Inter- and intraobserver variability in the evaluation of dynamic breast cancer MRI

Evaluation of Breast Cancer Size Measurement by Computer-Aided Diagnosis (CAD) and a Radiologist on Breast MRI

Purpose: This study aimed to evaluate cancer size measurement by computer-aided diagnosis (CAD) and radiologist on breast magnetic resonance imaging (MRI) relative to histopathology and to determine clinicopathologic and MRI factors that may affect measurements. Methods: Preoperative MRI of 208 breast cancers taken between January 2017 and March 2021 were included. We evaluated correlation between CAD-generated size and pathologic size as well as that between radiologist-measured size and pathologic size. We classified size discrepancies into accurate and inaccurate groups. For both CAD and radiologist, clinicopathologic and imaging factors were compared between accurate and inaccurate groups. Results: The mean sizes as predicted by CAD, radiologist and pathology were 2.66 ± 1.68 cm, 2.54 ± 1.68 cm, and 2.30 ± 1.61 cm, with significant difference (p < 0.001). Correlation coefficients of cancer size measurement by radiologist and CAD in reference to pathology were 0.898 and 0.823. Radiologist’s measurement was more accurate than CAD, with statistical significance (p < 0.001). CAD-generated measurement was significantly more inaccurate for cancers of larger pathologic size (>2 cm), in the presence of an extensive intraductal component (EIC), with positive progesterone receptor (PR), and of non-mass enhancement (p = 0.045, 0.045, 0.03 and 0.002). Radiologist-measured size was significantly more inaccurate for cancers in presence of an in situ component, EIC, positive human epidermal growth factor receptor 2 (HER2), and non-mass enhancement (p = 0.017, 0.008, 0.003 and <0.001). Conclusion: Breast cancer size measurement showed a very strong correlation between CAD and pathology and radiologist and pathology. Radiologist-measured size was significantly more accurate than CAD size. Cancer size measurement by CAD and radiologist can both be inaccurate for cancers with EIC or non-mass enhancement.

Features from MRI texture analysis associated with survival outcomes in triple-negative breast cancer patients

PURPOSE

The purpose of the study is to evaluate the associations between intratumoral or peritumoral textural features derived from pretreatment magnetic resonance imaging (MRI) and recurrence-free survival (RFS) in triple-negative breast cancer (TNBC) patients.

METHODS

Forty-three patients with TNBC who underwent preoperative MRI between February 2008 and March 2014 were included. We performed two-dimensional texture analysis on the intratumoral or peritumoral region of interest (ROI) on axial of T2-weighted image (T2WI), dynamic contrast-enhanced (DCE)-MRI and DCE-MRI subtraction images. We also analyzed histopathological data. Cox proportional hazards models were used to investigate associations with survival outcomes.

RESULTS

Twelve of the 43 patients (27.9%) had recurrence disease, at a median of 32.5 months follow-up (1.4-61.5 months). In univariate analysis, nine texture features in T2WI and DCE-MRI subtraction images were significantly associated with RFS. In multivariate analysis, intratumoral difference entropy in DCE-MRI subtraction images in the initial phase (hazard ratio 11.71; 95% confidence interval (CI) [1.41, 97.00]; p value 0.023) and, peritumoral difference variance in DCE-MRI subtraction images in the delayed phase (hazard ratio 9.60; 95% CI [1.98, 46.51]; p value 0.005), were both independently associated with RFS. Moreover, multivariate analysis revealed the presence of lymphovascular invasion as independently associated with RFS (hazard ratio 8.13; 95% CI [2.16, 30.30]; p value 0.002).

CONCLUSIONS

At pretreatment MRI, an intratumoral and peritumoral quantitative approach using texture analysis has the potential to serve as a prognostic marker in patients with TNBC.

