Measures of angular spread and entropy for the detection of architectural distortion in prior mammograms

Purpose

Architectural distortion is an important sign of early breast cancer. We present methods for computer-aided detection of architectural distortion in mammograms acquired prior to the diagnosis of breast cancer in the interval between scheduled screening sessions.

Methods

Potential sites of architectural distortion were detected using node maps obtained through the application of a bank of Gabor filters and linear phase portrait modeling. A total of 4,224 regions of interest (ROIs) were automatically obtained from 106 prior mammograms of 56 interval-cancer cases, including 301 true-positive ROIs, and from 52 mammograms of 13 normal cases. Each ROI was represented by three types of entropy measures of angular histograms composed with the Gabor magnitude response, angle, coherence, orientation strength, and the angular spread of power in the Fourier spectrum, including Shannon’s entropy, Tsallis entropy for nonextensive systems, and Rényi entropy for extensive systems.

Results

Using the entropy measures with stepwise logistic regression and the leave-one-patient-out method for feature selection and cross-validation, an artificial neural network resulted in an area under the receiver operating characteristic curve of 0.75. Free-response receiver operating characteristics indicated a sensitivity of 0.80 at 5.2 false positives (FPs) per patient.

Conclusion

The proposed methods can detect architectural distortion in prior mammograms taken 15 months (on the average) before clinical diagnosis of breast cancer, with a high sensitivity and a moderate number of FPs per patient. The results are promising and may be improved with additional features to characterize subtle abnormalities and larger databases including prior mammograms.     

Detection of architectural distortion in prior screening mammograms using Gabor filters, phase portraits, fractal dimension, and texture analysis

Objective Mammography is a widely used screening tool for the early detection of breast cancer. One of the commonly missed signs of breast cancer is architectural distortion. The purpose of this study is to explore the application of fractal analysis and texture measures for the detection of architectural distortion in screening mammograms taken prior to the detection of breast cancer. Materials and methods A method based on Gabor filters and phase portrait analysis was used to detect initial candidates for sites of architectural distortion. A total of 386 regions of interest (ROIs) were automatically obtained from 14 “prior mammograms”, including 21 ROIs related to architectural distortion. From the corresponding set of 14 “detection mammograms”, 398 ROIs were obtained, including 18 related to breast cancer. For each ROI, the fractal dimension and Haralick’s texture features were computed. The fractal dimension of the ROIs was calculated using the circular average power spectrum technique. Results The average fractal dimension of the normal (false-positive) ROIs was significantly higher than that of the ROIs with architectural distortion (p = 0.006). For the “prior mammograms”, the best receiver operating characteristics (ROC) performance achieved, in terms of the area under the ROC curve, was 0.80 with a Bayesian classifier using four features including fractal dimension, entropy, sum entropy, and inverse difference moment. Analysis of the performance of the methods with free-response receiver operating characteristics indicated a sensitivity of 0.79 at 8.4 false positives per image in the detection of sites of architectural distortion in the “prior mammograms”. Conclusion Fractal dimension offers a promising way to detect the presence of architectural distortion in prior mammograms.     

A pilot study of architectural distortion detection in mammograms based on characteristics of line shadows

Objective  We present herein a novel algorithm for architectural distortion detection that utilizes the point convergence index with the likelihood of lines (e.g., spiculations) relating to architectural distortion. Materials and methods  Validation was performed using 25 computed radiography (CR) mammograms, each of which has an architectural distortion with radiating spiculations. The proposed method comprises five steps. First, the lines were extracted on mammograms, such as spiculations of architectural distortion as well as lines in the mammary gland. Second, the likelihood of spiculation for each extracted line was calculated. In the third step, point convergence index weighted by this likelihood was evaluated at each pixel to enhance distortion only. Fourth, local maxima of the index were extracted as candidates for the distortion, then classified based on nine features in the last step. Results  Point convergence index without the proposed likelihood generated 84.48/image false-positives (FPs) on average. Conversely, the proposed index succeeded in decreasing this number to 12.48/image on average when sensitivity was 100%. After the classification step, number of FPs was reduced to 0.80/image with 80.0% sensitivity. Conclusion  Combination of the likelihood of lines with point convergence index is effective in extracting architectural distortion with radiating spiculations.     

