HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2283011906v1 228/3/851    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Malich, A. Articles by Kaiser, W. A. Search for Related Content PubMed PubMed Citation Articles by Malich, A. Articles by Kaiser, W. A.

Influence of Breast Lesion Size and Histologic Findings on Tumor Detection Rate of a Computer-aided Detection System¹

¹ From the Institute of Diagnostic and Interventional Radiology, Friedrich-Schiller-University Jena, Bachstr 18, 07740 Jena, Germany. Received November 27, 2001; revision requested February 25; final revision received December 26; accepted January 10, 2003. Address correspondence to A.M. (e-mail: ansgar.malich@med.uni-jena.de).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PURPOSE: To evaluate associations between histopathologic findings, tumor size, and detection rate of malignant mammographic findings by using a computer-aided detection (CAD) system.

MATERIALS AND METHODS: The study included 208 mammographically detected histologically proven malignant breast lesions in 208 women. Findings were 150 masses and 114 microcalcifications; 56 lesions showed both findings; 94 lesions, mass only; and 58 lesions, microcalcification only. CAD was used to evaluate mammograms in two views retrospectively. Also, corresponding histopathologic findings and lesion size were evaluated. CAD marks were considered positive if, on at least one view, they correctly identified the corresponding mammographic lesion location.

RESULTS: Ninety percent (135 of 150) of masses and 93.0% (106 of 114) of microcalcifications were marked correctly by the CAD system. Overall tumor detection rate was 93.8% (195 of 208). Size-related detection rate for masses was 83.3% (25 of 30) for lesions up to 10 mm, 100% (45 of 45) for lesions 11–20 mm, 100% (46 of 46) for lesions 21–30 mm, 83.3% (10 of 12) for lesions 31–40 mm, and 52.9% (nine of 17) for lesions larger than 40 mm. Size-related tumor detection rate for microcalcifications was 92.5% (37 of 40) for microcalcifications up to 10 mm, 93.1% (27 of 29) for lesions 11–20 mm, 100% (20 of 20) for lesions 21–30 mm, 87.5% (seven of eight) for lesions 31–40 mm, and 88.2% (15 of 17) for larger microcalcifications. Detection rates for mammographically visible masses (invasive ductal carcinoma, invasive lobular carcinoma, invasive tubular carcinoma, noninvasive cancers, mucinoid cancers, and others) were 92.3% (84 of 91), 89.3% (25 of 28), 75.0% (six of eight), 100% (15 of 15), 33.3% (one of three), and 80.0% (four of five), respectively. Detectability rates for mammographically visible areas suspicious for microcalcifications (invasive ductal carcinoma, invasive lobular carcinoma, invasive tubular carcinoma, and noninvasive cancers) were 92.3% (60 of 65), 100% (eight of eight), 100% (five of five), and 91.9% (31 of 34), respectively. Highest overall detection rates were observed for invasive ductal carcinomas (96.6% [112 of 116]) and noninvasive cancers (92.9% [39 of 42]).

CONCLUSION: Highest detection rates were observed for 10-30-mm tumor masses and for invasive ductal carcinomas and noninvasive cancers.

© RSNA, 2003

Index terms: Breast, diseases, 00.327, 00.329 • Breast neoplasms, calcification, 00.812 • Breast neoplasms, localization, 00.125 • Breast radiography, 00.111, 00.115 • Computers, diagnostic aid, 00.1299

