high resolution computedtomographic findings in...

Thorax 1996;51:397-402

High resolution computed tomographic findingsin pulmonary tuberculosis

O N Hatipoglu, E Osma, M Manisali, E S Uqan, P Balci, A Akkoclu, 0 Akpinar,C Karlikaya, C Yuksel

Chest DepartmentO N HatipogluE S UqanA AkkoqluO AkpinarC Karlikaya

RadiodiagnosticDepartmentE OsmaM ManisaliP BalciC Yuksel

Faculty of Medicine,Dokuz EylulUniversity, Inciralti,Izmir, Turkey

Correspondence to:Dr 0 N HatipogluChest MedicineDepartment, Faculty ofMedicine, Trakya University,Edirne, Turkey.Received 6 July 1994Returned to authors19 October 1994Revised version received3 January 1995Accepted for publication9 November 1995

AbstractBackground- Although chest radiographsusually provide adequate information forthe diagnosis of active pulmonary tuber-culosis, minimal exudative tuberculosiscan be overlooked on standard chestradiographs. The aim of the present studywas to assess the findings of active pul-monary tuberculosis on high resolutioncomputed tomographic (HRCT) scans,and to evaluate their possible use in de-termining disease activity.Methods - Thirty two patients with newlydiagnosed active pulmonary tuberculosisand 34 patients with inactive pulmonarytuberculosis were examined. The diag-nosis ofactive pulmonary tuberculosis wasbased on positive acid fast bacilli in spu-tum and bronchial washing smears or cul-tures and/or changes on serial radiographsobtained during treatment.Results - With HRCT scanning centri-lobular lesions (n = 29), "tree-in-bud"appearance (n = 23), and macronodules5-8mm in diameter (n = 22) were mostcommonly seen in cases of active pul-monary tuberculosis. HRCT scans showedfibrotic lesions (n = 34), distortion ofbronchovascular structures (n = 32), em-physema (n = 28), and bronchiectasis (n =24) in patients with inactive tuberculosis.Conclusions - Centrilobular densities inand around the small airways and "tree-in-bud" appearances were the most char-acteristic CT features of disease activity.HRCT scanning clearly differentiated oldfibrotic lesions from new active lesions anddemonstrated early bronchogenic spread.These findings maybe ofvalue in decisionson treatment.(Thorax 1996;51:397-402)

Keywords: tuberculosis, imaging, high resolution com-puted tomography.

The most common form of pulmonary tuber-culosis in adults is postprimary disease orreinfection. Although chest radiographs areextremely good for the diagnosis of active pul-monary tuberculosis, minimal exudative tuber-culosis can be overlooked on standard chestradiography.'-3 Computed tomography (CT) issuperior to conventional radiography in de-tecting activity and CT features of pulmonarytuberculosis have been described.4 6 In recentyears, because of high resolution power andminimal partial volume effect, high resolutioncomputed tomography (HRCT) has been

found to be superior to conventional chestradiography and standard CT in the local-isation of disease in the pulmonary lobule andin the evaluation of pulmonary parenchymaldisease. Using these advantages of HRCT, wehave studied patients with active and inactivepulmonary tuberculosis to assess the con-tribution of HRCT to the evaluation of tuber-culosis.

MethodsPATIENTSThirty two patients with active pulmonarytuberculosis and 36 patients with inactive dis-ease were studied prospectively at our de-partment from February 1993 to March 1994usingHRCT scanning. Two cases with inactivetuberculosis whose chest radiographs andHRCT images showed findings consistent withactivity were excluded as these new lesionswere due to non-tuberculous infection in oldtuberculosis lesions. The group with activetuberculosis included 23 men of mean age 44years (range 19-80). The diagnosis of activepulmonary tuberculosis was based on (1) de-tection of acid fast bacilli in sputum smears(n = 17, 53%); (2) detection of acid fast bacilliin cultures of sputum (n=21, 65%) or bron-chial washings (n = 3); and (3) radiographicand clinical improvement after administrationof two or more antituberculous drugs forpatients whose clinical and radiographic find-ings suggested a diagnosis of pulmonary tuber-culosis but smear and culture results werenegative (n = 8, 25%). In addition, pleural bi-opsy confirmed the diagnosis in two patientsin whom sputum smears were also positivefor acid fast bacilli. Bronchial washings wereobtained from 15 of the patients with activedisease who could not expectorate sputum(n = 2) or whose sputum smears were negativefor acid fast bacilli (n = 13). Fourteen patientshad a history of previous antituberculosis ther-apy and, for the remaining 18, it was the initialdiagnosis.The inactive group included 25 men ofmean