Reduction of false positives in the detection of architectural distortion in mammograms by using a geometrically constrained phase portrait model

Objective One of the commonly missed signs of breast cancer is architectural distortion. We have developed techniques for the detection of architectural distortion in mammograms, based on the analysis of oriented texture through the application of Gabor filters and a linear phase portrait model. In this paper, we propose constraining the shape of the general phase portrait model as a means to reduce the false-positive rate in the detection of architectural distortion. Material and methods The methods were tested with one set of 19 cases of architectural distortion and 41 normal mammograms, and with another set of 37 cases of architectural distortion. Results Sensitivity rates of 84% with 4.5 false positives per image and 81% with 10 false positives per image were obtained for the two sets of images. Conclusion The adoption of a constrained phase portrait model with a symmetric matrix and the incorporation of its condition number in the analysis resulted in a reduction in the false-positive rate in the detection of architectural distortion. The proposed techniques, dedicated for the detection and localization of architectural distortion, should lead to efficient detection of early signs of breast cancer.     

Mid-treatment Ultrasound Descriptors as Qualitative Imaging Biomarkers of Pathologic Complete Response in Patients with Triple-Negative Breast Cancer

This study aimed to investigate mid-treatment breast tumor ultrasound characteristics that may predict eventual pathologic complete response (pCR) in triple-negative breast cancer; specifically, we examined associations between pCR and two parameters: tumor response pattern and tumor appearance. Ultrasound was performed at mid-treatment, defined as the completion of four cycles of anthracycline-based chemotherapy and before receiving taxane-based chemotherapy. Consensus imaging review was performed while blinded to pathology results (i.e., pCR/non-pCR) from surgery. Tumor response pattern was described as “complete,” “concentric,” “fragmented,” “stable” or “progression.” Tumor appearance was designated as “mass,” “architectural distortion,” “flat tumor bed” or “clip only.” Univariate and multivariate regression analyses of 144 participants showed significant associations between mid-treatment response pattern and pCR (p = 0.0348 and p = 0.0173, respectively), with complete and concentric response patterns more likely to achieve pCR than other patterns. Univariate and multivariate regression analyses further showed significant associations between mid-treatment tumor appearance and pCR (p < 0.0001 for both), with persistent appearance of mass less likely than other appearances to achieve pCR. To conclude, our study demonstrated strong associations between pCR and both tumor response pattern and tumor appearance, thereby suggesting that these parameters have potential as qualitative imaging biomarkers of pCR in triple-negative breast cancer.     

Gray-scale and geometric registration of full-field digital and film-screen mammograms

During the transition from traditional film-screen (FS) mammography to full-field digital (FFD) mammography, images from both modalities are used in hospitals and in mammography screening centers, as comparison of mammograms from subsequent examinations of a client is an important part of the diagnostic procedure. A parametric method is presented to register a FS mammogram and a FFD mammogram of the same woman with respect to geometry and gray-scales. The main motivation for the study is to lessen irrelevant differences between mammograms due to acquisition. First, a technique like this might increase the radiologist's ability to detect relevant differences like abnormal growth in breast tissue that signal breast cancer. Second, applications may be found in subtraction radiology or in computer-aided detection of abnormalities in temporal mammograms. The proposed method is based on a parametric model of the most important aspects of acquisition, which relates the pixel values of two images. This encompasses (1) breast positioning; (2) breast compression; (3) exposure time; (4) incident radiation intensity; and (5a) film properties and digitization for FS mammograms, or (5b) detector response for FFD mammograms. The method does not require a priori knowledge about specific settings of acquisition; the parameters are estimated from the two mammograms themselves.     