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Mammography is a well-established method for early detection of breast cancer. However, mammographic interpretation varies with the experience of the radiologist. Results of several studies have shown that breast cancer detection rates can be improved by up to 15% by using a second reader (1–5). Because of financial, technical, and logistic restraints, however, it may not be feasible to perform second readings routinely. Efforts were made asearly as 1967 to develop a computer-aided diagnosis (CAD) system for mammography (6). These systems are designed to help radiologists in the detection of suspicious masses and microcalcifications earlier and more accurately during screening mammography. Funovics et al showed that the sensitivity for breast cancer detection increases significantly when a radiologist uses a CAD system (7). Since the histopathologic features and size of a cancerous lesion influence radiologist sensitivity (8–11), they may also influence the tumor detection rate of a CAD system. In addition to these characteristics, the differences in case selection criteria may also affect CAD performance, which may explain the differences in sensitivity values obtained in previous studies (8–11). To date, only a few investigators have assessed the association between histologic findings, tumor size, and detection rate for CAD systems. Their results suggest that the tumor detection rate may be associated with lesion size (8,12,13). The aim of the present study was to evaluate the association between histopathologic findings, tumor size, and detection rate of malignant mammographic findings by using the Second Look (CADX, Laval, Quebec, Canada) CAD system.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
A retrospective analysis of mammograms from 208 patients (with 208 mammographically detected histologically proven breast cancer lesions) was performed. Every fifth biopsy-proven cancer case from September 1992 to March 2000 was selected by using the institutional tumor case sampler. Mean age of the women was 55 years 5 months (SD, 30 years 7 months).

All mammograms that showed tumor-induced changes that were histologically proven in our department and that led to the diagnosis of breast cancer had been stored separately in the internal tumor case sampler. Most of these cases were cancers detected at screening. A wide range of tumor sizes was included in the study, without any preselection. Cases first detected in external mammography centers were excluded from the tumor sampler. All patients with more than one mammographically visible suspicious lesion per image and all patients with bilateral cancers were excluded from the study to avoid any methodologic problems. No other preselection was performed. Tumors of all sizes with various histologic findings were included, and all lesions were verified surgically. The local ethical board gave its approval for the study, and informed consent was obtained.

Two hundred eight malignant lesions were verified histopathologically, as shown in Table 1. Most of the lesions were invasive ductal carcinomas. Within the subgroup of other types of cancers, metastasis, metaplastic cancers, undifferentiated invasive cancer, and neuroendocrine tumor were included. Invasive tubular carcinomas and mucinoid cancers were uncommon (10 and four cases, respectively, of 208).

The largest mammographically visible diameter of the detected masses and microcalcifications was also determined by two experienced radiologists (A.M., C.M.) in consensus. For statistical analysis, lesion sizes of masses and microcalcifications were divided into five groups: group 1, diameter of less than 10 mm (30 masses, 40 microcalcifications); group 2, diameter between 10 and 20 mm (45 masses, 29 microcalcifications); group 3, diameter between 21 and 30 mm (46 masses, 20 microcalcifications); group 4, diameter between 31 and 40 mm (12 masses, eight microcalcifications); and group 5, larger than 40 mm (17 masses, 17 microcalcifications).

Tumor Detection by the CAD System
All mammographic examinations were conducted with Mammodiagnost UC (Philips, Best, the Netherlands) or Senographe DMR (GE Medical Systems, Milwaukee, Wis) units. Each mammographic examination consisted of acquisition of two images: craniocaudal and mediolateral oblique views of the right or left breast with one suspicious lesion each. The mammograms were processed in two views by the Second Look CAD system.

In practice, the system is used by the radiologist after initial review of mammograms. The images are then loaded into the digitizer of the CAD system for computer processing. Second Look uses proprietary algorithms to detect potential areas of concern and generates a printout of the screening mammogram on which potentially suspicious masses and microcalcifications are identified. The system marks potential areas of concern in two ways: an ellipse for masses and a rectangle for microcalcifications, both corresponding to the approximate size of the lesion. With the CAD system, there is no size limitation for masses or microcalcifications, which are determined independently from each other (Figure), as stated in the Second Look manual.

View larger version (71K): [in this window] [in a new window] [Download PPT slide]   Example of a CAD printout of a mammographic analysis of both breasts; a rectangle is set to mark microcalcifications, and an ellipse is set to mark masses. From left to right: right and left craniocaudal views of the right and left breasts, mediolateral oblique views of the right and left breasts.

