Most neurocytomas are well differentiated, being associated with better long-term survival than the more aggressive atypical lesions. Atypical neurocytomas are characterized by an MIB-1 labeling index >3% or atypical histologic features. This analysis focuses on well differentiated neurocytomas in order to define the optimal treatment. A case with a follow-up of 132 months is presented. The patient developed two recurrences two and four years after first surgery, each showing an increasing proliferation activity. Furthermore, all published well-differentiated neurocytoma cases were reviewed for surgery, radiotherapy, and prognosis. Additional relevant data were obtained from the authors. Complete resection (CTR), complete resection plus radiotherapy (CTR + RT), incomplete resection (ITR), and incomplete resection plus radiotherapy (ITR + RT) were compared for outcome by using the Kaplan-Meier method and the log-rank test. Data were complete in 301 patients (CTR, 108; CTR + RT, 27; ITR, 81; ITR + RT, 85). Local control and survival were better after CTR than after ITR (P < 0.0001 and P = 0.0085, respectively). Radiotherapy improved local control after ITR (P < 0.0001) and after CTR (P = 0.0474), but not survival (P = 0.17 and P = 1.0, respectively). In the ITR + RT group, doses < or =54 Gy (n = 33) and >54 Gy (n = 32) were not significantly different for local control (P = 0.88) and survival (P = 0.95). The data demonstrated CTR to be superior to ITR for local control and survival. After CTR and ITR, radiotherapy improved local control, but not survival. A radiation dose of 54 Gy appeared sufficient. Application of postoperative radiotherapy should be decided individually, taking into account the risk of local failure, the need for another craniotomy, and potential radiation toxicity.

Deep learning algorithms, in particular convolutional networks, have rapidly become a methodology of choice for analyzing medical images. This paper reviews the major deep learning concepts pertinent to medical image analysis and summarizes over 300 contributions to the field, most of which appeared in the last year. We survey the use of deep learning for image classification, object detection, segmentation, registration, and other tasks. Concise overviews are provided of studies per application area: neuro, retinal, pulmonary, digital pathology, breast, cardiac, abdominal, musculoskeletal. We end with a summary of the current state-of-the-art, a critical discussion of open challenges and directions for future research.Copyright © 2017 Elsevier B.V. All rights reserved.

The segmentation of infant brain tissue images into white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) plays an important role in studying early brain development in health and disease. In the isointense stage (approximately 6-8 months of age), WM and GM exhibit similar levels of intensity in both T1 and T2 MR images, making the tissue segmentation very challenging. Only a small number of existing methods have been designed for tissue segmentation in this isointense stage; however, they only used a single T1 or T2 images, or the combination of T1 and T2 images. In this paper, we propose to use deep convolutional neural networks (CNNs) for segmenting isointense stage brain tissues using multi-modality MR images. CNNs are a type of deep models in which trainable filters and local neighborhood pooling operations are applied alternatingly on the raw input images, resulting in a hierarchy of increasingly complex features. Specifically, we used multi-modality information from T1, T2, and fractional anisotropy (FA) images as inputs and then generated the segmentation maps as outputs. The multiple intermediate layers applied convolution, pooling, normalization, and other operations to capture the highly nonlinear mappings between inputs and outputs. We compared the performance of our approach with that of the commonly used segmentation methods on a set of manually segmented isointense stage brain images. Results showed that our proposed model significantly outperformed prior methods on infant brain tissue segmentation. In addition, our results indicated that integration of multi-modality images led to significant performance improvement. Copyright © 2014 Elsevier Inc. All rights reserved.

Automatic segmentation in MR brain images is important for quantitative analysis in large-scale studies with images acquired at all ages. This paper presents a method for the automatic segmentation of MR brain images into a number of tissue classes using a convolutional neural network. To ensure that the method obtains accurate segmentation details as well as spatial consistency, the network uses multiple patch sizes and multiple convolution kernel sizes to acquire multi-scale information about each voxel. The method is not dependent on explicit features, but learns to recognise the information that is important for the classification based on training data. The method requires a single anatomical MR image only. The segmentation method is applied to five different data sets: coronal T-weighted images of preterm infants acquired at 30 weeks postmenstrual age (PMA) and 40 weeks PMA, axial T-weighted images of preterm infants acquired at 40 weeks PMA, axial T-weighted images of ageing adults acquired at an average age of 70 years, and T-weighted images of young adults acquired at an average age of 23 years. The method obtained the following average Dice coefficients over all segmented tissue classes for each data set, respectively: 0.87, 0.82, 0.84, 0.86, and 0.91. The results demonstrate that the method obtains accurate segmentations in all five sets, and hence demonstrates its robustness to differences in age and acquisition protocol.