age 58 years (range 24-78). The diagnosis ofinactive pulmonary tuberculosis was establish-ed on the basis of the following criteria: (1)findings of residual fibrotic changes on chestradiography; (2) absence of radiological progres-sion comparing chest radiographs at recruit-ment with radiographs obtained six monthspreviously or at follow up; and (3) absence ofacid fast bacilli in sputum or bronchial washingson smear or culture. Bronchial washings wereobtained from 12 patients who could notexpectorate sputum and from two who had

397

on 11 July 2018 by guest. Protected by copyright.

http://thorax.bmj.com

/T

horax: first published as 10.1136/thx.51.4.397 on 1 April 1996. D

ownloaded from

http://thorax.bmj.com/

Hatipoglu, Osma, Manisali, Ufan, Balci, Akkoflu, Akpinar, Karlikaya, Yiiksel

radiographic evidence of active pulmonarytuberculosis but had negative smear and culturefor Mycobacterium tuberculosis. These twopatients were excluded from the study (onehad a positive culture of Staphylococcus aureusand the other a positive culture of Klebsiellapneumoniae). Both showed clinical and radio-graphic improvement after administration ofappropriate antibiotics. Serial follow up radio-graphs were taken monthly in the inactivetuberculosis group for 3-6 months. Follow upradiographs showed no progression.

IMAGINGHRCTPatients with active or suspected reactivationof tuberculosis on conventional chest radio-graphs or suspicious clinical findings under-went examination with HRCT within sevendays of diagnosis; in the inactive tuberculosisgroup HRCT was undertaken within 3-32days.HRCT scans were obtained with a 9800

(GE Medical Systems, Milwaukee, USA) CTscanner. Scanning was performed with 1-5 mmcollimation, 120 kvP-170 mA, two second scantime, 512 x 512 matrix, and a bone algorithmat 10 mm intervals from the lung apices to thebases. Intravenous contrast medium was notadministered. Images were obtained with a35-40 cm field of view; some were targetedretrospectively with a 20 cm field of view.The terms used to interpret HRCT findings

in the active cases were:(1) Centrilobular nodule or linear structures:well defined lesions 1-4 mm thick, separatedby more than 2 mm from the pleural surfaceor interlobular septa.(2) "Tree-in-bud" appearance: a branchinglinear structure with more than one contiguousbranching site.(3) Macronodule: a nodule 5-8 mm in dia-meter.

In addition, consolidation (lobular, sub-segmental, segmental, lobar), bronchial wallthickening, cavitation (single, multiple), em-physema, bronchovascular distortion, fibroticchanges, pleural thickening, bronchiectasis,lymphadenopathy, parenchymal calcification,pleural effusion, miliary nodules, and groundglass appearances were noted.

Chest radiographyIn patients with active disease chest radiographsat the time ofinitial examination were reviewed,focusing on the presence of cavities (single ormultiple), the distribution of lesions (uni-laterally or bilaterally), and findings of activity(infiltration with or without cavitation).

STATISTICAL ANALYSIS

Comparisons between active and inactive dis-ease were made using x2 analysis and a valueof p<0 05 was regarded as significant.

