Prediction of breast cancer molecular subtypes on DCE-MRI using convolutional neural network with transfer learning between two centers.

OBJECTIVES: To apply deep learning algorithms using a conventional convolutional neural network (CNN) and a recurrent CNN to differentiate three breast cancer molecular subtypes on MRI. METHODS: A total of 244 patients were analyzed, 99 in training dataset scanned at 1.5 T and 83 in testing-1 and 62 in testing-2 scanned at 3 T. Patients were classified into 3 subtypes based on hormonal receptor (HR) and HER2 receptor: (HR+/HER2-), HER2+, and triple negative (TN). Only images acquired in the DCE sequence were used in the analysis. The smallest bounding box covering tumor ROI was used as the input for deep learning to develop the model in the training dataset, by using a conventional CNN and the convolutional long short-term memory (CLSTM). Then, transfer learning was applied to re-tune the model using testing-1(2) and evaluated in testing-2(1). RESULTS: In the training dataset, the mean accuracy evaluated using tenfold cross-validation was higher by using CLSTM (0.91) than by using CNN (0.79). When the developed model was applied to the independent testing datasets, the accuracy was 0.4-0.5. With transfer learning by re-tuning parameters in testing-1, the mean accuracy reached 0.91 by CNN and 0.83 by CLSTM, and improved accuracy in testing-2 from 0.47 to 0.78 by CNN and from 0.39 to 0.74 by CLSTM. Overall, transfer learning could improve the classification accuracy by greater than 30%. CONCLUSIONS: The recurrent network using CLSTM could track changes in signal intensity during DCE acquisition, and achieved a higher accuracy compared with conventional CNN during training. For datasets acquired using different settings, transfer learning can be applied to re-tune the model and improve accuracy. KEY POINTS: • Deep learning can be applied to differentiate breast cancer molecular subtypes. • The recurrent neural network using CLSTM could track the change of signal intensity in DCE images, and achieved a higher accuracy compared with conventional CNN during training. • For datasets acquired using different scanners with different imaging protocols, transfer learning provided an efficient method to re-tune the classification model and improve accuracy.

Development of U-Net Breast Density Segmentation Method for Fat-Sat MR Images Using Transfer Learning Based on Non-Fat-Sat Model.

To develop a U-net deep learning method for breast tissue segmentation on fat-sat T1-weighted (T1W) MRI using transfer learning (TL) from a model developed for non-fat-sat images. The training dataset (N = 126) was imaged on a 1.5 T MR scanner, and the independent testing dataset (N = 40) was imaged on a 3 T scanner, both using fat-sat T1W pulse sequence. Pre-contrast images acquired in the dynamic-contrast-enhanced (DCE) MRI sequence were used for analysis. All patients had unilateral cancer, and the segmentation was performed using the contralateral normal breast. The ground truth of breast and fibroglandular tissue (FGT) segmentation was generated using a template-based segmentation method with a clustering algorithm. The deep learning segmentation was performed using U-net models trained with and without TL, by using initial values of trainable parameters taken from the previous model for non-fat-sat images. The ground truth of each case was used to evaluate the segmentation performance of the U-net models by calculating the dice similarity coefficient (DSC) and the overall accuracy based on all pixels. Pearson's correlation was used to evaluate the correlation of breast volume and FGT volume between the U-net prediction output and the ground truth. In the training dataset, the evaluation was performed using tenfold cross-validation, and the mean DSC with and without TL was 0.97 vs. 0.95 for breast and 0.86 vs. 0.80 for FGT. When the final model developed with and without TL from the training dataset was applied to the testing dataset, the mean DSC was 0.89 vs. 0.83 for breast and 0.81 vs. 0.81 for FGT, respectively. Application of TL not only improved the DSC, but also decreased the required training case number. Lastly, there was a high correlation (R2 > 0.90) for both the training and testing datasets between the U-net prediction output and ground truth for breast volume and FGT volume. U-net can be applied to perform breast tissue segmentation on fat-sat images, and TL is an efficient strategy to develop a specific model for each different dataset.

Evaluation of the association between quantitative mammographic density and breast cancer occurred in different quadrants.