Early prediction of neoadjuvant chemotherapy response by exploiting a transfer learning approach on breast DCE-MRIs

The dynamic contrast-enhanced MR imaging plays a crucial role in evaluating the effectiveness of neoadjuvant chemotherapy (NAC) even since its early stage through the prediction of the final pathological complete response (pCR). In this study, we proposed a transfer learning approach to predict if a patient achieved pCR (pCR) or did not (non-pCR) by exploiting, separately or in combination, pre-treatment and early-treatment exams from I-SPY1 TRIAL public database. First, low-level features, i.e., related to local structure of the image, were automatically extracted by a pre-trained convolutional neural network (CNN) overcoming manual feature extraction. Next, an optimal set of most stable features was detected and then used to design an SVM classifier. A first subset of patients, called fine-tuning dataset (30 pCR; 78 non-pCR), was used to perform the optimal choice of features. A second subset not involved in the feature selection process was employed as an independent test (7 pCR; 19 non-pCR) to validate the model. By combining the optimal features extracted from both pre-treatment and early-treatment exams with some clinical features, i.e., ER, PgR, HER2 and molecular subtype, an accuracy of 91.4% and 92.3%, and an AUC value of 0.93 and 0.90, were returned on the fine-tuning dataset and the independent test, respectively. Overall, the low-level CNN features have an important role in the early evaluation of the NAC efficacy by predicting pCR. The proposed model represents a first effort towards the development of a clinical support tool for an early prediction of pCR to NAC.

MRI-based radiomics in breast cancer: feature robustness with respect to inter-observer segmentation variability

Radiomics is an emerging field using the extraction of quantitative features from medical images for tissue characterization. While MRI-based radiomics is still at an early stage, it showed some promising results in studies focusing on breast cancer patients in improving diagnoses and therapy response assessment. Nevertheless, the use of radiomics raises a number of issues regarding feature quantification and robustness. Therefore, our study aim was to determine the robustness of radiomics features extracted by two commonly used radiomics software with respect to variability in manual breast tumor segmentation on MRI. A total of 129 histologically confirmed breast tumors were segmented manually in three dimensions on the first post-contrast T1-weighted MR exam by four observers: a dedicated breast radiologist, a resident, a Ph.D. candidate, and a medical student. Robust features were assessed using the intraclass correlation coefficient (ICC > 0.9). The inter-observer variability was evaluated by the volumetric Dice Similarity Coefficient (DSC). The mean DSC for all tumors was 0.81 (range 0.19–0.96), indicating a good spatial overlap of the segmentations based on observers of varying expertise. In total, 41.6% (552/1328) and 32.8% (273/833) of all RadiomiX and Pyradiomics features, respectively, were identified as robust and were independent of inter-observer manual segmentation variability.

Preoperative tumor size measurement in breast cancer patients: which threshold is appropriate on computer-aided detection for breast MRI?

Background

Computer-aided detection (CAD) can detect breast lesions by using an enhancement threshold. Threshold means the percentage of increased signal intensity in post-contrast imaging compared to precontrast imaging. If the pixel value of the enhanced tumor increases above the set threshold, CAD provides the size of the tumor, which is calculated differently depending on the set threshold. Therefore, CAD requires the accurate setting of thresholds. We aimed to compare the diagnostic accuracy of tumor size measurement using MRI and CAD with 3 most commonly used thresholds and to identify which threshold is appropriate on CAD in breast cancer patients.

Methods

A total of 130 patients with breast cancers (80 invasive cancers and 50 ductal carcinoma in situ [DCIS]) who underwent preoperative MRI with CAD and surgical treatment were included. Tumor size was manually measured on first contrast-enhanced MRI and acquired by CAD using 3 different thresholds (30, 50, and 100%) for each tumor. Tumor size measurements using MRI and CAD were compared with pathological sizes using Spearman correlation analysis. For comparison of size discrepancy between imaging and pathology, concordance was defined as estimation of size by imaging within 5 mm of the pathological size. Concordance rates were compared using Chi-square test.

Results

For both invasive cancers and DCIS, correlation coefficient rho (r) between tumor size on imaging and pathology was highest at CAD with 30% threshold, followed by MRI, CAD with 50% threshold, and CAD with 100% threshold (all p <  0.05). For invasive cancers, the concordance rate of 72.5% at CAD with 30% threshold showed no difference with that of 62.5% at MRI (p = 0.213). For DCIS, the concordance rate of 30.0% at CAD with 30% threshold showed no difference with that of 36.0% at MRI (p = 0.699). Compared to MRI, higher risk of underestimation was noted when using CAD with 50% or 100% threshold for invasive cancers and when using CAD with 100% threshold for DCIS.

Conclusion

For CAD analysis, 30% threshold is the most appropriate threshold whose accuracy is comparable to manual measurement on MRI for tumor size measurement. However, clinicians should be aware of the higher risk of underestimation when using CAD with 50% threshold for tumor staging in invasive cancers.