Stepwise discriminant analysis as used inthis study identifies the discriminating HRCTfindings that most usefully measure the char-

HRCTfindings ofpatients with active and inactivepulmonary tuberculosis

Active Inactivetuberculosis tuberculosis

Findings (n = 32) (n 34)

Centrilobular nodule and/or 29 (91) 0 (0)branching linear structure*

Tree-in-bud appearance* 23 (71) 0 (0)Macronodule* 22 (69) 0 (0)Cavity* 16 (50) 4 (12)Consolidation* 14 (44) 0 (0)Bronchial wall thickening 14 (44) 22 (65)Interlobular septal thickening 10 (34) 3 (9)Ground glass opacity* 12 (38) 4 (11)Bronchiectasis 18 (56) 24 (71)Emphysema* 14 (44) 28 (82)Bronchovascular distortion* 20 (63) 32 (94)Fibrotic changes* 21 (66) 34 (100)Calcified mediastinal lymph 7 (22) 15 (44)node enlargement*

Parenchymal calcification 14 (44) 18 (53)Pleural thickening or retraction 20 (63) 24 (71)Lymphadenopathy (>1O mm) 5 (16) 0 (0)Pleural effusion 2 (6) 0 (0)Miliary nodules 1 (3) 0 (0)

Values in parentheses are percentages.* p<O-05.

acteristics in which the two groups are expectedto differ. The objective is to maximise sep-aration of groups by weighting and combiningthe discriminating variables in some linearform. A user selected criterion is applied, andthe first variable included in the analysis hasthe largest acceptable value for the selectioncriterion. After the first variable is entered, thevalue of the criterion is re-evaluated for allvariables not in the model and the variable withthe largest acceptable criterion value is enterednext. The variable entered first is then re-evaluated to determine whether it meets theremoval criterion. If it does it is removed fromthe model. The next step is to examine thevariables not in the equation for removal. Vari-ables are removed until nine remain that meetthe removal criterion. Variable selection ter-minates when no more variables meet entry orremoval criteria.

In this study the ability to classify casescorrectly with the use of the discriminant func-tion was measured; pooled within group cor-relations between discriminating variables andcanonical discriminate functions were cal-culated by SPSS/PC +.The study was approved by the hospital eth-

ics committee.

ResultsHRCT FEATURES OF ACTIVE DISEASEThe HRCT findings in patients with activepulmonary tuberculosis are summarised in thetable. Centrilobular nodular or linear struc-tures (n=29, 91%) (p<0 001) (figs 1 and 2),"tree-in-bud" appearance (n = 23, 71%)(p<0001) (figs 1 and 3), and macronodules(n=22, 69%) (p<0001) (fig 4) were the mostcommon HRCT findings seen in active pul-monary tuberculosis.

Centrilobular nodules, branching linearstructures, and/or "tree-in-bud" appearanceswere detected in 29 of the 32 patients withactive tuberculosis (91 %). The exceptions wereone patient with miliary tuberculosis, one

398



/T


ownloaded from


High resolution computed tomographic findings in pulmonary tuberculosis

p* _

Figure 1 HRCT section at the level of the bifurcation of the trachea in a 36 year oldman with sputum positive for acid fast bacilli. Centrilobular nodular densities (curvedblack arrow) and "tree-in-bud" appearance (white arrow) are common findings in activepulmonary tuberculosis. The appearances are due to endobronchial spreading. Coalescenceof granulomas and acinar-lobular infiltrates usually cause subsegmental consolidationareas (black arrow with asterisk). Interlobular septal thickening (straight black arrow) isalso a common entity observed in active tuberculosis.

Figure 2 HRCT section at the mid lung level in a 42year old man with sputum positive for acid fast bacilli.The long white arrow shows a centrilobular linearbranching structure; centrilobular nodules (thick whitearrow) and cavities are also seen.