BACKGROUND: To investigate the relationship between mammographic density measured in four quadrants of a breast with the location of the occurred cancer. METHODS: One hundred and ten women diagnosed with unilateral breast cancer that could be determined in one specific breast quadrant were retrospectively studied. Women with previous cancer/breast surgery were excluded. The craniocaudal (CC) and mediolateral oblique (MLO) mammography of the contralateral normal breast were used to separate a breast into 4 quadrants: Upper-Outer (UO), Upper-Inner (UI), Lower-Outer (LO), and Lower-Inner (LI). The breast area (BA), dense area (DA), and percent density (PD) in each quadrant were measured by using the fuzzy-C-means segmentation. The BA, DA, and PD were compared between patients who had cancer occurring in different quadrants. RESULTS: The upper-outer quadrant had the highest BA (37 ± 15 cm2) and DA (7.1 ± 2.9 cm2), with PD = 20.0 ± 5.8%. The order of BA and DA in the 4 separated quadrants were: UO > UI > LO > LI, and almost all pair-wise comparisons showed significant differences. For tumor location, 67 women (60.9%) had tumor in UO, 16 (14.5%) in UI, 7 (6.4%) in LO, and 20 (18.2%) in LI quadrant, respectively. The estimated odds and the 95% confidence limits of tumor development in the UO, UI, LO and LI quadrants were 1.56 (1.06, 2.29), 0.17 (0.10, 0.29), 0.07 (0.03, 0.15), and 0.22 (0.14, 0.36), respectively. In these 4 groups of women, the order of quadrant BA and DA were all the same (UO > UI > LO > LI), and there was no significant difference in BA, DA or PD among them (all p > 0.05). CONCLUSIONS: Breast cancer was most likely to occur in the UO quadrant, which was also the quadrant with highest BA and DA; but for women with tumors in other quadrants, the density in that quadrant was not the highest. Therefore, there was no direct association between quadrant density and tumor occurrence.

A Rare and Life-Threatening Complication of Infective Endocarditis: Pseudoaneurysm of the Mitral - Aortic Intervalvular Fibrosa.

UNLABELLED: A 59 year-old previously healthy male was admitted to the hospital with fever reportedly several days in duration. His physical examination was unremarkable at first. Pneumonia was initially diagnosed, but acute pulmonary edema with a new grade III to and fro murmur developed 1 week later. Transesophageal echocardiography (TEE) disclosed a pseudoaneurysm of the mitral-aortic intervalvular fibrosa (P-MAIVF). Subsequent consultation with a cardiovascular surgeon resulted in a repaired aorta with otherwise uneventful results. P-MAIVF is a very rare complication of prosthetic aortic valve (AV) infective endocarditis, and even in native AV. Therefore a careful and through physical examination of patients and early TEE examination are essential in this rare complication of infective endocarditis. KEY WORDS: Echocardiography; Infective endocarditis; Mitral-aortic intervalvular fibrosa; Pseudoaneurysm.

The application of cardiac magnetic resonance imaging (CMR) in myocarditis after COVID-19 vaccines : Case series from single medical center in Taiwan.

During the coronavirus disease 2019 (COVID-19) pandemic, reports of vaccine-induced myocarditis, particularly messenger ribonucleic acid (mRNA)-based myocarditis, were widely spread. This case series describes various cases of COVID-19 vaccine-induced myocarditis confirmed by cardiac magnetic resonance imaging (MRI), including those who were administered rare protein-based vaccines. Eleven patients comprising eight males and three females with suspected myocarditis underwent cardiac MRI at Taichung Veterans General Hospital between October 2021 and May 2022. The median age of the patients was 33.5 years old (range: 22-57 years). The onset of myocarditis was mainly observed following mRNA vaccine inoculation. One patient received the MVC-COV1901 vaccine, a unique protein-based COVID-19 vaccine in Taiwan, and met the 2018 Lake Louise Criteria for the diagnosis of myocarditis, confirmed by cardiac MRI. Most patients reported chest discomfort after receiving various vaccine types. Among four patients with reduced left ventricular ejection fraction (LVEF), two showed LVEF restoration during the follow-up period, and the other two were lost to follow-up. Cardiac MRI characterizes myocardial features such as edema, inflammation, and fibrosis, and has been proven to diagnose myocarditis accurately with a sensitivity of 87.5% and a specificity of 96.2% according to the 2018 Lake Louise criteria. This diagnosis was achieved without invasive procedures such as endomyocardial biopsy or coronary angiography.

A Deep Learning Approach to Classify Fabry Cardiomyopathy from Hypertrophic Cardiomyopathy Using Cine Imaging on Cardiac Magnetic Resonance.

