Giulio Tagliafico

Mammographic density estimation: Comparison among BI-RADS categories, a semi-automated software and a fully automated one

Although breast density is considered a strong predictor of breast cancer risk, its quantitative assessment is difficult. The aim of this study is to demonstrate that breast density assessment with a fully automated software is feasible and correlates with the semi-automated evaluation and the quantitative BI-RADS standards. A data set of 160 mammograms was evaluated by three blinded radiologists. Intra-observer (reader 1: k=0.71; reader 2: k=0.76; reader 3: k=0.62) and inter-observer (reader 1 vs reader 2: k=0.72; reader 2 vs reader 3: k=0.80; reader 3 vs reader 1: k=0.72) variability for the semi-automated software were good on a four-grade scale (D1/D2/D3/D4) and correlated with BI-RADS evaluation made by other two blinded radiologists (r=0.65, p<0.01). Inter-observer (reader 1 vs reader 2: k=0.85; reader 2 vs reader 3: k=0.91; reader 3 vs reader 1: k=0.85) variability for the semi-automated software was very good on a two-grade scale (D1-D2/D3-D4). The use of the fully automated software eliminated intra- and inter-observer differences, correlated with BI-RADS categories (r=0.62, p<0.01) and can replace the semi-automated one (Bland-Altman statistics). Our study demonstrates that automated estimation of breast density is feasible and eliminates subjectivity. Furthermore both the semi-automated and the fully automated density estimation are more accurate than BI-RADS quantitative evaluation and could also be used in the daily clinical practice.

Nerve density: a new parameter to evaluate peripheral nerve pathology on ultrasound. Preliminary study.

The possibility to realize a quantitative evaluation of nerve density on ultrasound is clinically important to enhance the evaluation of peripheral nerve disorders. We developed software that quantifies the ratio between the hypoechoic and hyperechoic areas of peripheral nerves on ultrasound. Nerve density was defined as (hypoechoic pixels)/(total pixels) and the purpose of our study was to asses if nerve density can be used to differentiate pathologic conditions affecting peripheral nerves. Ultrasound images of peripheral nerves were obtained with a high-frequency probe (17-5 MHz, 288 elements). Sixty-five different patients and (n = 65) controls (age range, 35-81 years; mean 55 years) were prospectively evaluated. Thirty-five patients had carpal tunnel syndrome and 30 patients had neurofibromas. Three radiologists performed a semiautomated evaluation with intra and interobserver agreement. A complete automatic evaluation was performed with no need of intra and interobserver evaluation. With the semiautomated evaluation, mean intraobserver agreement was good (K = 0.85). Interobserver agreements was good as well (reader 1 vs reader 2: k = 0.72; reader 2 vs reader 3: k = 0.80; reader 3 vs reader 1: k = 0.72). Differences among value of nerve density in normal nerves, CTS and neurofibromas were statistically significant (p < 0.0001). There were no statistically significant differences between the results obtained using the automatic or the semiautomatic method. Nerve density is capable of discriminating between normal and pathologic nerves of patients affected by carpal tunnel syndrome or neurofibromas. Moreover, nerve density measure is useful to discriminate between patients with mild and severe CTS.

Breast Density Assessment Using a 3T MRI System: Comparison among Different Sequences

Purpose To compare MRI sequences for breast density measurements on a 3T MRI system using IDEAL (Iterative Decomposition of water and fat with Echo Asymmetry and Least squares estimation) as possible physiology-like reference. Materials and Methods MRI examination was performed in 48 consecutive patients (mean age 41, years; range, 35–67 years) on a 3.0T scanner and 46 were included. All (fertile) women, were examined between days 5 and 15 of their menstrual cycle. MRI protocol included: T1-turbo spin-echo (T1-tSE), T2-turbo spin-echo (T2-tSE), VIBRANT (Volume Imaging for Breast Assessment) before and after injection of contrast media and IDEAL. Breast density was calculated with semi-automated software. Statistical analysis was performed with non-parametric tests. Results Mean percentage of breast density calculated in each sequence was: T1-tSE  = 56%; T2-tSE  = 52%; IDEAL FatOnly  = 55%; IDEAL WaterOnly  = 53%, VIBRANT  = 55%. Significant differences were observed between T2-tSE and both T1-tSE (p<0.001), VIBRANT sequences (p = 0.009), T1-tSE and both IDEAL WaterOnly (p = 0.007) and IDEAL FatOnly (p = 0.047). Breast density percentage showed a positive linear correlation among different sequences: r≥0.93. Conclusions Differences exist between MRI sequences used to assess breast density percentage. T1-weighted sequences values were similar to IDEAL sequences.