174

Figure 3 HRCT section at the apices of the lungs in a woman aged 51 years withsputum positive for acid fast bacilli. "Tree-in-bud" appearances (white arrows) andaccompanying cavities are shown.

patient with lobar consolidation, and onepatient with a history of previous anti-tuberculous therapy. Since lympho-haematogenous dissemination occurs in miliarytuberculosis, we could not detect findings ofearly bronchogenic spread. In the patient withlobar consolidation we thought that small nod-ular lesions had resulted in lobular con-solidation and that these had led to segmentaland finally lobar consolidation. In the thirdpatient, although we could not observe cen-trilobular nodules, branching linear structures,or "tree-in-bud" appearance, a macronoduleand an area of lobular consolidation were pres-ent. In these three patients the diagnosis ofactive pulmonary tuberculosis had been con-firmed by positive acid fast bacilli on smearand/or culture.The characteristic findings of bronchogenic

spread were detected in 30 of31 patients (97%)with postprimary tuberculosis (all except theindividual with lobar consolidation). None ofthese lesions was detected in patients with in-active pulmonary tuberculosis. In one of thetwo cases excluded, a macronodule was presentand subsegmental consolidation was shown inthe other case. Bronchiectasis in the upperlobes was present in both.Pulmonary consolidation (fig 1) was seen

in 14 (44%) patients with active tuberculosis(p<O0OO1) and it was not encountered in thegroup with inactive disease. A ground glassappearance was present in 12 (38%) of thepatients with active tuberculosis and was com-mon in areas surrounding nodular lesions(p<005). This appearance was seen, associatedwith fibrotic changes, in a minority of patientswith inactive disease.

Cavitation was present in 16 (50%) of thepatients with active tuberculosis (p<OOO1).The cavity walls ranged from very thin andsmooth to very thick and nodular. Bron-chogenic spread of tuberculous material fromcavities to other parts of the lung was alwayspresent in patients with active disease. Ca-vitation was present in four of the patients withinactive disease. Most cavity walls were thinand evidence of bronchogenic spread was ab-sent. Five patients with active tuberculosis hadnon-calcified mediastinal lymphadenopathy(>10 mm). In most of these cases the findingsof disseminated pulmonary tuberculosis werepresent. In patients with inactive tuberculosisno mediastinal lymphadenopathy over 10mmwas detected.

Pleural effusion was found in two of thepatients with active tuberculosis. The effusionwas associated with parenchymal infiltrationand was positive for acid fast bacilli; pleuralbiopsy confirmed the diagnosis.

Multiple discrete miliary nodules (fig 5) wereseen in an even distribution throughout thelung in one case with miliary tuberculosis. Thediagnosis was confirmed by a positive culturefor Mycobacterium tuberculosis.

HRCT FEATURES IN INACTIVE DISEASEThe HRCT findings in inactive pulmonarytuberculosis are summarised in the table. The

399



/T


ownloaded from


Hatipoglu, Osma, Manisali, U,can, Balci, Akkoflu, Akpinar, Karlikaya, Yiiksel

i

Figure 4 HRCT section at the level of the bifurcation of the trachea in a man aged 36years with sputum positive for acid fast bacilli. Increased spatial resolution offers theopportunity of demonstrating drainage bronchioles of tuberculous cavities (long whitearrow); inferior to the cavity a macronodule (thick white arrow) is seen. Bar indicates1 cm.

Ad

Figure 5 HRCT section targeted to the left upper lung ina woman aged 60 years with culture positive for acid fastbacilli. In this patient with miliary tuberculosis multiplenodules can be seen scattered throughout the lung.Subpleural nodules (black arrow), centrilobular nodules(curved arrow), and nodulaity of interlobular septa(white arrow) are also present. Bar indicates 1 cm.

most common abnormalities included fibroticchanges in all patients (p<0001), broncho-vascular distortion in 32 (94%) (p<0001), andemphysema in 28 (82%) (p<0-001) (fig 6).Whenever bronchovascular distortion was

severe, the associated emphysema had an ap-pearance similar to that of paracicatricial em-

physema. In areas where bronchovasculardistortion was not dense, emphysema appearedas local hyperlucent areas and, in some areas,included centrilobular densities (central dots).Centrilobular densities in the middle of an area

of emphysema could be easily discriminatedfrom centrilobular nodules in patients with act-ive tuberculosis. Bronchiectasis was detectedin 18 (56%) of the patients with active diseaseand in 24 (71 %) of those with inactive disease.