A challenge in accurately identifying and classifying left ventricular hypertrophy (LVH) is distinguishing it from hypertrophic cardiomyopathy (HCM) and Fabry disease. The reliance on imaging techniques often requires the expertise of multiple specialists, including cardiologists, radiologists, and geneticists. This variability in the interpretation and classification of LVH leads to inconsistent diagnoses. LVH, HCM, and Fabry cardiomyopathy can be differentiated using T1 mapping on cardiac magnetic resonance imaging (MRI). However, differentiation between HCM and Fabry cardiomyopathy using echocardiography or MRI cine images is challenging for cardiologists. Our proposed system named the MRI short-axis view left ventricular hypertrophy classifier (MSLVHC) is a high-accuracy standardized imaging classification model developed using AI and trained on MRI short-axis (SAX) view cine images to distinguish between HCM and Fabry disease. The model achieved impressive performance, with an F1-score of 0.846, an accuracy of 0.909, and an AUC of 0.914 when tested on the Taipei Veterans General Hospital (TVGH) dataset. Additionally, a single-blinding study and external testing using data from the Taichung Veterans General Hospital (TCVGH) demonstrated the reliability and effectiveness of the model, achieving an F1-score of 0.727, an accuracy of 0.806, and an AUC of 0.918, demonstrating the model's reliability and usefulness. This AI model holds promise as a valuable tool for assisting specialists in diagnosing LVH diseases.

Consistency of breast density measured from the same women in four different MR scanners.

PURPOSE: To compare the breast volume (BV), fibroglandular tissue volume (FV), and percent density (PD) measured from breast MRI of the same women using four different MR scanners. METHODS: The study was performed in 34 healthy Asian volunteers using two 1.5T (GE and Siemens) and two 3T (GE and Philips) MR scanners. The BV, FV, and PD were measured on nonfat-suppressed T1-weighted images using a comprehensive computer algorithm-based segmentation method. The scanner-to-scanner measurement difference, and the coefficient of variation (CV) among the four scanners were calculated. The measurement variation between two density morphological patterns presenting as the central type and the intermingled type was separately analyzed and compared. RESULTS: All four scanners provided satisfactory image quality allowing for successful completion of the segmentation processes. The measured parameters between each pair of MR scanners were highly correlated, with R(2) ≥ 0.95 for BV, R(2) ≥ 0.99 for FV, and R(2) ≥ 0.97 for PD in all comparisons. The mean percent differences between each pair of scanners were 5.9%-7.8% for BV, 5.3%-6.5% for FV, 4.3%-7.3% for PD; with the overall CV of 5.8% for BV, 4.8% for FV, and 4.9% for PD. The variation of FV was smaller in the central type than in the intermingled type (p = 0.04). CONCLUSIONS: The results showed that the variation of FV and PD measured from four different MR scanners is around 5%, suggesting the parameters measured using different scanners can be used for a combined analysis in a multicenter study.

Automatic Detection and Segmentation of Breast Cancer on MRI Using Mask R-CNN Trained on Non-Fat-Sat Images and Tested on Fat-Sat Images.

RATIONALE AND OBJECTIVES: Computer-aided methods have been widely applied to diagnose lesions on breast magnetic resonance imaging (MRI). The first step was to identify abnormal areas. A deep learning Mask Regional Convolutional Neural Network (R-CNN) was implemented to search the entire set of images and detect suspicious lesions. MATERIALS AND METHODS: Two DCE-MRI datasets were used, 241 patients acquired using non-fat-sat sequence for training, and 98 patients acquired using fat-sat sequence for testing. All patients have confirmed unilateral mass cancers. The tumor was segmented using fuzzy c-means clustering algorithm to serve as the ground truth. Mask R-CNN was implemented with ResNet-101 as the backbone. The neural network output the bounding boxes and the segmented tumor for evaluation using the Dice Similarity Coefficient (DSC). The detection performance, and the trade-off between sensitivity and specificity, was analyzed using free response receiver operating characteristic. RESULTS: When the precontrast and subtraction image of both breasts were used as input, the false positive from the heart and normal parenchymal enhancements could be minimized. The training set had 1469 positive slices (containing lesion) and 9135 negative slices. In 10-fold cross-validation, the mean accuracy = 0.86 and DSC = 0.82. The testing dataset had 1568 positive and 7264 negative slices, with accuracy = 0.75 and DSC = 0.79. When the obtained per-slice results were combined, 240 of 241 (99.5%) lesions in the training and 98 of 98 (100%) lesions in the testing datasets were identified. CONCLUSION: Deep learning using Mask R-CNN provided a feasible method to search breast MRI, localize, and segment lesions. This may be integrated with other artificial intelligence algorithms to develop a fully automatic breast MRI diagnostic system.