乳腺导管原位癌X线表现与病理及年龄对照研究

目的探讨乳腺导管原位癌的X线表现及其与病理分级、患病年龄的相关性。方法经手术、病理证实的94例乳腺导管原位癌患者,观察其X线表现中钙化形态及软组织异常（分类包括肿块、不对称致密、结构扭曲）与病理分级、患病年龄的相关性。结果（1）乳腺导管原位癌的X线摄影表现为钙化者占80．4％。低、中级别中不定型模糊钙化和粗糙不均质钙化占45．8％,高级别中细小多形性钙化和线样分支状钙化占71．1％（P〈O．05）。线段样分布占53．6％,簇状分布占33．3％,区域分布占13．1％。本研究中有63．8％的导管原位癌患者每平方厘米钙化数目超过20枚。（2）肿块、局灶性不对称致密、结构扭曲（伴或不伴钙化）在乳腺导管原位癌的X线摄影中分别占39．1％、21．7％、3．2％。其中,病理为高级别者占64．4％,中、低级别分别占18．6％、17．0％。（3）50岁以前钙化出现的几率是50以岁后的1.3倍（P〈0．05）,单纯钙化容易出现在40-50岁（12．0％）;50岁以后出现软组织异常的比例是50岁以前人群的1．4倍（P〈0．05）。结论乳腺导管原位癌常见的X线摄影表现为单纯钙化、钙化伴肿块、局灶性不对称致密和结构扭曲,与病理分级和年龄有一定相关性,在一定程度上可为临床诊治提供参考。     

计算机辅助检测系统对乳腺癌检出的价值评估

目的：评估全数字化乳腺X线摄影的计算机辅助检测系统（CAD）对乳腺癌检出的临床应用价值.材料与方法：收集97例经手术病理证实的乳腺癌全数字化乳腺X线摄影图像,所有病例均经CAD软件检测,记录乳腺癌的X线征象、BI-RADS分类,病理类型并评估CAD检出的敏感性.结果：共有41例（42.3％）肿块,18例（18.5％）钙化,30例（31％）肿块合并钙化,7例（7.2％）结构扭曲和1例（1％）结构不对称.CAD检出乳腺癌X线征象总敏感性为88.7％（肿块,85.4％;钙化,94.4％;肿块合并钙化,100％;结构扭曲,57.1％）,97例乳腺癌有11例（6例肿块;3例结构扭曲;1例钙化;1例结构不对称）未被CAD检出.肿块形状、边缘、BI-RADS分类及病理类型与CAD检测没有显著差异性.CAD检测有27.2％特异性及46.7％假阳性率）.结论：CAD检测显示了高敏感性及低特异性.乳腺癌肿块形状、边缘、BI-RADS分类及病理类型不影响CAD检出敏感性,肿块密度、结构扭曲的毛刺粗细影响CAD的检出率.     

乳腺黏液癌的影像学特征分析

目的 评价乳腺黏液癌的超声、X线等影像学特征及其与病理组织类型的相关性.方法 回顾性分析21例经手术病理证实的乳腺黏液癌患者(22个病灶)的超声、X线特征及与病理组织类型之间的关系.结果 病理组织学分类包括14个单纯型(6个富细胞型,8个少细胞型)和8个混合型病灶.超声:所有病例均存在实性肿块,85.71％(12/14)的单纯型肿块境界清晰,回声等或略低于皮下脂肪,92.86％(13/14)的单纯型病灶后方回声增强;75.00％(6/8)的混合型和14.29％(2/14)的单纯型肿块边界较模糊并细小毛刺,内部回声较脂肪回声低.超声和X线片术前怀疑恶性的比例均为63.64％(14/22).恶性X线表现包括肿块(10个)、局限性不对称致密影(2个)、结构扭曲并恶性钙化和单纯不定性钙化(各1个).肿块主要为高密度,单纯型边界清楚或呈浅分叶状,混合型边界不规则和毛刺改变.81.82％(18/22)的病灶被超声或X线之一疑诊恶性,45.45％(10/22)的病灶术前超声和X线均疑为恶性.结论 乳腺黏液癌尤其单纯型影像学特征不典型,超声和X线诊断均可能诊断为良性病变;肿块边缘特征是鉴别良恶性的重要依据,混合型肿块较单纯型更具有浸润性特征;超声和X线联合诊断有利于避免误诊,两者之一怀疑恶性时,即应行穿刺活检以明确诊断.     