The purpose of this study was to assess the potential of a deep learning model to discriminate between benign and malignant breast lesions using magnetic resonance imaging (MRI) and characterize different histological subtypes of breast lesions.We developed a deep learning model that simultaneously learns to detect lesions and characterize them. We created a lesion-characterization model based on a single two-dimensional T1-weighted fat suppressed MR image obtained after intravenous administration of a gadolinium chelate selected by radiologists. The data included 335 MR images from 335 patients, representing 17 different histological subtypes of breast lesions grouped into four categories (mammary gland, benign lesions, invasive ductal carcinoma and other malignant lesions). Algorithm performance was evaluated on an independent test set of 168 MR images using weighted sums of the area under the curve (AUC) scores.We obtained a cross-validation score of 0.817 weighted average receiver operating characteristic (ROC)-AUC on the training set computed as the mean of three-shuffle three-fold cross-validation. Our model reached a weighted mean AUC of 0.816 on the independent challenge test set.This study shows good performance of a supervised-attention model with deep learning for breast MRI. This method should be validated on a larger and independent cohort.Copyright © 2019 Soci showét showé françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.

In this paper, we present a fully automatic brain tumor segmentation method based on Deep Neural Networks (DNNs). The proposed networks are tailored to glioblastomas (both low and high grade) pictured in MR images. By their very nature, these tumors can appear anywhere in the brain and have almost any kind of shape, size, and contrast. These reasons motivate our exploration of a machine learning solution that exploits a flexible, high capacity DNN while being extremely efficient. Here, we give a description of different model choices that we've found to be necessary for obtaining competitive performance. We explore in particular different architectures based on Convolutional Neural Networks (CNN), i.e. DNNs specifically adapted to image data. We present a novel CNN architecture which differs from those traditionally used in computer vision. Our CNN exploits both local features as well as more global contextual features simultaneously. Also, different from most traditional uses of CNNs, our networks use a final layer that is a convolutional implementation of a fully connected layer which allows a 40 fold speed up. We also describe a 2-phase training procedure that allows us to tackle difficulties related to the imbalance of tumor labels. Finally, we explore a cascade architecture in which the output of a basic CNN is treated as an additional source of information for a subsequent CNN. Results reported on the 2013 BRATS test data-set reveal that our architecture improves over the currently published state-of-the-art while being over 30 times faster.Copyright © 2016 Elsevier B.V. All rights reserved.

为解决脑脊液病理图像中部分细胞膜较为模糊，与图像背景难以区分的问题，采用了基于注意力机制的U-Net深度学习方法对脑脊液病理图像做全自动分割.在深度学习网络中加入注意力机制对细胞进行定位，抑制无关信息，提高语义的特征表达，提高对细胞整体分割的精确性.通过镜像、旋转等操作对数据集进行扩充预处理.采用VGG16预训练模型进行迁移学习，交叉熵与Dice损失相结合作为损失函数，分别在脑脊液临床图像与公开数据集2018 Data Science Bowl上进行验证；并与Otsu， PSPnet， Segnet， DeeplabV3+， U-Net进行对比，结果表明， 本文方法在各项指标上均优于其他分割方法.

为解决脑脊液病理图像中部分细胞膜较为模糊，与图像背景难以区分的问题，采用了基于注意力机制的U-Net深度学习方法对脑脊液病理图像做全自动分割.在深度学习网络中加入注意力机制对细胞进行定位，抑制无关信息，提高语义的特征表达，提高对细胞整体分割的精确性.通过镜像、旋转等操作对数据集进行扩充预处理.采用VGG16预训练模型进行迁移学习，交叉熵与Dice损失相结合作为损失函数，分别在脑脊液临床图像与公开数据集2018 Data Science Bowl上进行验证；并与Otsu， PSPnet， Segnet， DeeplabV3+， U-Net进行对比，结果表明， 本文方法在各项指标上均优于其他分割方法.