Menstrual cycle-related fluctuations in breast density measured by using three-dimensional MR imaging.

PURPOSE: To investigate the fluctuation of fibroglandular tissue volume (FV) and percentage of breast density (PD) during the menstrual cycle and compare with postmenopausal women by using three-dimensional magnetic resonance (MR)-based segmentation methods. MATERIALS AND METHODS: This study was approved by the Institutional Review Board and was HIPAA compliant. Written informed consent was obtained. Thirty healthy female subjects, 24 premenopausal and six postmenopausal, were recruited. All subjects underwent MR imaging examination each week for 4 consecutive weeks. The breast volume (BV), FV, and PD were measured by two operators to evaluate interoperator variation. The fluctuation of each parameter measured over the course of the four examinations was evaluated on the basis of the coefficient of variation (CV). RESULTS: The results from two operators showed a high Pearson correlation for BV (R(2) = 0.99), FV (R(2) = 0.98), and PD (R(2) = 0.96). The interoperator variation was 3% for BV and around 5%-6% for FV and PD. In the respective premenopausal and postmenopausal groups, the mean CV was 5.0% and 5.6% for BV, 7.6% and 4.2% for FV, and 7.1% and 6.0% for PD. The difference between premenopausal and postmenopausal groups was not significant (all P values > .05). CONCLUSION: The fluctuation of breast density measured at MR imaging during a menstrual cycle was around 7%. The results may help the design and interpretation of future studies by using the change of breast density as a surrogate marker to evaluate the efficacy of hormone-modifying drugs for cancer treatment or cancer prevention.

A new bias field correction method combining N3 and FCM for improved segmentation of breast density on MRI.

PURPOSE: Quantitative breast density is known as a strong risk factor associated with the development of breast cancer. Measurement of breast density based on three-dimensional breast MRI may provide very useful information. One important step for quantitative analysis of breast density on MRI is the correction of field inhomogeneity to allow an accurate segmentation of the fibroglandular tissue (dense tissue). A new bias field correction method by combining the nonparametric nonuniformity normalization (N3) algorithm and fuzzy-C-means (FCM)-based inhomogeneity correction algorithm is developed in this work. METHODS: The analysis is performed on non-fat-sat T1-weighted images acquired using a 1.5 T MRI scanner. A total of 60 breasts from 30 healthy volunteers was analyzed. N3 is known as a robust correction method, but it cannot correct a strong bias field on a large area. FCM-based algorithm can correct the bias field on a large area, but it may change the tissue contrast and affect the segmentation quality. The proposed algorithm applies N3 first, followed by FCM, and then the generated bias field is smoothed using Gaussian kernal and B-spline surface fitting to minimize the problem of mistakenly changed tissue contrast. The segmentation results based on the N3+FCM corrected images were compared to the N3 and FCM alone corrected images and another method, coherent local intensity clustering (CLIC), corrected images. The segmentation quality based on different correction methods were evaluated by a radiologist and ranked. RESULTS: The authors demonstrated that the iterative N3+FCM correction method brightens the signal intensity of fatty tissues and that separates the histogram peaks between the fibroglandular and fatty tissues to allow an accurate segmentation between them. In the first reading session, the radiologist found (N3+FCM > N3 > FCM) ranking in 17 breasts, (N3+FCM > N3 = FCM) ranking in 7 breasts, (N3+FCM = N3 > FCM) in 32 breasts, (N3+FCM = N3 = FCM) in 2 breasts, and (N3 > N3+FCM > FCM) in 2 breasts. The results of the second reading session were similar. The performance in each pairwise Wilcoxon signed-rank test is significant, showing N3+FCM superior to both N3 and FCM, and N3 superior to FCM. The performance of the new N3+FCM algorithm was comparable to that of CLIC, showing equivalent quality in 57/60 breasts. CONCLUSIONS: Choosing an appropriate bias field correction method is a very important preprocessing step to allow an accurate segmentation of fibroglandular tissues based on breast MRI for quantitative measurement of breast density. The proposed algorithm combining N3+FCM and CLIC both yield satisfactory results.