Fascicular ratio: a new parameter to evaluate peripheral nerve pathology on magnetic resonance imaging: a feasibility study on a 3T MRI system.

AbstractThe objective of the study was to define and quantitatively evaluate the fascicular ratio (FR) on magnetic resonance imaging (MRI) in patients with peripheral neuropathies compared with healthy controls.Forty control subjects (20 women, 20 men; age, 44.6 ± 13.4 years) and 40 patients with peripheral neuropathy (22 women, 18 men; age, 50.3 ± 10.2 years) were examined with a standard 3T MRI protocol. With customized software (with semiautomatic and automatic interface), the hypointense and hyperintense areas of the peripheral nerves corresponding to fascicular and nonfascicular tissue were examined on T1-weighted sequences. The ratio of fascicular pixels to total pixels was called FR. Correlation with FR calculated on high-resolution ultrasound was performed. The statistical analysis included the Mann–Whitney U test of controls versus patients, the receiver operating characteristic (ROC) analysis, and the subgroup analysis of patients according to etiologies of neuropathy. Intraobserver and interobserver agreement was calculated based on the evaluation made by 3 readers. Finally, a complete automatic evaluation was performed.On MRI, FRs were significantly increased in patients compared with controls (FR, 76.7 ± 15.1 vs 56 ± 12.3; P < 0.0001 for the semiautomatic interface; and FR 66.3 ± 17.5 vs 47.8 ± 18.4; P < 0.0001 for the automatic interface). The increase in FR was caused mainly by an increase in the hypointense part of the nerve. This observation was valid for all causes of neuropathies. ROC analysis found an area under the curve of 0.75 (95% confidence interval, 0.44–0.81) for FR to discriminate neuropathy from control. The correlation coefficient between MRI and ultrasound was significant (r = 0.49; 95% confidence interval for r, 0.21–0.70; P = 0.012).With the semiautomated evaluation, the mean intraobserver agreement was good (K = 0.86). The interobserver agreements were also good (reader 1 vs reader 2, k = 0.71; reader 2 vs reader 3, k = 0.78; reader 3 vs reader 1, k = 0.71). There were no statistically significant differences between the results obtained using the 2 methods.FR calculation on MRI is feasible, and it may be used in adjunct to standard MRI evaluation in peripheral nerve disorders.

Increased mammographic breast density in acromegaly: quantitative and qualitative assessment.

CONTEXT Mammographic density is a strong independent risk factor for breast cancer, whose prevalence in acromegaly is still controversial. OBJECTIVE To compare breast density in premenopausal acromegalic patients and controls and to determine whether density correlated with disease duration, GH, and IGF1 levels. DESIGN, SETTING AND PARTICIPANTS A prospective study involving 30 patients and 60 controls matched for age and body mass index. INTERVENTIONS A quantitative computer-aided mammographic density estimation (MDEST) and a qualitative blind evaluation by two experienced radiologists using the breast imaging reporting and data system (BI-RADS) was performed. Totally, 60 (acromegaly) and 120 (controls) craniocaudal and mediolateral oblique mammograms were evaluated in both patients and controls. MAIN OUTCOME MEASURES Breast density. RESULTS Patients showed a significantly (P<0.01) increased mammographic breast density with both methods (MDEST: 0.33 ± 0.21% and BI-RADS category: 2.81 ± 0.78) in comparison with controls (MDEST: 0.26 ± 0.19% and BI-RADS category: 2.35 ± 0.61). The agreement between the two methods and inter-observer agreement between the two radiologists were excellent (k=0.63 and k=0.85). In patients grouped according to disease activity (17 controlled and 13 uncontrolled) and medical therapy (15 treated and 15 untreated), no differences were found. All these groups had significantly increased mammographic breast density compared with controls (P<0.01). A positive correlation was found between mammographic breast density, IGF1 values and disease duration (r=0.29 and r=0.39), whereas it was not found with GH (r=-0.02). CONCLUSIONS Mammographic breast density in premenopausal acromegalic patients is significantly higher than controls and positively correlated with IGF1 and disease duration.