Metaplastic carcinoma of the breast: mammographic and sonographic findings

PURPOSE: We investigated the mammographic, sonographic, and pathologic findings in metaplastic carcinoma of the breast. METHODS: The mammographic (n = 16) and sonographic (n = 11) findings in 16 patients with metaplastic carcinoma of the breast were analyzed retrospectively along with pathologic findings. Whenever possible, results of preoperative fine-needle aspiration biopsy and immunohistochemical studies were obtained. RESULTS: All patients presented with a palpable breast mass. The mean size of the lesions at pathologic examination was 4.2 cm. On mammography, 15 patients had a mass (1 patient had 2 masses), and 1 patient had only clustered microcalcifications without an associated mass. The mean longest diameter of the 16 masses on mammography was 4.6 cm. Eleven lesions (69%) were round to ovoid in shape, 13 lesions (81%) showed ill-defined or obscured margins, and 10 lesions (63%) showed associated architectural distortion. On sonography, 6 (55%) of 11 lesions were round to ovoid, 9 lesions (82%) had well-defined margins, and 6 lesions (55%) showed complex echogenicity with solid and cystic components. At pathologic examination, 4 of these 6 lesions showed hemorrhagic or cystic necrosis. Axillary lymph nodes were positive in 6 (40%) of 15 patients in whom axillary node dissection was performed. CONCLUSIONS: Metaplastic carcinoma of the breast manifests as a rapidly growing, mammographically ill-defined round mass with associated architectural distortion on mammograms. Complex echogenicity with solid and cystic components may be seen sonographically and is related to hemorrhagic or cystic necrosis seen pathologically.     

Relationship between mammographic features and hormone receptor content in patients with breast cancer

Recurrences of breast cancer are more responsive to hormone therapy if the tumors are positive for estrogen receptors or progesterone receptors. To assess the relationship between hormone receptor content, mammographic tumor morphology, and breast parenchymal patterns, we reviewed charts and mammograms of 210 patients with primary unilateral breast cancer. Mammograms of tumors in 97 patients were divided morphologically into five groups: (1) spiculated mass, (2) architectural distortion, (3) calcifications only, (4) circumscribed mass, and (5) tumor not visible. Estrogen receptor positivity was 81% (39/48) in group 1, 37% (7/19) in group 2, 17% (2/12) in group 3, 31% (4/13) in group 4, and 60% (3/5) in group 5 (P less than .001). Mean estrogen receptor content was also significantly different among groups (P less than .001). There was no statistically significant association between tumor morphology and progesterone receptors, or between calcifications and receptor status. In all 210 patients, hormone-receptor-positive tumors showed no association with mammographic parenchymal pattern. When direct assay of estrogen receptors is unavailable, mammographic appearance of the tumor may suggest the estrogen receptor status.     

Longitudinal enhancement of the hyperechoic regions in ultrasonography of muscles using a Gabor filter bank approach: a preparation for semi-automatic muscle fiber orientation estimation

In this study, to complement our previously proposed method for estimating muscle fiber orientation, the Gabor filter bank (GF) technique was applied to sonograms of the biceps and forearm muscles to longitudinally enhance the coherently oriented and hyperechoic perimysiums regions. The method involved three steps: orientation field estimation, frequency map computation and Gabor filtering. The method was evaluated using a simulated image distorted with multiplicative speckle noises where the “muscles” were arranged in a bipennate fashion with an “aponeurosis” located in the middle. After enhancement using the GF approach, most of the original hyperechoic bands in the simulated image could be recovered. The proposed method was also tested using a group of biceps and forearm muscle sonograms collected from healthy adult subjects. Compared with the sonograms without enhancement, the enhanced images led to the detection of more linear patterns including muscle fascicles and smaller angle differences compared with the mean of manual results from two operators, therefore, were better prepared for the automatic estimation of muscle fiber orientation. The proposed method has the potential of assisting in the visualization of strongly oriented patterns in skeletal muscle sonograms as well as in the semi-automatic estimation of muscle fiber orientations. (E-mail: yongjin.zhou@inet.polyu.edu.hk)     

