Objective
To identify specific mammographic appearances that reduce the mammographic detection of breast cancer.
Materials and Methods
This study received institutional board review approval and all readers gave informed consent. A set of 60 mammograms each consisting of craniocaudal and mediolateral oblique projections were presented to 129 mammogram Breastscreen readers. The images consisted of 20 positive cases with single and multicentric masses in 16 and 4 cases, respectively (resulting in a total of 24 cancers), and readers were asked to identify and locate the lesions. Each lesion was then ranked according to a detectability rating (ie, the number of observers who correctly located the lesion divided by the total number of observers), and this was correlated with breast density, lesion size, and various descriptors of lesion shape and texture.
Results
Negative and positive correlations between lesion detection and density ( r = −0.64, P = .007) and size ( r = 0.65, P = .005), respectively, were demonstrated. In terms of lesion size and shape, there were significant correlations between the probability of detection and area ( r = 0.43, P = .04), perimeter ( r = 0.66, P = .0004), lesion elongation ( r = 0.49, P = .02), and lesion nonspiculation ( r = 0.78, P < .0001).
Conclusions
The results of this study have identified specific lesion characteristics associated with shape that may contribute to reduced cancer detection. Mammographic sensitivity may be adversely affected without appropriate attention to spiculation.
The detection of breast masses on mammograms is a difficult task for human observers or machines . The heterogeneity and potential variability of normal breast tissue often produce a number of localized findings that may simulate mass lesions or, depending on the observer, create distractions during the search process . Strategies must therefore be developed to ensure a reasonable performance in detection, and these should include the identification of lesion or image features that make the detection of the cancer more difficult. Such features could be lesion-specific such as size, shape, texture and local signal-to-noise ratio (SNR) or image-specific such as mammographic breast density and global SNR.
Even though incorrect radiographic positioning or exposure may be considered a principal culprit for missing cancers in mammography , previous findings suggest that a comprehensive investigation of other features is required . Bird and Majid and their colleagues found that missed cancers on mammograms tend to occur in denser breasts and with lesions that present as developing densities. Other authors found that the missed cancers were significantly lower in density and were smaller in size compared to those lesions more easily detected. The previous work implies that attenuation properties of the normal and abnormal structures within the breast can affect detection of disease; however, there are few data on the impact of other image radiographic appearances of disease such as overall lesion morphology and texture . This is surprising because radiologically, shape features of cancers such as spiculations are commonly used to differentiate between malignant and benign lesions and provide an indicator of the aggressiveness of an invasive tumor . Shape characteristics for sublesion elements, such as calcified regions, are well reported and details on calcification morphology, texture, area, perimeter, and elongation have been used to good effect in designing computer aided diagnosis (CAD) algorithms . Unfortunately, a large number of cancers present as mass lesions without calcifications .
Get Radiology Tree app to read full this article<
Materials and methods
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Data Analysis
Get Radiology Tree app to read full this article<
Table 1a
Lesion Shape Measurements
Lesion Feature
Feature Description Measure Measure Description Size Area The number of pixels contained within and including the boundary of the lesion . Size and shape Perimeter The circumferential distance around the lesion boundary . Elongation Eccentricity Measures the ratio of the major axis to minor axis of the best fit ellipse that outlines the lesion . Nonspiculation Solidity Describes the extent to which the external shape is smooth , calculated by the overall area divided by the convex area.
Table 1b
Lesion Texture Measurements
Lesion Feature Measure Measure Description Contrast ∗ GLCM Measures the local variations in intensity between a pixel and its neighbor within whole lesion . Homogeneity ∗ GLCM Measures the closeness of the distribution in the intensity between a pixel and its neighbor over the entire lesion .
Table 1c
Lesion Contrast Measurements
Lesion Measure Measure Description Lesion contrast † SNR local (Mean gray level value of lesion − mean gray level value of the local background that comprised a region of 1 diameter immediately outside the lesion)/√(standard deviation of the lesion) 2 + (standard deviation of the local background) 2 Lesion contrast † SNR global (Mean gray level value of lesion − mean gray level value of whole breast without the pectoralis major muscle)/√(standard deviation of the lesion 2 + standard deviation of whole breast without the pectoralis major muscle 2 )
GLCM, gray-level co-occurrence matrix; SNR, signal-to-noise ratio.
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Results
Get Radiology Tree app to read full this article<
Table 2
Details on Lesion Features
Feature Mean Minimum Maximum Breast density Lesion size Breast density 2.56 2 4 Diameter (mm) 11.44 6 25 Lesion shape measurements Area (pixels) 307.7 92 684 Perimeter 477.6 255 966 Lesion nonspiculation 0.82 0.43 0.98 Lesion elongation 0.67 0.41 0.98 Lesion texture measurements Contrast 0.006 0.002 0.01 Homogeneity 0.992 0.79 1 Lesion contrast measurements Local SNR 5.82 0.25 14.75 Global SNR 0.87 0.07 3.69
SNR, signal-to-noise ratio.
Table 3
The Correlation between the Probability of Lesion Detection and the Lesion Features
Feature_r_ Value_P_ Value Breast density lesion size Density−0.64.007 Diameter (mm)0.65.005 Lesion shape measurements Area (pixels)0.43.04 Perimeter0.66.0004 Lesion elongation0.49.02 Lesion nonspiculation0.78.0001 Lesion texture measurements Contrast 0.007 .97 Homogeneity 0.18 .39 Lesion contrast measurements Local SNR −0.29 .18 Global SNR −0.33 .11
SNR, signal-to-noise ratio.