Digital Analysis in Breast Imaging

Breast imaging is a multimodal approach that plays an essential role in the diagnosis of breast cancer. Mammography, sonography, magnetic resonance, and image-guided biopsy are imaging techniques used to search for malignant changes in the breast or precursors of malignant changes in, e.g., screening programs or follow-ups after breast cancer treatment. However, these methods still have some disadvantages such as interobserver variability and the mammography sensitivity in women with radiologically dense breasts. In order to overcome these difficulties and decrease the number of false positive findings, improvements in imaging analysis with the help of artificial intelligence are constantly being developed and tested. In addition, the extraction and correlation of imaging features with special tumor characteristics and genetics of the patients in order to get more information about treatment response, prognosis, and also cancer risk are coming more and more in focus. The aim of this review is to address recent developments in digital analysis of images and demonstrate their potential value in multimodal breast imaging.

Classification of breast mass lesions on dynamic contrast-enhanced magnetic resonance imaging by a computer-assisted diagnosis system based on quantitative analysis

The aim of the current study was to develop a semi-automatic and quantitative method for the analysis of a time-intensity curve (TIC) from breast dynamic contrast-enhanced magnetic resonance imaging. The performance of the proposed method, based on the level set segmentation algorithm, was evaluated by comparison with the traditional method. In the traditional method, the lesion area is delineated manually and the corresponding mean TIC is classified subjectively as one of three washout patterns. In addition, only one quantitative parameter, the maximum slope of increase (MSI), is calculated. In the proposed method, the lesion region was determined semi-automatically and the corresponding mean TIC was categorized quantitatively. In addition to MSI, a number of quantitative parameters were derived from the mean TIC and lesion area, including signal intensity slope (SI_slope), initial percentage of enhancement (E_initial), percentage of peak enhancement (E_peak), early signal enhancement ratio (ESER) and second enhancement percentage (SEP). Wilcoxon signed-rank test and receiver operating characteristic analyses were performed for statistical analysis. For TIC categorization the accuracy was 61.54% for the traditional method and 82.05% for the proposed method. Using the proposed method, mean curve accuracies were 84.0% for SI_slope, 66.7% for MSI, 66.0% for E_initial, 66.0% for E_peak, 68.0% for ESER and 44.9% for SEP. In the lesion region, the accuracies for the aforementioned parameters were 80.8, 65.4, 66.7, 62.2, 69.2 and 57.1%, respectively. Accuracy of the MSI value derived from the traditional method was 63.4%. Compared with the traditional method, the proposed semi-automatic method in the current study may provide results with a higher accuracy to differentiate benign and malignant lesions. Therefore, the proposed method should be considered as a supplementary tool for the diagnosis of breast lesions.

Breast cancer MRI radiomics: An overview of algorithmic features and impact of inter-reader variability in annotating tumors

PURPOSE

To review features used in MRI radiomics of breast cancer and study the inter-reader stability of the features.

METHODS

We implemented 529 algorithmic features that can be extracted from tumor and fibroglandular tissue (FGT) in breast MRIs. The features were identified based on a review of the existing literature with consideration of their usage, prognostic ability, and uniqueness. The set was then extended so that it comprehensively describes breast cancer imaging characteristics. The features were classified into 10 groups based on the type of data used to extract them and the type of calculation being performed. For the assessment of inter-reader variability, four fellowship-trained readers annotated tumors on preoperative dynamic contrast-enhanced MRIs for 50 breast cancer patients. Based on the annotations, an algorithm automatically segmented the image and extracted all features resulting in one set of features for each reader. For a given feature, the inter-reader stability was defined as the intraclass correlation coefficient (ICC) computed using the feature values obtained through all readers for all cases.

RESULTS

The average inter-reader stability for all features was 0.8474 (95% CI: 0.8068-0.8858). The mean inter-reader stability was lower for tumor-based features (0.6348, 95% CI: 0.5391-0.7257) than FGT-based features (0.9984, 95% CI: 0.9970-0.9992). The feature group with the highest inter-reader stability quantifies breast and FGT volume. The feature group with the lowest inter-reader stability quantifies variations in tumor enhancement.

CONCLUSIONS

Breast MRI radiomics features widely vary in terms of their stability in the presence of inter-reader variability. Appropriate measures need to be taken for reducing this variability in tumor-based radiomics.