Quantitative Measurement of Breast Tumors Using Intravoxel Incoherent Motion (IVIM) MR Images.

Breast magnetic resonance imaging (MRI) is currently a widely used clinical examination tool. Recently, MR diffusion-related technologies, such as intravoxel incoherent motion diffusion weighted imaging (IVIM-DWI), have been extensively studied by breast cancer researchers and gradually adopted in clinical practice. In this study, we explored automatic tumor detection by IVIM-DWI. We considered the acquired IVIM-DWI data as a hyperspectral image cube and used a well-known hyperspectral subpixel target detection technique: constrained energy minimization (CEM). Two extended CEM methods-kernel CEM (K-CEM) and iterative CEM (I-CEM)-were employed to detect breast tumors. The K-means and fuzzy C-means clustering algorithms were also evaluated. The quantitative measurement results were compared to dynamic contrast-enhanced T1-MR imaging as ground truth. All four methods were successful in detecting tumors for all the patients studied. The clustering methods were found to be faster, but the CEM methods demonstrated better performance according to both the Dice and Jaccard metrics. These unsupervised tumor detection methods have the advantage of potentially eliminating operator variability. The quantitative results can be measured by using ADC, signal attenuation slope, D*, D, and PF parameters to classify tumors of mass, non-mass, cyst, and fibroadenoma types.

Age- and race-dependence of the fibroglandular breast density analyzed on 3D MRI.

PURPOSE: The purpose of this study was to evaluate the age- and race-dependence of the breast fibroglandular tissue density based on three-dimensional breast MRI. METHODS: The normal breasts of 321 consecutive patients including Caucasians, Asians, and Hispanics were studied. The subjects were separated into three age groups: Younger than 45, between 45 and 55, and older than 55. Computer algorithms based on body landmarks were used to segment the breast, and fuzzy c-means algorithm was used to segment the fibroglandular tissue. Linear regression analysis was applied to compare mean differences among different age groups and race/ethnicity groups. The obtained parameters were not normally distributed, and the transformed data, natural log (ln) for the fibroglandular tissue volume, and the square root for the percent density were used for statistical analysis. RESULTS: On the average, the transformed fibroglandular tissue volume and percent density decreased significantly with age. Racial differences in mean transformed percent density were found among women older than 45, but not among women younger than 45. Mean percent density was higher in Asians compared to Caucasians and Hispanics; the difference remained significant after adjustment for age, but not significant after adjusted for both age and breast volume. There was no significant difference in the density between the Caucasians and the Hispanics. CONCLUSIONS: The results analyzed using the MRI-based method show age- and race-dependence, which is consistent with literature using mammography-based methods.

Arteriovenous fistula and graft evaluation in hemodialysis patients using MDCT: a primer.

OBJECTIVE: Patent arteriovenous fistula (AVF) is related to better prognosis and quality of life for patients on long-term dialysis. When AVF dysfunction is suspected, MDCT is a good noninvasive tool for evaluating the entire AVF structure and determining reversible conditions for treatment. The aim of this article is to introduce the scanning and interpretation techniques and to illustrate the conditions related to early and late fistula failures. CONCLUSION: MDCT is a fast, noninvasive, and accurate technique for diagnosing AVF complications. Radiologists familiar with these techniques can help to improve the prognosis and quality of life for hemodialysis patients.

Adjuvant whole breast radiotherapy with simultaneous integrated boost to tumor bed with intensity modulated radiotherapy technique in elderly breast cancer patients.