Simulation of Two-Dimensional Ultrasonic Imaging of Biological Tissues in the Presence of Phase Aberrations

A simulator of phase aberrations for mathematical modeling of two-dimensional (2-D) ultrasonic medical imaging is developed. Principal characteristics of expected phase aberrations were put into the model to investigate the distorting influence of intervening tissues on the quality of conventional medical B-scan images. Information necessary for numerical simulations, including the form of the phase correlation function, correlation length and distortion magnitude, was obtained from analysis of known experimental data on abdominal and breast imaging in vivo. Examples of simulated acoustical images of some simple phantoms are presented. Improvement of image quality due to one simple phase adaptation algorithm is also presented.     

Clinical and Imaging Characteristics of Papillary Neoplasms of the Breast Associated with Malignancy: A Retrospective Cohort Study

Papillary neoplasms of the breast comprise a broad range of pathologies ranging from papillomas to papillary carcinomas and have been associated with breast cancers. In this study, we evaluated the clinical, mammographic and sonographic features of papillary breast neoplasms from benign papillary breast lesions to malignancy-associated papillary lesions. A total of 194 lesions in 179 patients were analyzed, including 117 benign papillomas, 24 atypical papillomas, 41 benign papillomas with malignancies and 12 papillary carcinomas found between January 2003 and August 2011 in our institution. Statistically significant clinical factors included patient age (p = 0.001), lesion multiplicity (p = 0.009) and peripheral location (p = 0.003). Among these factors, the odds ratio for malignancy was 8.9 for bilateral multiple lesions. Visibility (p = 0.001) and density (p = 0.039) were significant factors for malignancy in mammograms, and echo patterns (p = 0.006), boundary (p = 0.001) and vascularity (p = 0.005) were significant features on ultrasound that differentiated malignancies from benign lesions. Overall, when papillary breast lesions are located bilaterally and peripherally in older patients, they are correlated with breast cancers. Additionally, for papillary breast lesions that appear highly dense on mammograms and/or exhibit positive vascularity on ultrasound, the probability of malignancy is relatively high.     

全数字化乳腺X线摄影在乳腺癌诊断中的作用

目的分析经手术病理证实的乳腺病变在全数字化乳腺摄影（FFDM）中的影像学表现,提高X线在乳腺癌诊断中的准确性。方法搜集2008年1月-2010年10月313例行FFDM检查、手术和病理证实的乳腺病变患者的临床资料及乳腺X线片,由放射科医师对患者的X线片观察并分析,参照乳腺影像学报告和数据系统（BI-RADS）进行分级,以术后病理结果为金标准,评价FFDM检查诊断乳腺癌的灵敏度、特异度、准确率、阳性预测值和阴性预测值,并分析误诊和漏诊的原因。结果病理结果显示,313例乳腺标本中,乳腺癌194例,乳腺良性病变119例。在乳腺恶性病变中,X线主要表现为单纯肿块型83例,钙化型95例,结构扭曲6例,不对称致密影8例,乳腺内弥漫性腺体结构模糊2例。X线诊断假阳性18例,假阴性20例。FFDM检查对乳腺癌诊断的灵敏度、特异度、准确率、阳性预测值、阴性预测值分别约89.2%、84.9%、87.7%、90.6%、82.8%。误诊的主要原因是肿块的边缘形态、微小钙化等类似于恶性;漏诊的主要原因是乳腺腺体致密、不对称致密影及微小结构扭曲。结论数字化乳腺X线摄影在乳腺癌诊断中具有较高的价值。     

Referral Patterns of Patients for Palliative Radiation Therapy in British Columbia: A Comparison Between Rural and Urban Family Physicians