Biomarkers and Imaging of Breast Cancer

OBJECTIVE

The goals of this review are to provide background information on the definitions and applications of the general term "biomarker" and to highlight the specific roles of breast imaging biomarkers in research and clinical breast cancer care. A search was conducted of the main electronic biomedical databases (PubMed, Cochrane, Embase, MEDLINE [Ovid], Scopus, and Web of Science). The search was focused on review literature in general radiology and biomedical sciences and on reviews and primary research articles on biomarkers in breast imaging over the 15 years ending in June 2017. The keywords included "biomarker," "trial endpoints," "breast imaging," "breast cancer," "radiomics," and "precision medicine" in the titles and abstracts of the papers.

CONCLUSION

Clinical breast care and breast cancer-related research rely on imaging biomarkers for decision support. In the era of precision medicine and big data, the practice of radiology is likely to change. A closer integration of breast imaging with related biomedical fields and the creation of large integrated and shareable databases of clinical, molecular, and imaging biomarkers should allow the field to continue guiding breast cancer care and research.

Magnetic Resonance Imaging Features of Idiopathic Granulomatous Mastitis: A Retrospective Analysis

Background

Granulomatous mastitis is a rare and benign inflammatory breast disease that may clinically and radiologically mimic breast cancer.

Objectives

The aim of this study was to evaluate the features of idiopathic granulomatous mastitis (IGM) on breast magnetic resonance imaging (MRI) with mammographic and sonographic findings.

Patients and Methods

A retrospective analysis was conducted on 20 patients with IGM who had been diagnosed by needle core or excisional biopsy at a single institution between 2006 and 2012. All of the patients underwent MRI for suspicious breast findings prior to biopsy. MRI examinations were performed on a 1.5 T scanner. The MRI findings were evaluated in accordance with the breast imaging-reporting and data system (BI-RADS) MRI lexicon established by the American College of Radiology.

Results

MRI detected a total of 29 lesions in the 20 patients. Fourteen of these lesions were seen as mass enhancements, with the remaining 15 identified as non-mass enhancements (NMEs). The median size of all lesions was 3.6 cm (range, 0.7 - 6.7 cm). The most frequently observed features were masses with a round shape (9 out of 14, 64%), smooth contour (11 out of 14, 78%), and a rim enhancement pattern (10 out of 14, 71%). The most common features of the 15 NME lesions were segmental distribution (6 out of 15, 40%) and heterogeneous enhancement patterns (8 out of 15, 53%). The time-intensity curves of the dynamic studies showed benign type one signal intensity (persistent enhancement pattern) in the majority of lesions (10 out of 20, 50%).

Conclusion

Our study suggests that breast MRI findings of IGM have a wide spectrum. Rim enhancement patterns are frequently seen on contrast enhanced images, but the imaging findings are nonspecific and cannot be used definitively to distinguish between benign and malignant lesions.

A comparison of quantitative methods for clinical imaging with hyperpolarized (13)C-pyruvate

Dissolution dynamic nuclear polarization (DNP) enables the metabolism of hyperpolarized (13)C-labelled molecules, such as the conversion of [1-(13)C]pyruvate to [1-(13)C]lactate, to be dynamically and non-invasively imaged in tissue. Imaging of this exchange reaction in animal models has been shown to detect early treatment response and correlate with tumour grade. The first human DNP study has recently been completed, and, for widespread clinical translation, simple and reliable methods are necessary to accurately probe the reaction in patients. However, there is currently no consensus on the most appropriate method to quantify this exchange reaction. In this study, an in vitro system was used to compare several kinetic models, as well as simple model-free methods. Experiments were performed using a clinical hyperpolarizer, a human 3 T MR system, and spectroscopic imaging sequences. The quantitative methods were compared in vivo by using subcutaneous breast tumours in rats to examine the effect of pyruvate inflow. The two-way kinetic model was the most accurate method for characterizing the exchange reaction in vitro, and the incorporation of a Heaviside step inflow profile was best able to describe the in vivo data. The lactate time-to-peak and the lactate-to-pyruvate area under the curve ratio were simple model-free approaches that accurately represented the full reaction, with the time-to-peak method performing indistinguishably from the best kinetic model. Finally, extracting data from a single pixel was a robust and reliable surrogate of the whole region of interest. This work has identified appropriate quantitative methods for future work in the analysis of human hyperpolarized (13)C data.