BACKGROUND: Adjuvant whole breast radiotherapy is the standard of care for breast cancer patients after partial mastectomy. Intensity-modulated radiation therapy (IMRT) has been reported to reduce acute toxicities compared to conventional radiotherapy. IMRT with simultaneous integrated boost (SIB) technique can deliver higher doses to tumor bed and irradiate whole breast with a lower dose level to shorten overall treatment duration. This study presents the long-term results of adjuvant IMRT with SIB in elderly breast cancer patients who received partial mastectomy. METHODS: From January 2007 to January 2018, 93 elder breast cancer patients (≥65-year-old) who received IMRT with SIB technique after partial mastectomy were reviewed retrospectively. The axillary areas were managed with either sentinel lymph node biopsies or axillary lymph node dissection. The dose to whole breast was 50.4 Gy in 28 fractions in all patients and the dose to tumor bed was 61.6 to 66.4 Gy in 28 fractions. The primary end point is locoregional control. Secondary end points include: overall survival, breast cancer-specific survival, distant-metastases-free survival, disease-free survival, and acute and chronic toxicities. RESULTS: The median follow-up was 56.1 months. One patient had ipsilateral breast tumor recurrence, 3 patients had regional lymph node recurrence, and 9 patients had distant metastases. Death occurred in 5 patients, including 3 patients died of breast cancer progression. Five-year overall survival is 96.3% and 5-year locoregional recurrence-free survival is 96.4%. The 5-year breast cancer specific survival and 5-year distant metastases-free survival is 97.5% and 87.2%, respectively. Seven patients developed second primary cancer after RT. Eighty-one point seven percent patients had acute grade 1 dermatitis while 18.3% suffered from grade 2 dermatitis. The incidence of grade 1 pneumonitis and grade 1 stomatitis was 4.3% and 8.6%, respectively. CONCLUSIONS: Adjuvant IMRT with SIB technique is a safe and effective treatment strategy for elderly breast cancer patients after partial mastectomy.

CT-guided core biopsy of lung lesions: a primer.

OBJECTIVE: CT-guided core biopsy is playing an increasing role in the diagnosis of benign disease, cellular differentiation, somatic mutation analysis, and molecular fingerprint analysis. CONCLUSION: In this article, we summarize the basic concepts, protocols, and techniques that we use for CT-guided core biopsy of lung lesions to assist radiologists in obtaining diagnostic specimens while reducing preventable complications.

Spontaneous spinal and intracranial subdural hematoma.

Subdural hematoma (SDH) of the spine following intracranial hemorrhage is extremely rare. We present a 35-year-old woman who suffered from headache and dizziness initially, and then lower back pain, lower limb weakness and paraparesis gradually developed within 1-2 weeks. Magnetic resonance imaging revealed intracranial and spinal SDH. No vascular abnormality was seen by brain and spinal angiography. Platelet count, prothrombin time, activated partial thromboplastin time, and inflammatory markers, including C-reactive protein, were normal. A diagnosis of spontaneous spinal and intracranial SDH was then confirmed surgically. Postoperative recovery was uneventful.

Evaluation of breast stiffness measured by ultrasound and breast density measured by MRI using a prone-supine deformation model.

BACKGROUND: This study evaluated breast tissue stiffness measured by ultrasound elastography and the percent breast density measured by magnetic resonance imaging to understand their relationship. METHODS: Magnetic resonance imaging and whole breast ultrasound were performed in 20 patients with suspicious lesions. Only the contralateral normal breasts were analyzed. Breast tissue stiffness was measured from the echogenic homogeneous fibroglandular tissues in the central breast area underneath the nipple. An automatic, computer algorithm-based, segmentation method was used to segment the whole breast and fibroglandular tissues on three dimensional magnetic resonanceimaging. A finite element model was applied to deform the prone magnetic resonance imaging to match the supine ultrasound images, by using the inversed gravity loaded transformation. After deformation, the tissue level used in ultrasound elastography measurement could be estimated on the deformed supine magnetic resonance imaging to measure the breast density in the corresponding tissue region. RESULTS: The mean breast tissue stiffness was 2.3 ± 0.8 m/s. The stiffness was not correlated with age (r = 0.29). Overall, there was no positive correlation between breast stiffness and breast volume (r = - 0.14), or the whole breast percent density (r = - 0.09). There was also no correlation between breast stiffness and the local percent density measured from the corresponding region (r = - 0.12). CONCLUSIONS: The lack of correlation between breast stiffness measured by ultrasound and the whole breast or local percent density measured by magnetic resonance imaging suggests that breast stiffness is not solely related to the amount of fibroglandular tissue. Further studies are needed to investigate whether they are dependent or independent cancer risk factors.

Automatic Breast and Fibroglandular Tissue Segmentation in Breast MRI Using Deep Learning by a Fully-Convolutional Residual Neural Network U-Net.

