Catálogo de publicaciones - libros

The Digital Database for Screening Mammography (DDSM) is an invaluable resource for digital mammography research. However, there are two particular shortcomings that can pose a significant barrier to many of those who may want to use the resource: 1) the actual mammographic image data is encoded using a non-standard lossless variant of the JPEG image format; 2) although detailed metadata is provided, it is not in a form that permits it to be searched, manipulated or reasoned over by standard tools. This paper describes web services that will allow both humans and computers to query for, and obtain, mammograms from the DDSM in a standard and well-supported image file format. Further, this paper describes how these and other services can be used within grid-based workflows, allowing digital mammography researchers to make use of distributed computing facilities.

In this work the implementation of a database of digitized mammograms is described. The digitized images were collected since 1999 by a community of physicists in collaboration with radiologists in several Italian hospitals, as a first step in order to develop and implement a Computer Aided Detection (CAD) system. 3369 mammograms were collected from 967 patients; they were classified according to the type and the morphology of the lesions, the type of the breast tissue and the type of pathologies. A dedicated Graphical User Interface was developed for mammography visualization and processing, in order to support the medical diagnosis directly on a high-resolution screen. The database has been the starting point for the development of other medical imaging applications such as a breast CAD, currently being upgraded and optimized for the use in conjunction of the GRID technology in the framework of the INFN-funded MAGIC-5 project.

Mass screening of breast cancer utilizing mammography (MMG) has been widely carried out. However, MMG might not be able to depict small impalpable masses in dense breast tissue clearly. We have developed a computer-aided detection (CAD) scheme in whole breast ultrasound (US) system for mass screening which has been developed by ALOKA CO., LTD., Japan. Our CAD scheme and image processing techniques have the following three benefits. 1 Indication of mass candidates by our CAD scheme. 2 Visualization of breast US images in two views of B-planes (CC View and ML View) and C-plane. 3 Comparison of left and right breast images as in MMG. The performance of the CAD scheme in detecting malignant masses on an initial study has a true positive fraction of 0.91 (10/11) at a 0.69 (633/924) false positive per image. Although mass screening utilizing US was not appropriate because images acquired by conventional hand probe were poor in reproduction, the problem could be solved in our system.

Medical ontologies are being developed with some of these specifically for mammographic computer aided diagnosis (CAD) systems. However, to provide full functionality for such mammographic CAD systems it is essential that the ontology information is fully linked to the image information. This linking can be through problem specific image attributes. However, such an approach tends to be non-generic. Here, we propose a framework that will use generic image structures and the topology that links the image structures. In the process we describe a comparison approach which takes the classes, attributes and semantics into account.

Mammographic risk assessment provides an indication of the likelihood of women developing breast cancer. A number of mammographic image based classification methods have been developed, such as Wolfe, Boyd, BI-RADS and Tabár based assessment. We provide a comparative study of these four approaches. Results on the full MIAS database are presented, which indicate strong correlation (Spearman’s > 0.9) between Wolfe, Boyd and BI-RADS based classification, whilst the correlation with Tabár based classification is less straight forward (Spearman’s < 0.5, but low correlations mainly caused by one of the classes).

In digital mammography it is of utmost importance that the quality of screen devices is checked on a regularly basis. The EUREF guidelines propose to do this daily using the AAPMtg18-QC pattern. In this paper we report our initial results with the use of an alternative, recently developed, dynamic pattern (“MoniQA”) and a scoring scheme. As soon as the observers are familiar with the procedure, the measurements are very stable and we could not observe big variations in the quality of the monitor. In order to control the intrinsic quality of the monitor, the number of quality control checks could thus be reduced. The global working condition (such as the ambient light level) is controlled as well with the proposed procedure and this may be of great interest, especially during the start-up of digital mammography (screening) units: it is very informative to trace the influences of different light sources (such as (occasional) viewing boxes).

To establish a practical CAD (Computer-Aided Diagnosis) system to facilitate the diagnosis of microcalcifications, we propose a filter-based technique to detect microcalcifications. Via examination of an existing optimal filter-based technique, it is found that its performance on highlighting the energy of mammograms is seriously affected by artefacts and the background of breast. As a result, four methods in pre and post-processing are described in this paper to improve the optimal filtering, leading to an adaptive selection of thresholds for input mammograms. These methods have been tested by using 30 mammograms (with 25 microcalcifications) from the MIAS database and 23 mammograms from DDSM database. Comparing with the original optimal filter-based technique, our technique reduces the false detections (FD), eliminates the influence of the background in mammograms and is able to adaptively select the threshold for the detection of microcalcifications.

The ability of human observers to discriminate between textures is related to the contrast between key structural elements and their repeating patterns. Here we have developed an automatic texture classification approach based on this principle. Local contrast information is modelled and a hybrid metric, based on probability density distributions and transportation estimation, are used to classify unseen samples. Quantitative and qualitative evaluation, based on mammographic images and Wolfe classification, is presented and shows segmentation results in line with the various classes.

To compare image quality and dose of a photon-counting multi-slit scanner (PC) and a system based on amorphous silicon (aSi), images of the CDMAM 3.4 were taken in standard mode. For 3cm PMMA, the PC used 29kV/11.7mAs, aSi used 27kV/ 50mAs. For 5cm PMMA, PC used 35kV /14.8mAs, aSi used 31kV/ 50 mAs. Exposure was manually increased for PC and lowered for aSi-system. Average glandular dose and an image quality index (IQI) were calculated over the diameter ranges 0.06 – 2.0mm and 0.1-1. In standard mode with 3cm PMMA, IQI for PC was 35% lower than for aSi at 80% lower dose. Increased dose of PC resulted in 13% lower IQI at 57% lower dose. With 5cm PMMA IQI in standard mode was 18 % lower with PC at a 69% lower dose. Increasing the dose of PC resulted a 7% lower IQI at 54% lower dose. In conclusion the PC-system might reduce dose by up to 54% at equivalent image quality, although maximal quality of aSi could not be reached.

Numerous studies have investigated the relation between mammographic density and breast cancer risk. These studies indicate that women with high breast density have a four to six fold risk increase. An investigation of whether or not this relation is causal is important for, e.g., hormone replacement therapy (HRT), which has been shown to actually increase the density. No gold standard for automatic assessment of mammographic density exists. Manual methods such as Wolfe patterns and BI-RADS are helpful for communication of diagnostic sensitivity, but they are both time consuming and crude. For serial, temporal analysis it is necessary to be able to detect more subtle changes. In previous work, a method for measuring the effect of HRT w.r.t. changes in biological density in the breast is described. The method provides structural information orthogonal to intensity-based methods. Hessian-based features and a clustering of these is employed to divide a mammogram into four structurally different areas. Subsequently, based on the relative size of the areas, a density score is determined. We have previously shown that this method can separate patients receiving HRT from patients receiving placebo. In this work, the focus is on deeper understanding of the methodology using tests on sets of artificial images of regular elongated structures.