Traction bronchiectasis was the predominantfeature in patients with inactive tuberculosisand in those cases with active disease and ahistory of previous tuberculosis.

Bronchial wall and interlobular septal thick-ening were detected adjacent to areas with otherfindings associated with activity in patients withactive disease. These appearances were seenassociated with the residual fibrotic changes insome patients with inactive disease.While calcified mediastinal/hilar lymph node

enlargement was present in seven patients, par-enchymal calcification in 14, and pleural thick-ening or retraction in 20 individuals with activetuberculosis, they were present in 15 (p<0O05),18, and 24 patients with inactive tuberculosis,respectively.Using discriminant analysis to determine the

HRCT features which best correlated with ac-tivity we found that the variables, ordered bymagnitude of correlation, were: centrilobularnodule (0 703), "tree in bud" appearance(0 539), macronodule (0A467), broncho-vascular distortion (-0-361), fibroticchanges (-0 290), parenchymal calcifica-tion (-0 262), bronchiectasis (-0-214),consolidation (0 199), lymphadenopathy(>10 mm) (0188), ground glass opacity(O-159), interlobular septal thickening (0 154),emphysema (-0-152), bronchial wall thick-ening (0 098), cavity (0 096), pleural effusion(0-081), pleural thickening (-0 045), miliarynodule (0,040), and calcified mediastinallymph node enlargement (-0-017).

COMPARISON WITH CHEST RADIOGRAPHYChest radiographic findings consistent withpulmonary tuberculosis were present in 17 ofthe 18 cases with newly diagnosed active pul-monary tuberculosis. In one patient the chestradiograph showed minimal change but fea-tures associated with activity were recognisedusing HRCT. New lesions around old fibroticlesions were easily demonstrated by chest radio-graphy in seven of 14 cases with a historyof previous antituberculous treatment and asuspicion of reactivation in the remaining sevencases. HRCT could easily distinguish newlesions from old fibrotic lesions. On chest radio-graphy lesions were observed in both lungs in13 patients and unilateral changes in 19. Chestradiography could not recognise focal bron-chogenic spread in the upper zone of the lungcontralateral to the site of major abnormalityin one case. Cavity formation was noted onchest radiography and onHRCT in 16 patients.Chest radiography was erroneously evaluatedas demonstrating cavity formation in onepatient. In another patient chest radiographycould not recognise the cavity which was shownon the HRCT scan. In 15 patients cavity form-ation was present both on the chest radiographand the HRCT scan. Multiple cavities wereseen on the chest radiographs of four patientsand the HRCT scans of nine.

DiscussionPostprimary or reactivation tuberculosis is themost common form of pulmonary tuberculosis

400



/T


ownloaded from


High resolution computed tomographic findings in pulmonary tuberculosis

Figure 6 HRCT section at the subapical level in a patient aged 62 years with inactivetuberculosis. On the left, irregular fibrotic changes causing parenchymal distortion andtraction bronchiectasis are seen. There are also emphysematous areas.

in adults. Various investigations have con-

cluded that postprimary tuberculosis resultsfrom reactivation of a previously dormant prim-ary infection in 90% of cases.7-9 A few casesrepresent a continuation of the primarydisease.810 Rarely, postprimary tuberculosis isa result ofexogenous superinfection on inactiveor even active original infection. The earlyparenchymal infiltrate of postprimary tuber-culosis results from an intense inflammatoryreaction in the hypersensitive host." This maylead to caseous necrosis which over time is aptto slough into a bronchus, leaving a cavity."'The most commonly involved areas ofthe lungsin postprimary disease are the apical and pos-terior segments of an upper lobe.'-310-12 Bron-chogenic dissemination is the most common