Statistical Analysis
The Fisher-Freeman-Halton exact test (Monte Carlo testing) and Kruskal-Wallis test were used and Kendall rank correlation was calculated to investigate whether there were statistically significant differences or associations between detection rate and lesion size or histologic findings. Exclusion of all patients with more than one lesion per image or more than one cancer simplified statistical analysis. Since the marking of masses and microcalcifications is performed separately by the system, all statistical analyses were done separately for microcalcifications and masses, including Fisher-Freeman-Halton exact testing if necessary. Overall tumor detection was scored as positive if any sign was marked on the location of the suspicious lesion by the CAD system on any image. A P value of less than .05 was considered to indicate a statistically significant difference. t tests were used to determine whether mean sizes of masses and microcalcifications with different histologic findings differed significantly. We used SPSS version 9.0.0 (SPSS, Chicago, Ill) and StatXact-4 (Cytel Software, Cambridge, Mass) statistical software.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
General Tumor Detection Rate
Of the 150 mammographically suspicious malignant masses, 135 were marked correctly by the CAD system (90.0% detection rate). In addition, 106 of 114 mammographically suspicious malignant microcalcifications were marked correctly by the system (93.0% detection rate).

The overall tumor detection rate (including any mark on the tumor location in at least one view) was 93.8%, with 195 of 208 cancers marked in at least one view with at least one sign. The size characteristics of the undetected lesions are described in Table 2. Most undetected masses were large malignancies (eight of 15 larger than 41 mm). Rounded densities were observed in six of 15 cases. Additional edema of the skin was visible in four of 15 cases. A spiculated mass was visible in seven of 15 cases.

Histologic Findings and Detection Rate
Table 3 shows the detection rate of the malignant lesions with histologic classification. The lowest overall detection rates were observed for mucinoid cancers (75.0%) and other malignancies (80.0%), but sample sizes for these types were low.

The CAD detection rate for microcalcifications was always higher than 90.0% for all histologic subgroups.

When comparing the detection rates (marking of the tumor with any sign), no significant differences were found by using Fisher-Freeman-Halton testing (Monte Carlo testing, P = .17).

Furthermore, it was verified by using t tests that sizes between invasive ductal carcinoma, invasive lobular carcinoma, invasive tubular carcinoma, and noninvasive cancers did not differ significantly (masses, P = .26; microcalcifications, P = .84).

When suspicious masses and microcalcifications were analyzed separately, tumor detection rate did vary significantly for the different histologic findings of masses (Fisher-Freeman-Halton Monte Carlo exact testing, P = .03), whereas detection rate of microcalcifications did not.

Mass Sizes
The distribution of mass sizes as classified into histologically defined groups is shown in Table 4.

Results of Kendall rank correlation did not indicate a statistically significant value for tumor size and detection rate.

Results of the Kruskal-Wallis test indicated a significantly different tumor detectability rate, which decreased with increasing mass size (P < .001). Very large masses are significantly less detectable by the CAD system.

Microcalcification Sizes
The distribution of microcalcification sizes according to histologic findings is shown in Table 5.

Results of the Kruskal-Wallis test showed that tumor detection rate did not differ significantly with increasing microcalcification size (P = .66).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Overall Tumor Detection Rate
A CAD system assists the radiologist in the early detection of breast cancer by highlighting suspicious areas. A true-positive mark by the CAD system is defined as correct identification of the cancer in at least one of the two views. The Second Look CAD system showed a detection rate of 90.0% for masses and 93.0% for microcalcifications. The overall detection rate of Second Look was 93.8%.

Several studies of tumor detection rate and sensitivity of various CAD systems have been conducted (10–17). The case selection protocols from these studies differ considerably. Since case selection can affect substantially the evaluation of a CAD system, it is therefore problematic to compare the results of these previous studies (18). However, the overall detection rate was >90.0% in all cases. An improvement in the accuracy of the radiologist could be verified (14). Brem et al stated in congruence to our findings (by using other selection criteria) that mass size does influence detection rate (13).

In the present study, the CAD system demonstrated a promising detection rate, especially considering that the most common types of lesions at the time of first detection are between 10 and 30 mm.

