were systematically evaluated per patient. A statistically significantly higher frequency of identification of 11 subsegmental arteries was observed in group 1. These subsegmental arteries included LA3a in the upper lobes; RA4a, LA4a, and LA5a in the right middle lobe and lingula; and RA8a, RA9a, RA10a, LA7b, LA8a, LA9a, and LA10a in the lower lobes.

On group 1 scans, the frequency of identification of subsegmental pulmonary arteries was 90%–100% for all but six subsegmental arteries, namely RA2a (60%) and RA3b (87%) in the upper lobes; RA4b (87%) and LA5b (70%) in the right middle lobe and lingula; and LA7a (87%) and LA9a (87%) in the lower lobes.

The causes of inadequate depiction of subsegmental arteries are summarized in Table 3 and were related to partial volume effects in 75% of cases on group 2

scans and to partial volume effects and anatomic variants in 43% and 39%, respectively, on group 1 scans.

Analysis of Fifth- and Sixth-Order Pulmonary Arteries

A total of 2,400 fifth-order (80 arteries per patient) and 4,800 sixth-order (160 arteries per patient) pulmonary arteries were individually analyzed in each group.

The frequency of identification of fifth- and sixth-order pulmonary arteries is summarized in Table 5. The percentage of analyzable fifth-order pulmonary arteries was significantly higher in group 1 (74%;

1,782 of 2,400 arteries) than in group 2 (47%; 1,128 of 2,400 arteries; P⬍.001) (Fig 2).

The percentage of identification of fifth-order pulmonary arteries was 80%

(966 of 1,200) in the right lung in group 1 (vs 52% [621 of 1,200] in group 2;P⬍ .001) and 68% (816 of 1,200) in the left lung in group 1 (vs 42% [507 of 1,200] in group 2;P⬍.001). On group 1 scans, the fifth-order pulmonary arteries were identified with a frequency greater than 60%

(range, 67% [240 of 360] to 89% [322 of 360]) in all the lobes except the lingula, where only 55% (131 of 240) of branches were adequately depicted. The fifth-order pulmonary arteries identified with the highest frequency were located in the apical segment of the right upper lobe in both group 1 (97% [116 of 120]) and group 2 (82.5% [99 of 120]). The least- identified fifth-order pulmonary arteries were located in the left paracardiac segment, both in group 1 (35% [42 of 120]) and group 2 (15% [18 of 120]).

The percentage of analyzable sixth-order pulmonary arteries was significantly higher in group 1 (35%; 1,667 of 4,800 arteries) than in group 2 (16%; 754 of 4,800 arteries;P⬍.001). The percentage of identification of sixth-order pulmonary arteries was 38% (914 of 2,400) in the right lung in group 1 (vs 17% [404 of 2,400] in group 2; P ⬍ .001) and 31%

(753 of 2,400) in the left lung in group 1 (vs 14.5% [350 of 2,400] in group 2;P⬍ .001). The sixth-order pulmonary arteries identified with the highest frequency were located in the apical segment of the right lower lobe in group 1 (59% [141 of 240]) and in the posterior segment of the right lower lobe in group 2 (37% [89 of 240]). The least-identified sixth-order pulmonary arteries were located in the left paracardiac segment, both in group 1 (7% [16 of 240]) and group 2 (2% [five of 240]).

The mean z-axis coverage enabling

analysis of peripheral pulmonary arteries down to the sixth order was 165 mm⫾ 19, extending from 11 mm⫾7.5 above the top of the aortic arch to 54 mm⫾ 11.5 below the level of the right inferior pulmonary vein.

DISCUSSION

In our study, adequate depiction of segmental arteries was attained in 88.5%

and 88% of the cases in groups 1 and 2, respectively. These findings confirm previously reported results, considering anatomic and clinical studies (10 –12), that showed that the mean percentage of analyzable segmental arteries was 83%–95%.

In the present investigation, we specifi- cally recorded the causes of inadequate depiction of pulmonary arteries. This led us to observe that the percentage of adequately depicted segmental arteries could reach 100% in group 1 and 96% in group 2, inasmuch as the lack of depiction of arteries due to anatomic variants can be considered within the normal range.

Therefore, a technically adequate multi–

detector row spiral CT angiogram is com- patible with the analysis of the entire segmental arterial bed on thin-collimated scans.

TABLE 5

Percentage of Analyzable Fifth- and Sixth-Order Pulmonary Arteries A: Upper Lobes

Group

Right Left

Fifth Order

Sixth Order

Fifth Order

Sixth Order

1 89 49 79 38

2 67* 22* 53* 17*

B: Right Middle Lobe and Lingula

Group

Right Middle

Lobe Lingula

Fifth Order

Sixth Order

Fifth Order

Sixth Order

1 69 21 55 16

2 33* 5^† 21* 3^†

C: Lower Lobes

Group

Right Left

Fifth Order

Sixth Order

Fifth Order

Sixth Order

1 80 38 67 34

2 50* 18* 44* 18*

Note.—Data are percentages of arteries (fifth order,n⫽2,400; sixth order,n⫽4,800).