RATIONALE AND OBJECTIVES: Breast segmentation using the U-net architecture was implemented and tested in independent validation datasets to quantify fibroglandular tissue volume in breast MRI. MATERIALS AND METHODS: Two datasets were used. The training set was MRI of 286 patients with unilateral breast cancer. The segmentation was done on the contralateral normal breasts. The ground truth for the breast and fibroglandular tissue (FGT) was obtained by using a template-based segmentation method. The U-net deep learning algorithm was implemented to analyze the training set, and the final model was obtained using 10-fold cross-validation. The independent validation set was MRI of 28 normal volunteers acquired using four different MR scanners. Dice Similarity Coefficient (DSC), voxel-based accuracy, and Pearson's correlation were used to evaluate the performance. RESULTS: For the 10-fold cross-validation in the initial training set of 286 patients, the DSC range was 0.83-0.98 (mean 0.95 ± 0.02) for breast and 0.73-0.97 (mean 0.91 ± 0.03) for FGT; and the accuracy range was 0.92-0.99 (mean 0.98 ± 0.01) for breast and 0.87-0.99 (mean 0.97 ± 0.01) for FGT. For the entire 224 testing breasts of the 28 normal volunteers in the validation datasets, the mean DSC was 0.86 ± 0.05 for breast, 0.83 ± 0.06 for FGT; and the mean accuracy was 0.94 ± 0.03 for breast and 0.93 ± 0.04 for FGT. The testing results for MRI acquired using four different scanners were comparable. CONCLUSION: Deep learning based on the U-net algorithm can achieve accurate segmentation results for the breast and FGT on MRI. It may provide a reliable and efficient method to process large number of MR images for quantitative analysis of breast density.

Breast Tumor Detection and Classification Using Intravoxel Incoherent Motion Hyperspectral Imaging Techniques.

Breast cancer is a main cause of disease and death for women globally. Because of the limitations of traditional mammography and ultrasonography, magnetic resonance imaging (MRI) has gradually become an important radiological method for breast cancer assessment over the past decades. MRI is free of the problems related to radiation exposure and provides excellent image resolution and contrast. However, a disadvantage is the injection of contrast agent, which is toxic for some patients (such as patients with chronic renal disease or pregnant and lactating women). Recent findings of gadolinium deposits in the brain are also a concern. To address these issues, this paper develops an intravoxel incoherent motion- (IVIM-) MRI-based histogram analysis approach, which takes advantage of several hyperspectral techniques, such as the band expansion process (BEP), to expand a multispectral image to hyperspectral images and create an automatic target generation process (ATGP). After automatically finding suspected targets, further detection was attained by using kernel constrained energy minimization (KCEM). A decision tree and histogram analysis were applied to classify breast tissue via quantitative analysis for detected lesions, which were used to distinguish between three categories of breast tissue: malignant tumors (i.e., central and peripheral zone), cysts, and normal breast tissues. The experimental results demonstrated that the proposed IVIM-MRI-based histogram analysis approach can effectively differentiate between these three breast tissue types.

Development of a quantitative method for analysis of breast density based on three-dimensional breast MRI.

Breast density has been established as an independent risk factor associated with the development of breast cancer. It is known that an increase of mammographic density is associated with an increased cancer risk. Since a mammogram is a projection image, different body position, level of compression, and the x-ray intensity may lead to a large variability in the density measurement. Breast MRI provides strong soft tissue contrast between fibroglandular and fatty tissues, and three-dimensional coverage of the entire breast, thus making it suitable for density analysis. To develop the MRI-based method, the first task is to achieve consistency in segmentation of the breast region from the body. The method included an initial segmentation based on body landmarks of each individual woman, followed by fuzzy C-mean (FCM) classification to exclude air and lung tissue, B-spline curve fitting to exclude chest wall muscle, and dynamic searching to exclude skin. Then, within the segmented breast, the adaptive FCM was used for simultaneous bias field correction and fibroglandular tissue segmentation. The intraoperator and interoperator reproducibility was evaluated using 11 selected cases covering a broad spectrum of breast densities with different parenchymal patterns. The average standard deviation for breast volume and percent density measurements was in the range of 3%-4% among three trials of one operator or among three different operators. The body position dependence was also investigated by performing scans of two healthy volunteers, each at five different positions, and found the variation in the range of 3%-4%. These initial results suggest that the technique based on three-dimensional MRI can achieve reasonable consistency to be applied in longitudinal follow-up studies to detect small changes. It may also provide a reliable method for evaluating the change of breast density for risk management of women, or for evaluating the benefits/risks when considering hormonal replacement therapy or chemoprevention.