Inter-reader reproducibility of dynamic contrast-enhanced magnetic resonance imaging in patients with non-small cell lung cancer treated with bevacizumab and erlotinib

Objectives When evaluating anti-tumor treatment response by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) it is necessary to assure its validity and reproducibility. This has not been well addressed in lung tumors. Therefore we have evaluated the inter-reader reproducibility of response classification by DCE-MRI in patients with non-small cell lung cancer (NSCLC) treated with bevacizumab and erlotinib enrolled in a multicenter trial.

MATERIALS AND METHODS

Twenty-one patients were scanned before and 3 weeks after start of treatment with DCE-MRI in a multicenter trial. The scans were evaluated by two independent readers. The primary lung tumor was used for response assessment. Responses were assessed in terms of relative changes in tumor mean trans endothelial transfer rate (K(trans)) and its heterogeneity in terms of the spatial standard deviation. Reproducibility was expressed by the inter-reader variability, intra-class correlation coefficient (ICC) and dichotomous response classification.

RESULTS

The inter-reader variability and ICC for the relative K(trans) were 5.8% and 0.930, respectively. For tumor heterogeneity the inter-reader variability and ICC were 0.017 and 0.656, respectively. For the two readers the response classification for relative K(trans) was concordant in 20 of 21 patients (k=0.90, p<0.0001) and for tumor heterogeneity in 19 of 21 patients (k=0.80, p<0.0001).

CONCLUSIONS

Strong agreement was seen with regard to the inter-reader variability and reproducibility of response classification by the two readers of lung cancer DCE-MRI scans.

Reproducibility of Dynamic Contrast-Enhanced MRI in Renal Cell Carcinoma: A Prospective Analysis on Intra- and Interobserver and Scan-Rescan Performance of Pharmacokinetic Parameters

The objective of this study was to investigate the intra- and interobserver as well as scan-rescan reproducibility of quantitative parameters of renal cell carcinomas (RCCs) with dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). A total of 21 patients with clear cell RCCs (17 men, 4 woman; age 37-69 years, mean age 54.6 years, mean size, 5.0 ± 2.2 cm) were prospectively recruited from September 2012 to November 2012. Patients underwent paired DCE-MRI studies on a 3.0 T MR system with an interval of 48 to 72 hours. The extended-Tofts model and population-based arterial input function were used to calculate kinetic parameters. Three observers defined the 2-dimensional whole-tumor region of interest at the slice with the maximum diameter of the RCC. Intraobserver and scan-rescan differences were assessed using paired t tests, whereas interobserver differences using two-way analysis of variance. Intra- and interobserver reproducibility and scan-rescan reproducibility were evaluated using within-subject coefficient of variation (wCoV) and intraclass correlation coefficient (ICC). There were no significant intra-, interobserver, or scan-rescan differences in parameters (all P > 0.05). All ICCs for intra- and interobserver agreements were >0.75 (P < 0.05), whereas the scan-rescan agreement was moderate to good; V(e) (0.764, 95% confidence interval [CI]: 0.378-0.925) and K(ep) (0.906, 95% CI: 0.710-0.972) had higher ICC than K(trans) (0.686; 95% CI: 0.212-0.898) and V(p) (0.657; 95% CI: 0.164-0.888). In intra- and interobserver variability analyses, all parameters except V(p) had low wCoV values. K(trans) and V(e) had slightly lower intraobserver wCoV (1.2% and 0.9%) compared with K(ep) (3.7%), whereas all 3 of these parameters had similar interobserver wCoV values (2.5%, 3.1%, and 2.9%, respectively). Regarding scan-rescan variability, K(trans) and K(ep) showed slightly higher variation (15.6% and 15.4%) than V(e) (10.1%). V(p) had the largest wCoV in all variability analyses (all >30%). DCE-MRI demonstrated good intra- and interobserver reproducibility and moderate to good scan-rescan performance in the assessment of RCC using K(trans), K(ep), and V(e) as parameters under noncontinuous scanning mode. V(p) showed poor reproducibility, and thus may not be suitable for this scanning protocol.

Computer-aided detection (CAD) system for breast MRI in assessment of local tumor extent, nodal status, and multifocality of invasive breast cancers: preliminary study

Background

We aimed to investigate the efficacy of computer-aided detection (CAD) for MRI in the assessment of tumor extent, lymph node status, and multifocality in invasive breast cancers in comparison with other breast imaging modalities.

