Use of a Method for Contour Analysis of Radiation Images of Malignant Breast Tumors on the Basis of Retrospective Material

Objective. To enhance the reliability of visual analysis of X-ray mammograms, by applying the mathematical models of neoplasms and a method for their processing based on the mathematical apparatus of contour analysis.

Material and methods. Two data sets were generated from X-ray mammograms obtained from 38–82 year old patients at routine examinations in the Republican Oncology Dispensary. The first set contained 100 packages of X-ray mammographic images that failed to reveal abnormal malignant changes. The second set was represented by 168 packages of X-ray mammographic images showing morphologically verified breast cancer. All the packages of mammographic images are presented in the standard direct craniocaudal and mediolateral oblique projections. The images were obtained using an analog mammograph. Digital copies of images having a resolution of 600 dpi were obtained for subsequent computer processing. The latter of digital mammographic images involved segmentation of space-occupying lesions, determination of the linearity factor of their outlines, and differential diagnosis of space-occupying lesions based on the calculated value of the linearity factor of their outlines.

Results. An algorithm was elaborated for identifying the outlines of space-occupying lesions on X-ray mammographic images. The sequence of complex-valued vectors approximating its curve was used as a mathematical model of the outline. The concept on the outline linearity factor, which quantitatively characterizes its shape, was introduced. A method was developed for the objective classification of malignant and benign space-occupying lesions based on the value of the introduced linearity factor. The outlines of benign space-occupying lesions in the breast were ascertained to be characterized by the higher linearity factor (in the region of 0.3–0.4) (BI-RADS category 2), while the outlines of malignant tumors had a much lower value of this factor (in the order of 0.05–0.1) (BI-RADS categories 4–5). The main quantitative measures (sensitivity, specificity, and accuracy) of the informative value of the proposed method were determined. The latter was shown to have a higher specificity than the traditional visual analysis carried out by a radiologist. This allows the proposed method to be used as an additional procedure in the visual analysis of mammograms to enhance the reliability of clinical findings.

Conclusion. The practical value of the method is in quantitatively evaluating the shapes of malignant breast neoplasms, in reducing the performance of a mammographic examination, and in increasing its objectivity. The proposed method makes it possible to reduce the time of analyzing X-ray mammograms and to enhance the reliability of clinical findings.

1. Shumakova T.A., Solntseva I.A., Safronova O.B., Savello V.E., Serebryakova S.V. Application of the international BI-RADS classification in mammology: a guide for physicians. Saint Petersburg: ELBI-SPb, 2018 (in Russ.).

2. Sinitsyn V.E. ACR BI-RADS: a system for describing and processing breast imaging data. Mammology atlas: mammography, ultrasound, and magnetic resonance imaging. Moscow: Medpraktika-M; 2010 (in Russ.).

3. Rozhkova N.I., Burdina I.I., Dabagov A.R., Mazo M.L., Prokopenko S.P., Jacobs O.E. X-ray diagnosis in mammology: guidelines for physicians. Moscow: SIMK; 2014 (in Russ.).

4. Moreira I.C., Amaral I., Domingues I., Cardoso A., Cardoso M.J., Cardoso J.S. INbreast: toward a full-field digital mammographic database. Acad. Radiol. 2012; 19 (2): 236–48. DOI: 10.1016/j.acra.2011.09.014

5. Furman Ya.A., Krevetskiy A.V., Peredreev A.K. Introduction to contour analysis and its applications to the processing of images and signals. 2nd ed. Moscow: Fizmatlit; 2003 (in Russ.).

6. Wentzel E.S., Ovcharov L.A. The theory of probability and its engineering applications. Moscow: Vysshaya shkola; 2006 (in Russ.).

7. Baker J.A., Rosen E.L., Lo J.Y., Gimenez E.I., Walsh R., Soo M.S. Computer-aided detection (CAD) in screening mammography: sensitivity of commercial CAD systems for detecting architectural distortion. Am. J. Roentgenol. 2003; 181 (4): 1083–8. DOI: 10.2214/ajr.181.4.1811083

8. Dheeba J., Albert Singh N., Tamil Selvi S. Computer-aided detection of breast cancer on mammograms: a swarm intelligence optimized wavelet neural network approach. J. Biomed. Inform. 2014; 49: 45–52. DOI: 10.1016/j.jbi. 2014.01.010

9. Lehman C.D., Wellman R.D., Buist D.S., Kerlikowske K., Tosteson A.N., Miglioretti D.L. Diagnostic accuracy of digital screening mammography with and without computeraided detection. JAMA Intern. Med. 2015; 175 (11): 1828–37. DOI: 10.1001/jamainternmed.2015.5231

10. Rozhkova N.I., Gorshkov V.A. (Eds.). Digital mammological clinic. Visualization technologies. M.: SIMK; 2013 (in Russ.).

11. Shakh B.A., Fundaro D.M., Mandava S. Radiation diagnosis of breast diseases. Moscow: BINOM; 2013 (in Russ.).

12. Romanycheva E.A., Batukhtin D.M., Ivanov K.O., Sevastyanov V.V., Furman Ya.A., Eruslanov R.V. Classification of objects according to their shape in mammographic images using loop analysis. Journal of Volga State University of Technology. Series: Radiotechnical and Infocommunication Systems. 2014; 4: 58–71 (in Russ.).

13. Mikhailov M.K., Sevastyanov V.V., Furman Ya.A., Romanycheva E.A. Differential diagnosis of images of neoplasms with a strong shape variability on X-ray mammograms by methods of contour analysis. Proceedings of the 9th AllRussian National Congress of Radiation Diagnosis and Therapy “Radiology-2015”. Russian Electronic Journal of Radiology. 2015; 5 (2): 112 (in Russ.).

14. Batukhtin D.M., Romanycheva E.A., Sevastyanov V.V., Ivanov K.O., Furman Ya.A. Combined method of classifying objects in mammographic images. Biomedical Engineering. 2016; 50 (2): 100–4.