Analysis of Histologic Findings
If CAD detectability is defined as marking the tumor with any sign (for mass or microcalcification), the detection rate was best for invasive ductal cancers at 96.6% (112 of 116) and for intraductal noninvasive cancers at 92.8% (39 of 42). This might be associated with the fact that invasive ductal carcinomas often show both signs of malignancy—mass and microcalcification—whereas the rate of microcalcifications observed with mucinoid cancer is lower and is stated in the literature as being three of 22 cases (20). The lowest overall CAD detection rate was observed for mucinoid cancers (75.0%, three of four) and invasive tubular carcinomas (80%, four of five), although low sample sizes do not allow further statistical interpretation.

Mucinoid cancers and squamous cell carcinoma are rare malignancies of the breast, and a high number (17.0%) of mucinoid cancers are mammographically occult, as reported by Goodman et al (19). These cancer types are associated with dense ill-defined or well-defined masses, mainly with regular borders (19–21). Invasive tubular carcinomas are usually irregularly shaped masses with spiculated margins, central densities, and spicules longer than the diameter of the central lesion (22). These large spicules might appear as a benign lesion. The low number of mucinoid adenocarcinomas and invasive tubular cancers included in the study does not allow any further statistical interpretation of the results; however, a significant altered detection rate could be verified, compared with the most common histologic findings. Most common malignancies, however, did not show any statistically significant alteration of detection rate. This can be viewed as promising, especially considering that the detection of invasive lobular carcinomas is sometimes difficult for the radiologist (23). The CAD system was designed to identify more common types of histologic findings because of the way these systems are developed and trained.

Analysis of Size-related Findings
When the sizes of malignant lesions are considered separately, the CAD system correctly identified all lesions ranging from 10 to 30 mm. Of note, this size group is the most relevant in screening mammography, as well as in second readings, because the mean size of most common cancers (such as invasive ductal carcinoma and invasive tubular carcinoma) detected at screening mammography is between 10 and 30 mm (24). The CAD system showed the lowest detection rates for cancers larger than 40 mm in their maximal diameter. Most of the malignancies larger than 40 mm were mammographically visible and palpable, and additional ultrasonographic examination was performed. In particular, some invasive lobular carcinomas and mucinoid cancers can be very large, resulting in poor mammographic detectability for the radiologist and also for the CAD system (19). It should be noted that because of technical reasons, other CAD systems usually have a size limitation (6–32 mm) for the detection of masses (7,17). This limitation for very large lesions may also exist in other CAD systems.

False-Positive Rate
The false-positive rate was underestimated in the present study, since the selection criteria included exclusively malignant lesions (25). The maximum number of marks per image is limited by the system. Although this maximum is not typically reached, the presence of malignant findings on all scanned images does lower the theoretically achievable number of false-positive marks compared with those in a normal population, as shown in a study by Malich et al (25). However, the false-positive number of microcalcification marks associated with a high tumor detection rate of microcalcifications should be judged as promising, especially in comparison to those obtained with other systems. As reported by Malich et al, the false-positive rate for masses depends on the examined group and was calculated as a mean of 0.97 mass mark per image, which is a high value (25).

Mammography is the single most effective method of early detection for breast cancer, since mammograms can demonstrate cancers several years before physical signs develop.

The results obtained with the Second Look CAD system in the present study are promising, with a high overall tumor detection rate. The CAD system showed the best detection rate for cancers between 10 and 30 mm (100%); however, lesions that were very large and very subtle were detected poorly by the system. In conclusion, we can assume that size is an important parameter in mammographically visible lesions that influences considerably the tumor detection rate, whereas most common histologically associated differences do not.

	FOOTNOTES

 
Abbreviation: CAD = computer-aided detection

Author contributions: Guarantors of integrity of entire study, A.M., W.A.K.; study concepts, A.M., C.M., T.B.; study design, A.M., D.S., C.M.; literature research, A.M., M. Fleck; clinical studies, A.M., C.M., M. Facius, M. Fleck, S.O.P.; data acquisition, A.M., C.M., M. Facius, T.B.; data analysis/interpretation, A.M., D.S., C.M., T.B., S.O.P.; statistical analysis, D.S., A.M.; manuscript preparation, A.M., D.S., T.B., C.M., M. Facius; manuscript definition of intellectual content, W.A.K., S.O.P., M. Facius, M. Fleck; manuscript editing, revision/review, and final version approval, all authors