Methods

Two radiologists measured the maximum tumor size, as well as, analyzed lymph node status and multifocality in 86 patients with invasive breast cancers using mammography, ultrasound, CT, MRI with and without CAD, and 18-fludeoxyglucose positron emission tomography (FDG-PET). The assessed data were compared with pathology.

Results

For tumor extent, there were no significant differences between pathological size and measured size using mammography, ultrasound, CT, or MRI with and without CAD (P > 0.05). For evaluation of lymph node status, ultrasound had the best kappa coefficients (0.522) for agreement between imaging and pathology, and diagnostic performance with 92.1% specificity and 90.0% positive predictive value. For multifocality, MRI with CAD had the highest area under the receiver operating characteristic curve (AUC = 0.888).

Conclusions

CAD for MRI is feasible to assess tumor extent and multifocality in invasive breast cancer patients. However, CAD is not effective in evaluation of nodal status.

Exploring temporospatial changes in glucose metabolic disorder, learning, and memory dysfunction in a rat model of diffuse axonal injury

Diffuse axonal injury (DAI) is the predominant effect of severe traumatic brain injury and contributes significantly to cognitive deficits. The mechanisms underlying these cognitive deficits are often associated with complex metabolic alterations. However, the relationships between temporospatial alterations in cerebral glucose metabolism and the pathophysiology of DAI-related learning and memory dysfunction are not yet completely understood. We used a small animal positron emission tomography (PET) scanner with 2-[F-18]-fluoro-2-deoxy-D-glucose (¹⁸F-FDG) as a molecular probe to evaluate temporospatial glucose metabolism in vulnerable areas of rats with DAI. The Morris water maze (MWM) was used to evaluate the development and progression of learning and memory dysfunction. Compared to the sham-treated group, PET-MRI fusion images showed that glucose metabolism was reduced in animals with DAI. In addition, the standardized uptake value (SUV) of ¹⁸F-FDG was significantly decreased in the sensorimotor cortex, hippocampus, corpus callosum, caudate putamen, brain stem, and cerebellum at days 1, 3, and 7 after injury. SUV returned to baseline levels by 30 days after injury. The escape latency of the injured group was significantly increased, and the percentages of distance travelled and time spent in the target quadrant were significantly decreased 1 month after injury. These effects persisted for 3 months. SUVs in the hippocampus at the acute stage were significantly correlated with MWM performance during the recovery stage of DAI. These results demonstrate that microstructural injury-induced hypometabolism in the hippocampus at the acute stage are all significantly correlated with learning and memory dysfunctions during the recovery stage of DAI.

Reproducibility of dynamic contrast-enhanced MR imaging. Part II. Comparison of intra- and interobserver variability with manual region of interest placement versus semiautomatic lesion segmentation and histogram analysis

PURPOSE

To compare the inter- and intraobserver variability with manual region of interest (ROI) placement versus that with software-assisted semiautomatic lesion segmentation and histogram analysis with respect to quantitative dynamic contrast material-enhanced (DCE) MR imaging determinations of the volume transfer constant (K(trans)).

MATERIALS AND METHODS

The study was approved by the institutional review board and compliant with HIPAA. The requirement to obtain informed consent was waived. Fifteen DCE MR imaging studies of the female pelvis defined the study group. Uterine fibroids were used as a perfusion model. Three varying types of lesion measurements were performed by five readers on each study by using DCE MR imaging perfusion analysis software with manual ROI placement and a semiautomatic lesion segmentation and histogram analysis solution. Intra- and interreader variability of measurements of K(trans) with the different measurement types was calculated.

RESULTS

The overall interobserver variability of K(trans) with manual ROI placement (mean, 28.5% ± 9.3) was reduced by 42.5% when the semiautomatic, software-assisted lesion measurement method was used (16.4% ± 6.2). Whole-lesion measurement showed the lowest interobserver variability with both measurement methods (20.1% ± 4.3 with the manual method vs 10.8% ± 2.6 with the semiautomatic method). The overall intrareader variability with the manual ROI method (7.6% ± 10.6) was not significantly different from that with the semiautomatic method (7.3% ± 10.8), but the intraclass correlation coefficient for intrareader reproducibility improved from 0.86 overall with the manual method to 0.99 with the semiautomatic method.