BackgroundPrevious studies have shown that palliative radiation therapy (PRT) is often underused, especially in rural and remote settings despite evidence supporting its effectiveness in managing symptoms from advanced or metastatic cancer.PurposeTo identify factors which influence family physicians (FPs) in British Columbia (BC) to refer patients for PRT at the BC Cancer Agency (BCCA) and to compare referral patterns between FPs in rural and urban areas.Methods and MaterialsA total of 1,001 questionnaires were sent to all FPs practicing in rural areas and randomly to FPs in urban areas (351 and 650, respectively). Rural and urban areas were chosen based on our previous study of utilization rates of PRT in BC. The questionnaire was adapted from a previously validated survey, and was used to obtain information on referral practices of FPs in BC. FPs who did not practice family medicine or where 80% of their practice was spent with either obstetrical or pediatric patients were excluded.ResultsThe overall response rate was 33% (44% rural vs. 28% urban). Rural FPs were more involved in both palliative care and metastatic cancer management of their patients (88% vs. 74%; P = .01 and 58% vs. 39%; P = .01). No difference was observed in the FPs' awareness of the BCCA's Radiation Oncology Program. The most significant factors influencing an FP to refer a patient for PRT were: poor functional status, inconvenience to travel and life expectancy. A higher proportion of rural FPs had 10 years or less of experience in family practice than the urban FPs (P = .03). There was no significant difference in the formal training or additional training between the rural and urban FPs.ConclusionsThis study found that FPs practicing in rural areas were more involved in palliative management of their patients and participated more in the care of patients with advanced or metastatic cancer than those in urban areas. They also more commonly referred patients for palliative radiotherapy than their urban counterparts. The reported factors that influenced rural and urban FPs to refer were patients' functional status and life expectancy, combined with uncertain benefit and potential side effects of radiotherapy. More than twice as many FPs from rural compared to urban areas were influenced by perceived inconvenience to travel for palliative radiotherapy. After controlling for potential confounding factors, FP awareness of the radiotherapy program, high participation in advanced, metastatic, or palliative care of cancer patients, formal training in radiation oncology, and additional training in palliative care were all associated with an increased probability of ever referring for palliative radiotherapy.     

磁共振及X线对隆乳术后乳腺改变及并发症的诊断价值

目的分析磁共振（MR）及x线对隆乳术后乳腺改变及并发症的诊断价值。方法对38例76个乳腺的MR、X线及临床资料进行回顾性分析。结果正常情况下,假体及聚丙烯酰胺水凝胶（PAHG）边缘光整,MR呈水样信号,x线表现密度均匀,PAHG与周围腺体密度相近。异常情况下：①包膜不均匀增厚挛缩,PAHG呈不规则团块影,腺体结构较乱,密度增高。②自体脂肪隆乳术,乳腺内多发颗粒样低密度影,部分边缘钙化。结论x线是隆乳术后随访首选方法,MR准确直观反应术后乳腺结构的变化及并发症表现,有理想的诊断价值。     

Simulation of mammographic breast compression in 3D MR images using ICP-based B-spline deformation for multimodality breast cancer diagnosis

Purpose

   Multimodality mammography using conventional 2D mammography and dynamic contrast-enhanced 3D magnetic resonance imaging (DCE-MRI) is frequently performed for breast cancer detection and diagnosis. Combination of both imaging modalities requires superimposition of corresponding structures in mammograms and MR images. This task is challenging due to large differences in (1) dimensionality and spatial resolution, (2) variations in tissue contrast, as well as (3) differences in breast orientation and deformation during the image acquisition. A new method for multimodality breast image registration was developed and tested.

Methods

   Combined diagnosis of mammograms and MRI datasets was achieved by simulation of mammographic breast compression to overcome large differences in breast deformation. Surface information was extracted from the 3D MR image, and back-projection of the 2D breast contour in the mammogram was done. B-spline-based 3D/3D surface-based registration was then used to approximate mammographic breast compression. This breast deformation simulation was performed on 14 MRI datasets with 19 corresponding mammograms. The results were evaluated by comparison with distances between corresponding structures identified by an expert observer.

Results

   The evaluation revealed an average distance of 6.46 mm between corresponding structures, when an optimized initial alignment between both image datasets is performed. Without the optimization, the accuracy is 9.12 mm.

Conclusion

   A new surface-based method that approximates the mammographic deformation due to breast compression without using a specific complex model needed for finite-element-based methods was developed and tested with favorable results. The simulated compression can serve as foundation for a point-to-line correspondence between 2D mammograms and 3D MR image data.