A preliminary study on computerized lesion localization in MR mammography using 3D nMITR maps, multilayer cellular neural networks, and fuzzy ‐partitioning

Abstract

Cellular neural networks (CNNs) are massively parallel cellular structures with learning abilities. They can be used to realize complex image processing applications efficiently and in almost real time. In this preliminary study, we propose a novel, robust, and fully automated system based on CNNs to facilitate lesion localization in contrast-enhanced MR mammography, a difficult task requiring the processing of a large number of images with attention paid to minute details. The data set consists of 1170 slices containing one precontrast and five postcontrast bilateral axial MR mammograms from 39 patients with 37 malignant and 39 benign mass lesions acquired using a 1.5 Tesla MR scanner with the following parameters: 3D FLASH sequence, TR/TE 9.80/4.76 ms, flip angle 250, slice thickness 2.5 mm, and 0.625 x 0.625 mm2 in-plane resolution. Six hundred slices with 21 benign and 25 malignant lesions of this set are used for training the CNNs; the remaining data are used for test purposes. The breast region of interest is first segmented from precontrast images using four 2D CNNs connected in cascade, specially designed to minimize false detections due to muscles, heart, lungs, and thoracic cavity. To identify deceptively enhancing regions, a 3D nMITR map of the segmented breast is computed and converted into binary form. During this process tissues that have low degrees of enhancements are discarded. To boost lesions, this binary image is processed by a 3D CNN with a control template consisting of three layers of 11 x 11 cells and a fuzzy c-partitioning output function. A set of decision rules extracted empirically from the training data set based on volume and 3D eccentricity features is used to make final decisions and localize lesions. The segmentation algorithm performs well with high average precision, high true positive volume fraction, and low false positive volume fraction with an overall performance of 0.93 +/- 0.05, 0.96 +/- 0.04, and 0.03 +/- 0.05, respectively (training: 0.93 +/- 0.04, 0.94 +/- 0.04, and 0.02 +/- 0.03; test: 0.93 +/- 0.05, 0.97 +/- 0.03, and 0.05 +/- 0.06). The lesion detection performance of the system is quite satisfactory; for the training data set the maximum detection sensitivity is 100% with false-positive detections of 0.28/lesion, 0.09/slice, and 0.65/case; for the test data set the maximum detection sensitivity is 97% with false-positive detections of 0.43/lesion, 0.11/slice, and 0.68/case. On the average, for a detection sensitivity of 99%, the overall performance of the system is 0.34/lesion, 0.10/slice, and 0.67/case. The system introduced does not require prior information concerning breast anatomy; it is robust and exceptionally effective for detecting breast lesions. The use of CNNs, fuzzy c-partitioning, volume, and 3D eccentricity criteria reduces false-positive detections due to artifacts caused by highly enhanced blood vessels, nipples, and normal parenchyma and artifacts from vascularized tissues in the chest wall due to oversegmentation. We hope that this system will fa...

A computerized volumetric segmentation method applicable to multi-centre MRI data to support computer-aided breast tissue analysis, density assessment and lesion localization

“…In order to assess the degree to which the new method represents an improvement over an algorithm previously demonstrated in the literature, we performed a comparison with the cellular neural network (CNN) method [9] for all 82 cases. The adjustable parameter b in the thresholding stage of that algorithm was here set to 0.55, which gave significantly improved results for the current cohort, compared with the value of 0.79 in [9].…”

“…The adjustable parameter b in the thresholding stage of that algorithm was here set to 0.55, which gave significantly improved results for the current cohort, compared with the value of 0.79 in [9]. The best CNN segmentation performance gave overall statistics RO = 0.88, TPVF = 0.94 and FPVF = 0.07 on these data, which is somewhat inferior to the results of the new BCFCM method (RO = 0.94, TVPF = 0.97, FPVF = 0.04).…”

“…These adjusted segmentations are clearly not “blind” to computer segmentations, and hence a comparison of segmentation performance with methods that use “pure” manual segmentation is not entirely fair. However, our prior experience, as reported in [9], is that, for the breast-to-air boundary, computerized segmentations and pure manual segmentations are almost the same. This method therefore allows us to process a large statistical sample, where the human reviewer performing the corrections can focus on the chest wall region, which is where automated algorithms face the greatest difficulties.…”

“…The breast–air boundary is identified easily by searching for a sharp increase in the image intensity from the air side provided that the background noise is low [9, 13, 22]. However, detection of the breast–chest wall boundary is a complicated problem, due to coil-related intensity inhomogeneity artefacts and partial volume issues, especially in the presence of dense breast tissue connected to the chest wall muscles and liver tissue beneath chest wall muscles [47].…”

“…Ertas et al [9] proposed a four-element cascaded cellular neural network that performs grey-level thresholding, detection of the largest region and morphological erosion followed by reconstruction using morphological dilation.…”