*P⬍ .001.

†P⬍.01.

(7)

A significant difference between groups 1 and 2 concerned the evaluation of the subsegmental pulmonary arteries. On group 1 scans, 94% of these arteries were adequately depicted, whereas 82% were coded as analyzable on group 2 scans.

The frequency of identification of subsegmental branches was 90%–100% for all but six subsegmental arteries on group 1 scans, whereas only half of the subseg-

mental arteries were seen with a frequency of 90%–100% in group 2. The main reason for coding these vessels as nonanalyzable was the presence of partial volume effects, more frequent in group 2 (75% of cases [165 of 219 arteries]) than in group 1 (43% of cases [32 of 75 arteries]).

On both series of reconstructed scans, the frequency of identification of subsegmental arteries was superior to that pre-

viously reported with single-section spiral CT (10,12). Remy-Jardin et al (10) found that 37% of the subsegmental bed was analyzable on collimated scans of 3-mm-thick sections, whereas this percentage reached 61% on collimated scans of 2-mm-thick sections. Comparing spiral CT and electron-beam CT on collimated scans of 5-mm-thick sections, Schoepf et al (12) recently reported that Figure 2. Multi– detector row spiral CT scans (4⫻1-mm collimation, pitch of 1.7, and administration of a 24% iodinated contrast material at a rate of 4 mL/sec) obtained at the level of the left upper lobe in a 74-year-old woman for comparison of reconstructed images of 1.25- and 3-mm-thick sections to depict fifth- and sixth-order pulmonary arteries.(a–f)Reconstructed scans of 1.25-mm-thick sections obtained at 1-mm intervals, photographed at mediastinal window settings, and displayed in a cephalocaudal direction.(g–i)Reconstructed scans of 3-mm-thick sections obtained at 2-mm intervals in the same volume of interest as that ina–f(mediastinal images displayed in a cephalocaudal direction). Note the adequate depiction of the lateral (LA2a; small open star) and anterior (LA2b; small solid star) subsegmental rami of the anterior segmental artery (large open star) in the left upper lobe on both series of mediastinal images. The anterior subsegmental ramus of LA2b gives rise to two symmetric dichotomous divisions: a lateral fifth-order branch (arrowhead) and an anterior fifth-order branch (long arrow). The anterior fifth-order branch then divides into two symmetric sixth-order branches (short arrows). All of the fifth- and sixth-order branches are analyzable on reconstructed scans of 1.25-mm-thick sections, whereas only the anterior rami are adequately depicted on reconstructed images of 3-mm-thick sections.

(8)

mental arteries were 73% in the right lung and 70% in the left lung. The use- fulness of narrow-collimation scanning for the analysis of subsegmental branches has also been recently underlined by Baile et al (15) in an experimental study based on single-section spiral CT.

The improvement in the evaluation of an anatomic compartment composed of branches 2–3 mm in diameter is directly related to the availability of multi– detector row spiral CT scanners, enabling scanning of the pulmonary vascular bed with a narrow collimation during acceptable scanning times, otherwise shorter than those commonly selected with single-section spiral CT. In the present study, our population underwent scanning with a four– detector row spiral CT scanner with a 0.5-second rotation time. The mean duration of data acquisition for the entire thorax was 16 seconds, whereas it was 26 seconds in the study by Schoepf et al (12) and 22 seconds for the coverage of a 10 – 12-cm region of interest in the study by Remy-Jardin et al (10).

As small arterial branches were easily depicted on narrowly collimated scans in routine clinical practice, we attempted to determine the frequency of identification of the fifth- and sixth-order arteries with multi– detector row spiral CT, which was previously not investigated in the radiology literature, to our knowledge. We observed that 74% of the fifth-order pulmonary arteries were analyzable in group 1, a percentage significantly higher than that found in group 2 (47%). As expected, the percentage of analyzable sixth-order pulmonary arteries was significantly higher in group 1 (35%) than in group 2 (16%). Despite an overall limited identification of these branches on narrowly collimated scans, it is notice- able that sixth-order pulmonary branches could be identified in 59% of cases in the apical segment of the right lower lobe in group 1 and in 37% of cases in the posterior segment of the right lower lobe in group 2. Obviously, multi– detector row spiral CT does not allow the evaluation of pulmonary arteries down to the capillary bed as on conventional or digital angiograms. However, our results suggest that the fifth- and sixth-order pulmonary arteries should no longer be considered beyond the scope of CT evaluation when

row spiral CT.