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Bird RE. Professional quality assurance for mammography screening programs. Radiology 1990; 177:8-10.
2. Chan HP, Doi K, Vyborny CJ, et al. Improvement in radiologist’s detection of clustered MC on mammograms: the potential of computer-aided diagnosis. Invest Radiol 1990; 25:1102-1110.[CrossRef][Medline]
3. Zheng B, Chang YH, Staiger M, Good W, Gur D. Computer-aided detection of clustered MC in digitized mammograms. Acad Radiol 1995; 2:655-662.[CrossRef][Medline]
4. Chan HP, Sahiner B, Helvie MA, et al. Improvement of radiologists’ characterization of mammographic masses by using computer-aided diagnosis: an ROC study. Radiology 1999; 212:817-827.[Abstract/Free Full Text]
5. Thurfjell EL, Lernevall KA, Taube AAS. Benefit of independent double reading in a population-based mammography screening program. Radiology 1994; 191:241-244.[Abstract/Free Full Text]
6. Winsberg F, Elkin M, Macy J, Bordaz V, Weymouth W. Detection of radiographic abnormalities in mammograms by means of optical scanning and computer analysis. Radiology 1967; 89:211-215.
7. Funovics M, Schamp S, Lackner B, Wunderbaldinger P, Lechner G, Wolf G. Computerassistierte diagnose in der mammographie: das R2 image checker—system in der detektion spikulierter läsionen. Wien Med Wochenschr 1998; 148:321-324.[Medline]
8. Ehrenstein T, Kenzel PP, Hadijuana J, et al. Computer assisted diagnosis in mammography: evaluation of an expert system. Eur Radiol 2000; 10(suppl):117.
9. Marx C, Schütze B, Fleck M, O’Shaughnessy K, Kaiser WA. Computer aided diagnosis in mammography. Eur Radiol 1997; 7(suppl):82.[CrossRef][Medline]
10. Jiang Y, Nishikawa RM, Schmidt RA, Metz CE, Doi K. Comparison of independent double reading and computer-aided diagnosis (CAD) for the diagnosis of breast lesions (abstr). Radiology 1999; 213(P):323.
11. Thurfjell E, Thurfjell MG, Egge E, Bjurstam N. Sensitivity and specificity of computer-assisted breast cancer detection in mammography screening. Acta Radiol 1998; 39:384-388.[Medline]
12. Chang YH, Zheng B, Gur D. Computer-aided detection of clustered microcalcifications on digitized mammograms: a robustness experiment. Acad Radiol 1994; 4:415-418.
13. Brem RF, Schoonjans JM, Hoffmeister J, Raza S, Baum JK. Evaluation of breast cancer with a computer-aided detection system by mammographic appearance, histology and lesion size (abstr). Radiology 2000; 217(P):400.
14. Warren Burhenne LJ, Wood SA, D’Orsi CJ, et al. Potential contribution of computer-aided detection to the sensitivity of screening mammography. Radiology 2000; 215:554-562.[Abstract/Free Full Text]
15. Ibrahim N, Fujita H, Hara T, Endo T. Automated detection of clustered MC on mammograms: CAD system application to MIAS database. Phys Med Biol 1997; 42:2577-2589.[CrossRef][Medline]
16. Veldkamp WJ, Karssmeijer N. Accurate segmentation and contrast measurement of microcalcifications in mammograms: a phantom study. Med Phys 1998; 25:1102-1110.[CrossRef][Medline]
17. Sittek H, Perlet C, Helmberger R, Linsmeier E, Kessler M, Reiser M. Computer assisted analysis of mammograms in routine clinical diagnosis. Radiologe 1998; 38:848-852. [German].[CrossRef][Medline]
18. Nishikawa RM, Giger ML, Doi K, et al. Effect of case selection on the performance of computer-aided detection schemes. Med Phys 1994; 21:265-269.[CrossRef][Medline]
19. Goodman DN, Boutross-Tadross O, Jong RA. Mammographic features of pure mucinous carcinoma of the breast with pathological correlation. Can Assoc Radiol J 1995; 46:296-301.[Medline]
20. Memis A, Ozdemir N, Parilad M, Ustun EE, Erhan Y. Mucinoid (colloid) breast cancer: mammographic and US features with histologic correlation. Eur J Radiol 2000; 35:39-43.[CrossRef][Medline]
21. Samuels TH, Miller NA, Manchul LA, De Freitas G, Panzarella T. Squamous cell carcinoma of the breast. Can Assoc Radiol J 1996; 47:177-182.[Medline]
22. Sheppard DG, Whitman GJ, Huynh PT, Sahin AA, Fornage BD, Stelling CB. Tubular carcinoma of the breast: mammographic and sonographic features. AJR Am J Roentgenol 2000; 174:253-257.[Abstract/Free Full Text]
23. Uchiyama N, Miyakawa K, Moriyama N, Kumazaki T. Radiographic features of invasive lobular carcinoma of the breast. Radiat Med 2001; 19:19-25.[Medline]
24. Porter PL, El-Bastawissi AY, Mandelson MT, et al. Breast tumor characteristics as predictors of mammographic detection: comparison of interval- and screen-detected cancers. J Natl Cancer Inst 1999; 91:2020-2028.[Abstract/Free Full Text]
25. Malich A, Marx C, Facius M, Boehm T, Fleck M, Kaiser WA. Tumor detection rate of a new commercially available computer-aided detection (CAD) system. Eur Radiol 2001; 12:2454-2459.

