MICROCALCIFICATION CLUSTERING PARAMETERS IN BREAST DISEASE - A MORPHOMETRIC ANALYSIS OF RADIOGRAPHS OF EXCISION SPECIMENS

Citation
Kh. Ng et al., MICROCALCIFICATION CLUSTERING PARAMETERS IN BREAST DISEASE - A MORPHOMETRIC ANALYSIS OF RADIOGRAPHS OF EXCISION SPECIMENS, British journal of radiology, 69(820), 1996, pp. 326-334
Citations number
39
Categorie Soggetti
Radiology,Nuclear Medicine & Medical Imaging
Journal title
British journal of radiology
ISSN journal
00071285 → ACNP
Volume
69
Issue
820
Year of publication
1996
Pages
326 - 334
Database
ISI
SICI code
Abstract
X-ray microradiography of surgically excised breast specimens offers t he possibility of morphological characterization of calcifications. Wh en combined with digital imaging techniques there exists added potenti al for obtaining valuable basic quantitative morphometric information regarding differences between microcalcifications in tissues exhibitin g evidence of fibrocystic change, benign and malignant tumours. A tota l of 157 excised breast specimens from 84 patients were microradiograp hed using a Softer Super Soft X-ray unit and Kodak AA high resolution industrial film. A Quantimet 570C image analysis system was used to di gitize and analyse the microradiographs. Of the 157 microradiographs, 51 (from 30 patients) revealed microcalcification clusters. The existe nce of significant differences between the three identified categories of tissue were indicated by clustering parameters. These included the number of particles per cluster, area of clusters, maximum distance t o nearest neighbour, and geometric mean distance to nearest neighbour. The distribution pattern index (DPI), another of the clustering param eters used in this study, has been observed to be a particularly power ful discriminator. The value for fibrocystic change was found to be si gnificantly smaller (0.514) than that for benign tumour (0.796) whilst that for benign tumour was observed to be significantly larger than t hat for malignant tumour (0.604) at a p-value of less than 0.05 (Krusk al-Wallis one-way analysis of variance).