means of spread in postprimary tuberculosis.The typical chest radiographic finding in post-primary tuberculosis is an acinar pattern - thatis, multiple fluffy nodules roughly 5 mm indiameter.'0 Such a radiographic pattern hasbeen reported in various diseases.'013 Itoh etall4 showed in their necroscopic studies thatthe disease develops initially in respiratorybronchioles followed by surrounding focal in-flammation, and that these pathological pro-cesses are reflected in the nodules seen on

radiological examination of the lung. Smallernodules evolved into larger lesions whose dia-meter was the same as that of the pulmonaryacinus, and it was inferred that direct extensionof inflammation through the pores of Kohn or

aspiration of inflammatory materials into thepulmonary acinus was the mechanism involved.In addition, caseous necrosis was shown toaffect the walls of the bronchioles and adjacentalveoli. Murata et al"5 showed that these peri-bronchiolar intralobular nodules could be re-

cognised as centrilobular nodules on HRCTscans and that, in tuberculosis, they were foundaround the respiratory bronchioles.Im et al'6 reported that the earliest HRCT

finding ofbronchogenic dissemination oftuber-culosis was a centrilobular nodule or branchinglinear structure 2-4 mm in diameter. Thesecentrilobular lesions consisted of solid caseous

material within or around the terminal or res-piratory bronchioles. Multiple branching linearstructures of similar calibre originating froma single "stalk" (the "tree-in-bud" appearance)were commonly seen in patients with extensivebronchogenic spread. Terminal tufts of the"tree-in-bud" structure might represent lesionswithin the bronchioles and alveolar ducts, whilethe "stalk" might represent a lesion that affec-ted the last order bronchus within the sec-ondary lobule. In addition they showed thatbronchial wall thickening occurred duringbronchogenic dissemination, that interlobularseptal thickening resulted from increasedlymphatic clearance, and that the ground glassappearance could be present around activelesions.

In several investigations chest radiographicfindings of bronchogenic spread have been re-ported in approximately 20% of instances.7 18With HRCT scanning we have detected fea-tures characteristic of bronchogenic spread in30 of 31 cases (97%) with postprimary tuber-culosis (all except for one case with lobar con-solidation). The results of our study areconsistent with those of Im et al. 6Our study has illustrated several advantages

ofHRCT scanning in cases of suspected activetuberculosis. HRCT demonstrated features as-sociated with disease activity in a newly diag-nosed patient with a normal chest radiograph,and in seven patients with indeterminateradiographs and a history of previous tuber-culosis. In only one of these eight patients wereacid fast bacilli detected by smear examinationof the sputum and/or bronchial washings.Thus, HRCT scanning enabled an early diag-nosis to be made and treatment started im-mediately by showing characteristic features ofearly bronchogenic spread. HRCT scanningwas also found to be superior to chest radio-graphy in the detection of dissemination oflesions and the presence and number of cav-ities.Healing of tuberculosis lesions results in the

development of a fibrotic scar and calcificationwith loss of lung parenchyma volume. Rarely,if lesions heal without development of necrosis,no residual fibrotic changes occur. Im et al'6showed that healing of cavitatory lesions causedmore cicatricial change than in other areaswithout cavitation and that distortion ofbronc-hovascular structures, emphysema, fibroticbands, or bronchiectasis tended to increase inprevalence on follow up CT scans. In our studythe frequency of bronchovascular distortion,fibrotic changes, emphysema, and bronchi-ectasis in patients with inactive pulmonarytuberculosis was high. This high frequencymight be due to the presence of residual fibroticchanges on the chest radiographs of all patientsincluded in the study with the diagnosis ofinactive pulmonary tuberculosis.

