University of Patras
Design and implementation of a digital density equalisation technique in mammography
Abstract
dc:descriptionThe inhomogeneity of the degree of blackening as well as contrast at film regions corresponding to breast periphery and dense mammary gland can be dealt with by applicationof equalization techniques. These techniques aim at equalizing image quality at the entirebreast area and can be categorised as either exposure equalization techniques, which aim atreducing the exposure dynamic range at the film, or density equalization techniques, whichequalize image quality by equalizing the value of a certain image quality parameter at theentire breast region. As a preliminary step, a quantitative data description study was conducted. Such a study had not been carried out by other researchers and was essential, in order to calculate certain image features of the target regions and facilitate the design of an efficient visualization improvement algorithm. This description incorporated a set of quantitative measurements at a sufficiently large sample of mammograms. A quantitative evaluation of image quality at breast periphery and dense mammary gland vs. normal mammary gland was carried out, by means of a technique based on the wavelet transform. Initially, a breast periphery density equalization technique (SCuDE technique) for mammography images was developed. This technique, which is driven by a model of the mammography image, takes into account the non-linear characteristic curve of the film digitiser system in order to restore the modulated beam information that is degraded by the recording and digitisation process. Application of the method demonstrated its capability to equalize the density of mammography images, while contrast improvement at breast periphery was found to be statistically highly significant. Visualisation improvement of breast periphery is obtained with no additional dose to the patient, as the need for acquiring a second mammography image, selecting exposure parameters that would result in good visualisation of breast periphery, is eliminated. SCuDE technique does not require any irradiation data specific for each mammography image, this fact making the technique more robust. Furthermore, it is independent of breast size, breast symmetry, imaging geometry and the type of mammography unit used. Subsequently, another digital density equalization technique, enhancing visualisation of both dense mammary gland and breast periphery, was designed and developed, taking into account the physical background of the SCuDE technique. The technique was based on a layer model of the breast region, in addition to exploiting the film-digitiser characteristic curve. The originality of the technique is attributed to its advantages with respect to the already proposed methods. The technique deals with both dense mammary gland and breast periphery problems. It is applicable to mammograms of both mammography views and is independent of breast size and symmetry. It does not require any irradiation data specific for each mammography image, this fact making the technique more robust. Additionally, it does not consider calcium to have been detected, therefore it can be applied for visualization improvement of micro calcification clusters. Furthermore, the performance of the technique was clinically as well as quantitatively evaluated and the resulting images were found to be density equalized. The visualisation improvement of both dense mammary gland and breast periphery was found to be statistically significant. Future work concerning the improvement of the second proposed technique remains to be carried out. First, selective application of the technique to dense breast regions should be investigated. Second, application of a more advanced segmentation method is expected to lead to more accurate skin edge determination and subsequent layer partitioning. Third, the technique must be appropriately extended to discriminate between signal and noise, so that a de-noising scheme can also be incorporated. Finally, a complete evaluation of the technique should be carried out. The complete anatomical features based performance evaluation should be extended to all breast patterns and performance results should be analyzed per pattern and per mammography view. Additionally, the technique should be further evaluated based on the detection of pathological features.
Degree
thesis:*- Grantor dc:publisher
- University of Patras
- Year dc:date
- 2002
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Stefanoyiannis, Antonios
- Στεφανογιάννης, Αντώνιος
Subjects
dc:subject × 14Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/13469