CONCLUSION

A semiautomatic lesion segmentation and histogram analysis approach can provide a significant reduction in interobserver variability for DCE MR imaging measurements of K(trans) when compared with manual ROI methods, whereas intraobserver reproducibility is improved to some extent.

Multicentre imaging measurements for oncology and in the brain

Multicentre imaging studies of brain tumours (and other tumour and brain studies) can enable a large group of patients to be studied, yet they present challenging technical problems. Differences between centres can be characterised, understood and minimised by use of phantoms (test objects) and normal control subjects. Normal white matter forms an excellent standard for some MRI parameters (e.g. diffusion or magnetisation transfer) because the normal biological range is low (<2-3%) and the measurements will reflect this, provided the acquisition sequence is controlled. MR phantoms have benefits and they are necessary for some parameters (e.g. tumour volume). Techniques for temperature monitoring and control are given. In a multicentre study or treatment trial, between-centre variation should be minimised. In a cross-sectional study, all groups should be represented at each centre and the effect of centre added as a covariate in the statistical analysis. In a serial study of disease progression or treatment effect, individual patients should receive all of their scans at the same centre; the power is then limited by the within-subject reproducibility. Sources of variation that are generic to any imaging method and analysis parameters include MR sequence mismatch, B(1) errors, CT effective tube potential, region of interest generation and segmentation procedure. Specific tissue parameters are analysed in detail to identify the major sources of variation and the most appropriate phantoms or normal studies. These include dynamic contrast-enhanced and dynamic susceptibility contrast gadolinium imaging, T(1), diffusion, magnetisation transfer, spectroscopy, tumour volume, arterial spin labelling and CT perfusion.

Pearls and pitfalls in breast MRI

At our academic institution, we have noticed repeated examples of both false-positive and false-negative MR diagnoses in breast cancer. The most common diagnostic errors in interpreting MRI of the breast are discussed in this review and experience-based advice is provided to avoid similar mistakes. The most common reasons for false-positive diagnoses are misinterpretation of artefacts, confusion between normal enhancing structures and tumours and, above all, insufficient use of the American College of Radiology breast imaging reporting and data system lexicon, whereas false-negative diagnoses are made as a result of missed tiny enhancement, a background-enhancing breast, or enhancement interpreted as benign rather than malignant.

Differential diagnosis of breast cancer using quantitative, label-free and molecular vibrational imaging

We present a label-free, chemically-selective, quantitative imaging strategy to identify breast cancer and differentiate its subtypes using coherent anti-Stokes Raman scattering (CARS) microscopy. Human normal breast tissue, benign proliferative, as well as in situ and invasive carcinomas, were imaged ex vivo. Simply by visualizing cellular and tissue features appearing on CARS images, cancerous lesions can be readily separated from normal tissue and benign proliferative lesion. To further distinguish cancer subtypes, quantitative disease-related features, describing the geometry and distribution of cancer cell nuclei, were extracted and applied to a computerized classification system. The results show that in situ carcinoma was successfully distinguished from invasive carcinoma, while invasive ductal carcinoma (IDC) and invasive lobular carcinoma were also distinguished from each other. Furthermore, 80% of intermediate-grade IDC and 85% of high-grade IDC were correctly distinguished from each other. The proposed quantitative CARS imaging method has the potential to enable rapid diagnosis of breast cancer.

Iterative active deformational methodology for tumor delineation: Evaluation across radiation treatment stage and volume

PURPOSE

To introduce, implement, and assess an iterative modification to the active deformational image segmentation method as applied to cervical cancer tumors.

MATERIALS AND METHODS

A comparison by Jaccard similarity (JS) between this active deformational method and manual segmentation was performed on tumors of various sizes across preradiation, 3 weeks postradiation, and 6 weeks postradiation using a General Linear Mixed Model across 121 studies from 52 patients with Stage IIB-IV cervical cancers.

RESULTS

The deformable segmentation method produced promising levels of agreement including JS factors of 0.71+/-0.11 in the preradiation studies. The analysis illustrated a rate of improvement in JS with increasing tumor volume that differed between the preradiation and 6 weeks postradiation stage (P=0.0474). In the large preradiated tumors each additional cm3 of volume was associated with an increase or improvement in JS of 0.0008 (95% confidence interval [CI]: 0.0003, 0.0014). In the smaller postradiation tumors, each additional cm3 of volume was associated with a more robust improvement in JS of 0.0046 (95% CI: 0.0009, 0.0082).