In conclusion, our data support the conceptthat HRCT scanning can be of value in situ-ations where chest radiographs do not con-tribute to the diagnosis of active disease andcannot distinguish new lesions from old fibroticchanges. If there is clinical suspicion, HRCTcan be used to show centrilobular lesions which

401



/T


ownloaded from


Hatipoglu, Osma, Manisali, Ufan, Balci, Akko1lu, Akpinar, Karlikaya, Yuiksel

are the most common findings of early bron-chogenic spread. This is particularly helpful inthe management of the disease because in only53% of our patients were acid fast bacilli pres-

ent in the sputum or bronchial washings.HRCT aided decisions about treatment.

1 Miller WT, MacGregor RR. Tuberculosis: Frequency ofunusual radiographic findings. AJR 1978;130:867-75.

2 Miller WT. Tuberculosis in adults. Postgrad Radiol 1981 ;1:147-67.

3 Woodring JH. The radiographic manifestations of pul-monary tuberculosis. Ky Med Assoc 1984;82: 17-23.

4 Naidich DP, McCouley DI, Leitman BS, Genesier NB,Hulnick DH. CT of pulmonary tuberculosis. In: Siegel-man SS, ed. Computed tomography of the chest. New York:Churchill Livingstone, 1984:175-217.

5 Kuhlman JE, Deutsch JH, Fishman EK, Siegelman SS. CTfeatures of thoracic mycobacterial disease. RadioGraphics1990;10:413-31.

6 Ikezoe J, Takeuchi N, Johkoh T, Kohno N, Tomiyama N,Kozuka T, et al. CT appearance ofpulmonary tuberculosisin diabetic and immunocompromised patients. AJ7R 1992;159:1175-9.

7 Stead WW. Pathogenesis ofthe sporadic case oftuberculosis.N Engl Med 1967;277:1008-12.

8 Stead WW, Kerby GR, Schlueter DP, Jordahl CW. Theclinical spectrum of primary tuberculosis in adults: con-

fusion with reinfection in the pathogenesis of chronic.tuberculosis. Ann Intern Med 1968;68:731-45.

9 Geppert EF, Leff A. The pathogenesis of pulmonary andmiliary tuberculosis. Arch Intern Med 1979;139:1381-3.

10 Fraser RG, Pare JAP. Diagnosis of diseases of the chest. Vol.2. 2nd edn. Philadelphia: Saunders, 1978: 731-61.

11 Goodwin RA, DesPrez RM. Apical localization of pul-monary tuberculosis, chronic pulmonary histoplasmosis,and progressive massive fibrosis of the lung. Chest 1983;83:801-5.

12 Jacobson HG, Shapiro JH. Pulmonary tuberculosis. RadiolClin North Am 1963;1:411-27.

13 Ziskind MM, Weill H, Payzant AR. The recognition andsignificance of acinus filling process of the lungs. Am RevRespirDis 1963;87:551-9.

14 Itoh H, Tokunaga S, Asamato H, Furata M, Funamoto Y,Kitachi M, et al. Radiologic-pathologic correlation ofsmalllung nodules with special reference to peribronchial nod-ules. AJrR 1978;130:233-31.

15 Murata K, Itoh H, Todo G, Kanaoka M, Noma S, Itoh T,et al. Centrilobular lesions of the lung: demonstration byhigh resolution CT and pathologic correlation. Radiology1986;161:641-5.

16 Im JG, Itoh H, Soo Shim Y, Lee HL, Ahn J, Han MC, etal. Pulmonary tuberculosis: CT findings, early active dis-ease and sequence change with antituberculous therapy.Radiology 1993;186:653-60.

17 Hadlock FP, Park SK, Awe RJ, Rivera M. Unusual radio-graphic findings in adult pulmonary tuberculosis. AJ7R1980;134:1015-8.

18 Woodring JH, Vandiviere HM, Fried AM, Dillon ML,Williams TD, Melvin IG. Update: the radiographic fea-tures of pulmonary tuberculosis. A_JR 1986;146:497-506.

402



/T


ownloaded from


high resolution computedtomographic findings in...

Documents