Bold-type values indicate statistically significant findings.
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Discussion
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
References
1. Vyborny C.J., Doi T., O’Shaughnessy K.F., et. al.: Breast cancer: importance of spiculation in computer-aided detection. Radiology 2000; 215: pp. 703-707.
2. Taplin S.H., Rutter C.M., Finder C., et. al.: Screening mammography. AJR Am J Roentgenol 2002; 178: pp. 797-803.
3. Bird R.E., Wallace T.W., Yankaskas B.C.: Analysis of cancers missed at screening mammography. Radiology 1992; 184: pp. 613-617.
4. Majid A.S., de Paredes E.S., Doherty R.D., et. al.: Missed breast carcinoma: pitfalls and pearls. Radiographics 2003; 23: pp. 881-895.
5. Goergen S.K., Evans J., Cohen G.P., et. al.: Characteristics of breast carcinomas missed by screening radiologists. Radiology 1997; 204: pp. 131-135.
6. Bassett L.W.: Diagnosis of diseases of the breast.1997.W.B. SaundersPhiladelphia
7. Huynh P.T., Jarolimek A.M., Daye S.: The false-negative mammogram. Radiographics 1998; 18: pp. 1137-1154.
8. Pohlman S., Powell K.A., Obuchowski N.A., et. al.: Quantitative classification of breast tumors in digitized mammograms. Med Phys 1996; 23: pp. 1337-1345.
9. Rangayyan R.M., Mudigonda N.R., Desautels J.E.: Boundary modelling and shape analysis methods for classification of mammographic masses. Med Biol Eng Comput 2000; 38: pp. 487-496.
10. Shi J., Sahiner B., Chan H.P., et. al.: Characterization of mammographic masses based on level set segmentation with new image features and patient information. Med Phys 2008; 35: pp. 280-290.
11. Shen L., Rangayyan R.M., Desautels J.L.: Application of shape analysis to mammographic calcifications. IEEE Trans Med Imaging 1994; 13: pp. 263-274.
12. Jiang Y, Nishikawa RM, Papaioannou J. Requirement of microcalcifications detection for computerized classification of malignant and benign clustered microcalcifications, Proc. SPIE—Int Soc Opt Eng 1998; 3338:313–317.
13. Strickland R.N., Hahn H.I.: Wavelet transforms for detecting microcalcifications in mammograms. IEEE Trans Med Imaging 1996; 15: pp. 218-229.
14. Jain A.: Artificial intelligence techniques in breast cancer diagnosis and prognosis.2000.World ScientificRiver Edge, NJ xvii, 330 p. p.179
15. Sickles E.A.: Breast masses: mammographic evaluation. Radiology 1989; 173: pp. 297-303.
16. D’Orsi C.J., Kopans D.B.: Mammographic feature analysis. Semin Roentgenol 1993; 28: pp. 204-230.
17. Tabar L., Dean P.B.: Teaching atlas of mammography.2nd ed.1985.Georg Thieme VerlagStuttgart, Germany 15–136
18. Wallis M.G., Walsh M.T., Lee J.R.: A review of false negative mammography in a symptomatic population. Clin Radiol 1991; 44: pp. 13-15.
19. Yankaskas B.C., Schell M.J., Bird R.E., et. al.: Reassessment of breast cancers missed during routine screening mammography: a community-based study. AJR Am J Roentgenol 2001; 177: pp. 535-541.
20. Bassett LW, Gold RH. Breast cancer detection: mammography and other methods in breast imaging, 2nd ed. 1987.
21. American College of Radiology: Breast Imaging Reporting and Data System (BIRADS).3rd ed.1998.American College of RadiologyReston, VA
22. Harris J.R.: Staging and natural history of breast cancer.Harris J.R.Lippman M.E.Morrow M.Osborne C.K.Diseases of the breast.2000.Lippincott Williams & WilkinsPhiladelphia, PA:pp. 403-406.
23. Kallergi M.: Computer-aided diagnosis of mammographic microcalcification clusters. Med Phys 2004; 31: pp. 314-326.
24. David M., Catarious J., Baydush A.H., et. al.: Incorporation of an iterative, linear segmentation routine into a mammographic mass CAD system. Med Phys 2004; 31: pp. 1512-1520.
25. Manning D.J., Ethell S.C., Donovan T.: Detection or decision errors? Missed lung cancer from the posteroanterior chest radiograph. Br J Radiol 2004; 77: pp. 231-235.
26. Fenton J.J., Abraham L., Taplin S.H., et. al.: Effectiveness of computer-aided detection in community mammography practice. J NCI 2011; 103: pp. 1152-1161.
27. Obenauer S., Luftner-Nagel S., von Heyden D., et. al.: Screen film vs full-field digital mammography: image quality, detectability and characterization of lesions. Eur Radiol 2002; 12: pp. 1697-1702.
28. Karahaliou A., Skiadopoulos S., Boniatis I., et. al.: Texture analysis of tissue surrounding microcalcifications on mammograms for breast cancer diagnosis. Br J Radiol 2007; 80: pp. 648-656.