Because our investigation was performed in optimal conditions, namely the analysis of a complete nondilated pulmonary arterial bed in both lungs on technically optimal spiral CT examinations, these results need further validation in clinical studies. However, con- versely to the anatomic studies (10,12) previously published, the present study did not compare scans from different populations. Therefore, differences observed between the two groups of patients in our study were not influenced by differences in arterial patterns but re- flected technical differences between CT scans.

Because the main goal of the present investigation was to determine whether an accurate analysis of subsegmental and smaller branches could be performed with multi– detector row CT, the consensus reading of two experienced chest radiologists was considered to be better suited to this anatomic study. Therefore, we did not attempt to evaluate interob- server variability in the recognition of peripheral pulmonary arteries. From a practical standpoint, the analysis of peripheral pulmonary arteries on hard cop- ies was judged time-consuming by the two readers, especially for the reconstructed scans of 1.25-mm-thick sections.

Further studies are needed to establish whether these vessels can be accurately depicted alternatively with diagnostic work- stations.

Our results demonstrate that peripheral pulmonary arteries down to the fifth- order branches can be accurately depicted with reconstructed scans of 1.25- mm-thick sections by using multi– detector row spiral CT. The ability to scan the entire thorax with narrow collimation is expected to modify the imaging protocol of the pulmonary circulation in routine clinical practice, as well as the radiolo- gist’s daily approach to viewing the number of images generated from each data set.

References

1. Remy-Jardin M, Remy J, Wattinne L, Gi- raud F. Central pulmonary thromboem- bolism: diagnosis with helical volumetric CT with the single breath hold technique— comparison with pulmonary angiography. Radiology 1992; 185:381–387.

et al. Detection of pulmonary embolism in patients with unresolved clinical and scintigraphic diagnosis: helical CT versus angiography. AJR Am J Roentgenol 1995;

164:1369 –1374.

3. Senac JP, Verhnet H, Bousquet C, et al.

Embolie pulmonaire: apport de la tomo- densitome´trie he´licoı¨dale. J Radiol 1995;

74:339 –345.

4. van Rossum AB, Pattynama PMT, Ton ER, et al. Pulmonary embolism: validation of helical CT angiography in 149 patients.

Radiology 1996; 201:467– 470.

5. van Rossum AB, Treuniet FEE, Kieft GJ, Smith SJ, Schepers-Bok R. Role of helical volumetric computed tomographic scanning in the assessment of patients with clinical suspicion of pulmonary embolism and an abnormal ventilation-perfusion scan. Thorax 1996; 51:23–28.

6. Remy-Jardin M, Remy J, Deschildre F, et al. Diagnosis of acute pulmonary embolism with helical CT: comparison with pulmonary angiography and scintigraphy. Radiology 1996; 200:699 –706.

7. Ferretti GR, Bosson JL, Buffaz PD, et al.

Acute pulmonary embolism: role of helical CT in 164 patients with intermediate probability at ventilation-perfusion scintigraphy and normal results at duplex US of the legs. Radiology 1997; 205:453– 458.

8. van Rossum AB, Pattynama PM, Mallens WM, Hermans J, Heijerman HG. Can helical CT replace scintigraphy in the diagnostic process in suspected pulmonary embolism? A retrolective-prolective co- hort study focusing on total diagnostic yield. Eur Radiol 1998; 8:90 –96.

9. Drucker EA, Rivitz SM, Shepard JAO, et al.

Acute pulmonary embolism: assessment of helical CT for diagnosis. Radiology 1998; 209:235–241.

10. Remy-Jardin M, Remy J, Artaud D, Des- childre F, Duhamel A. Peripheral pulmonary arteries: optimization of the acquisition protocol. Radiology 1997; 204:157–

163.

11. Remy-Jardin M, Baghaie F, Bonnel F, Mas- son P, Duhamel A, Remy J. Thoracic helical CT: influence of subsecond scan time and thin collimation on evaluation of peripheral pulmonary arteries. Eur Ra- diol 2000; 10:1297–1303.

12. Schoepf UJ, Helmberger T, Holzknecht, et al. Segmental and subsegmental pulmonary arteries: evaluation with electron- beam versus spiral CT. Radiology 2000;

214:433– 439.

13. Jackson CL, Huber JF. Correlated applied anatomy of the bronchial tree and lungs with a system of nomenclature. Dis Chest 1943; 9:319 –326.

14. Boyden EA. Segmental anatomy of the lungs. New York, NY: McGraw-Hill, 1955.

15. Baile EM, King GG, Muller NL, et al. Spi- ral computed tomography is comparable to angiography for the diagnosis of pulmonary embolism. Am J Respir Crit Care Med 2000; 161:1010 –1015.