This article has been cited by other articles:

				S. J. Kim, W. K. Moon, N. Cho, J. H. Cha, S. M. Kim, and J.-G. Im Computer-aided Detection in Full-Field Digital Mammography: Sensitivity and Reproducibility in Serial Examinations Radiology, December 1, 2007; 246(1): 71 - 80. [Abstract] [Full Text] [PDF]

				R. L. Ellis, A. A. Meade, M. A. Mathiason, K. M. Willison, and W. Logan-Young Evaluation of Computer-aided Detection Systems in the Detection of Small Invasive Breast Carcinoma Radiology, October 1, 2007; 245(1): 88 - 94. [Abstract] [Full Text] [PDF]

				S. K. Yang, W. K. Moon, N. Cho, J. S. Park, J. H. Cha, S. M. Kim, S. J. Kim, and J.-G. Im Screening Mammography-detected Cancers: Sensitivity of a Computer-aided Detection System Applied to Full-Field Digital Mammograms Radiology, July 1, 2007; 244(1): 104 - 111. [Abstract] [Full Text] [PDF]

				J. J. Fenton, S. H. Taplin, P. A. Carney, L. Abraham, E. A. Sickles, C. D'Orsi, E. A. Berns, G. Cutter, R. E. Hendrick, W. E. Barlow, et al. Influence of Computer-Aided Detection on Performance of Screening Mammography N. Engl. J. Med., April 5, 2007; 356(14): 1399 - 1409. [Abstract] [Full Text] [PDF]

				S. J. Kim, W. K. Moon, N. Cho, J. H. Cha, S. M. Kim, and J.-G. Im Computer-aided Detection in Digital Mammography: Comparison of Craniocaudal, Mediolateral Oblique, and Mediolateral Views Radiology, December 1, 2006; 241(3): 695 - 701. [Abstract] [Full Text] [PDF]

				V. R. Pai, N. E. Gregory, A. E. Swinford, and M. Rebner Ductal Carcinoma in Situ: Computer-aided Detection in Screening Mammography Radiology, December 1, 2006; 241(3): 689 - 694. [Abstract] [Full Text] [PDF]

				J. C. Dean and C. C. Ilvento Improved cancer detection using computer-aided detection with diagnostic and screening mammography: prospective study of 104 cancers. Am. J. Roentgenol., July 1, 2006; 187(1): 20 - 28. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2283011906v1 228/3/851    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Malich, A. Articles by Kaiser, W. A. Search for Related Content PubMed PubMed Citation Articles by Malich, A. Articles by Kaiser, W. A.