CONCLUSION

Agreement was strongly affected by tumor volume, and its performance was most impacted across volume in the later stages of radiation therapy. The deformation-based segmentation method appears to demonstrate utility for delineating cervical cancer tumors, particularly in the earliest stages of radiation treatment, where agreement is greatest.

Multispectral co-occurrence with three random variables in dynamic contrast enhanced magnetic resonance imaging of breast cancer

Presented is a new computer-aided multispectral image processing method which is used in three spatial dimensions and one spectral dimension where the dynamic, contrast enhanced magnetic resonance parameter maps derived from voxel-wise model-fitting represent the spectral dimension. The method is based on co-occurrence analysis using a 3-D window of observation which introduces an automated identification of suspicious lesions. The co-occurrence analysis defines 21 different statistical features, a subset of which were input to a neural network classifier where the assessments of the voxel-wise majority of a group of radiologist readings were used as the gold standard. The voxel-wise true positive fraction (TPF) and false positive fraction (FPF) results of the computer classifier were statistically indistinguishable from the TPF and FPF results of the readers using a one sample paired t-test. In order to observe the generality of the method, two different groups of studies were used with widely different image acquisition specifications.

The "laboratory" effect: comparing radiologists' performance and variability during prospective clinical and laboratory mammography interpretations

PURPOSE

To compare radiologists' performance during interpretation of screening mammograms in the clinic with their performance when reading the same mammograms in a retrospective laboratory study.

MATERIALS AND METHODS

This study was conducted under an institutional review board-approved, HIPAA-compliant protocol; the need for informed consent was waived. Nine experienced radiologists rated an enriched set of mammograms that they had personally read in the clinic (the "reader-specific" set) mixed with an enriched "common" set of mammograms that none of the participants had previously read in the clinic by using a screening Breast Imaging Reporting and Data System (BI-RADS) rating scale. The original clinical recommendations to recall the women for a diagnostic work-up, for both reader-specific and common sets, were compared with their recommendations during the retrospective experiment. The results are presented in terms of reader-specific and group-averaged sensitivity and specificity levels and the dispersion (spread) of reader-specific performance estimates.

RESULTS

On average, the radiologists' performance was significantly better in the clinic than in the laboratory (P = .035). Interreader dispersion of the computed performance levels was significantly lower during the clinical interpretations (P < .01).

CONCLUSION

Retrospective laboratory experiments may not represent either expected performance levels or interreader variability during clinical interpretations of the same set of mammograms in the clinical environment well.

Dynamic MRI for imaging tumor microvasculature: comparison of susceptibility and relaxivity techniques in pelvic tumors

PURPOSE

To assess the reproducibility of intrinsic relaxivity and both relaxivity- and susceptibility-based dynamic contrast enhanced (DCE) MRI in pelvic tumors; to correlate kinetic parameters obtained and to assess whether acute antivascular effects are seen in response to cisplatin- or taxane-based chemotherapy.

MATERIALS AND METHODS

T1-weighted and T2*-weighted DCE-MRI and basal R2* measurements were performed on three consecutive days in women with gynecological tumors. The third scan was 21.0 (range 17.3-23.5) hours after the first cycle of chemotherapy. Kinetic parameter estimates were obtained and correlated between techniques. Test-retest reproducibility and response to treatment were assessed.

RESULTS

Relative blood volume (rBV) and relative blood flow (rBF) correlated strongly with transfer constant (Ktrans), kep, and the initial area under the gadopentetate dimeglumine (Gd-DTPA) concentration-time curve (IAUGC) (all P<0.01). The group 95% confidence interval (CI) for change was -10.8 to +12.1%; +/-5.1%; -9.5 to +10.5%; +/-7.5%; for Ktrans, ve, kep, and IAUGC, respectively, and +/-13.6%, +/-2.4%, +/-11.6%, and +/-11.0%, for rBV, mean transit time (MTT), rBF, and R2*, respectively. There were no significant acute changes in kinetic parameter estimates in response to treatment on group analysis, apart from a small decrease in ve.

CONCLUSION

The results confirm the dominant influence of flow on Ktrans in untreated gynecological tumors. There is no evidence of an acute, large magnitude antivascular effect caused by cisplatin- or taxane-based chemotherapy.