Differences in breast density assessment using mammography, tomosynthesis and MRI and their implications for practice

Literature searching of PubMed highlights, through the number of articles related to breast density and also to digital breast tomosynthesis (DBT), that there is an increasing interest in breast density assessment and DBT. Indeed, searching at first the term “breast density” and then “digital breast tomosynthesis”, it is possible to note a positive trend on these topics (http://www.ncbi.nlm.nih.gov). The question then arises as to why there is an interest in breast density assessment and how this relates to new modalities for breast cancer detection? The answer may be that modern medicine is rapidly moving towards a personalised or tailored approach based on predefined risks of a particular disease, a concept that is particularly applicable to breast cancer. Identifying females with an increased risk of developing breast cancer is possible and is important because they may benefit from modified screening and diagnostic protocols [1]. The inclusion of breast density assessment in statistical models such as the Gail et al [2] and Claus et al [3] models may improve their accuracy or use because these methods include non-modifiable risk factors, such as age at inclusion, age at menarche, age at first full-term pregnancy, number of previous biopsies with a benign result and number of first-degree relatives with breast cancer. By contrast, breast density is considered to be an independent risk factor [4], and it is also indicative of changes in modifiable risk factors [5–9]. Moreover, breast density may be particularly suitable for individualised breast cancer risk estimation and not for population-level brief estimation only [10]. Breast density assessment may be carried out using different imaging modalities used in clinical practice, such as mammography, DBT and MRI. These commonly used imaging techniques may give similar results for density assessment, but they may not always be interchangeable for an individualised purpose [11]. Therefore, the purpose of this article is to give a brief overview on breast density percentage assessment with these imaging techniques, highlighting our perspectives on the differences and limitations of each technique.

Quantitative assessment of nerve echogenicity: a promising research tool?

We read with great interest the article by Boom and Visser: ‘‘Quantitative assessment of nerve echogenicity: Comparison of methods for evaluating nerve echogenicity in ulnar neuropathy at the elbow’’ (Boom and Visser, 2012). We are happy to see that the most reliable automatic thresholding method found in this study is very similar to the maximum entropy one used in our study (Tagliafico et al., 2010). However, we would like to add some possible ideas on the reasons why the authors were not able to differentiate normal from pathological ulnar nerves using the manual method. A first issue is represented by patient selection. In the study no mention is reported in case of subluxation of the ulnar nerve over the medial epicondyle. This condition may be responsible of a nonsymptomatic enlargement of the ulnar nerve cross-sectional area (Tagliafico, 2011). These patients may also have a more hyperechogenic nerve in case of long lasting frictions over the bone (Tagliafico et al., 2008). In a relatively small sample, inclusion of patients or controls with ulnar nerve subluxation may be sufficient to reduce the chances to obtain significant results. This hypothesis is reinforced by the fact the authors found a considerable amount of nerves without a clear fascicular pattern on ultrasound. This sign, for example, may be related to subluxation. Another issue may be represented by the different kind of training in analysing medical images. In our study the evaluation was made by three radiologists with formal training in medical imaging evaluation. This fact may have increased reliability and repeatability results of our study using the manual thresholding. Moreover, in a group of 15 patients with confirmed ulnar neuropathy at the elbow, with the same exclusion criteria, we were able to differentiate patients and controls using a manual thresholding method (mean hypoechoic fraction between patients and controls: 88% versus 69% (p < 0.02) if patients and controls with ulnar nerve subluxation were excluded (Fig. 1). We would like to encourage the discussion related to nerve echogenicity assessment because this seems to be a way to study the internal structure of a peripheral nerve using ultrasound. Moreover this topic seems a promising research area.

Local up-sampling and morphological analysis of low-resolution magnetic resonance images

Abstract Limitations in the resolution of acquired images, which are due to sensor manufacturing and acquisition conditions, are reduced with the help of algorithms that enhance the spatial resolution by assigning pixel values that are interpolated or approximated from known pixels. We propose a variant of the moving least-squares approximation for image up-sampling, with a specific focus on biomedical MR images. For each evaluation point, we locally compute the best approximation by minimizing a weighted least-squares error between the input data and their approximation with an implicit function. The proposed approach provides a continuous approximation, an accuracy and extrapolation capabilities higher than previous work, and a lower computational cost. As main application, we consider the up-sampling of low field MR images, where the volumetric and meshless properties of the approximation allow us to easily process images with anisotropic voxel size by rescaling the image and inter-slices resolution. Finally, we include the resolution rescaling into a pipeline that performs a morphological characterization of 3D anatomical districts, which has been developed with a focus on rheumatoid arthritis evolution and provides a more accurate segmentation as an input to quantitative analysis.