Mammary images are enhanced by algorithm of Contrast Limited Adaptive Histogram Equalization（CLAHE）.Details of mammograph，such as calcifications breast，duct and other tissues in mammary image are enhanced effectively.Parameters of algorithm are optimized to improve the enhancement so that diagnostician can analyze image expediently.Comparing with the result of histogram equalization，it is shown that CLAHE is an effective algorithm to enhance mammary image and it is valuable in computer-aided mammary diagnosis.

Mammary images are enhanced by algorithm of Contrast Limited Adaptive Histogram Equalization（CLAHE）.Details of mammograph，such as calcifications breast，duct and other tissues in mammary image are enhanced effectively.Parameters of algorithm are optimized to improve the enhancement so that diagnostician can analyze image expediently.Comparing with the result of histogram equalization，it is shown that CLAHE is an effective algorithm to enhance mammary image and it is valuable in computer-aided mammary diagnosis.

Recent outstanding results of supervised object detection in competitions and challenges are often associated with specific metrics and datasets. The evaluation of such methods applied in different contexts have increased the demand for annotated datasets. Annotation tools represent the location and size of objects in distinct formats, leading to a lack of consensus on the representation. Such a scenario often complicates the comparison of object detection methods. This work alleviates this problem along the following lines: (i) It provides an overview of the most relevant evaluation methods used in object detection competitions, highlighting their peculiarities, differences, and advantages; (ii) it examines the most used annotation formats, showing how different implementations may influence the assessment results; and (iii) it provides a novel open-source toolkit supporting different annotation formats and 15 performance metrics, making it easy for researchers to evaluate the performance of their detection algorithms in most known datasets. In addition, this work proposes a new metric, also included in the toolkit, for evaluating object detection in videos that is based on the spatio-temporal overlap between the ground-truth and detected bounding boxes.

The aim of this study was to identify factors associated with delayed cranial neuropathy following radiosurgery for vestibular schwannoma (VS or acoustic neuroma) and to determine how such factors may be manipulated to minimize the incidence of radiosurgical complications while maintaining high rates of tumor control.From July 1988 to June 1998, 149 cases of VS were treated using linear accelerator radiosurgery at the University of Florida. In each of these cases, the patient's tumor and brainstem were contoured in 1-mm slices on the original radiosurgical targeting images. Resulting tumor and brainstem volumes were coupled with the original radiosurgery plans to generate dose-volume histograms. Various tumor dimensions were also measured to estimate the length of cranial nerve that would be irradiated. Patient follow-up data, including evidence of cranial neuropathy and radiographic tumor control, were obtained from a prospectively maintained, computerized database. The authors performed statistical analyses to compare the incidence of posttreatment cranial neuropathies or tumor growth between patient strata defined by risk factors of interest. One hundred thirty-nine of the 149 patients were included in the analysis of complications. The median duration of clinical follow up for this group was 36 months (range 18-94 months). The tumor control analysis included 133 patients. The median duration of radiological follow up in this group was 34 months (range 6-94 months). The overall 2-year actuarial incidences of facial and trigeminal neuropathies were 11.8% and 9.5%, respectively. In 41 patients treated before 1994, the incidences of facial and trigeminal neuropathies were both 29%, but in the 108 patients treated since January 1994, these rates declined to 5% and 2%, respectively. An evaluation of multiple risk factor models showed that maximum radiation dose to the brainstem, treatment era (pre-1994 compared with 1994 or later), and prior surgical resection were all simultaneously informative predictors of cranial neuropathy risk. The radiation dose prescribed to the tumor margin could be substituted for the maximum dose to the brainstem with a small loss in predictive strength. The pons-petrous tumor diameter was an additional statistically significant simultaneous predictor of trigeminal neuropathy risk, whereas the distance from the brainstem to the end of the tumor in the petrous bone was an additional marginally significant simultaneous predictor of facial neuropathy risk. The overall radiological tumor control rate was 93% (59% tumors regressed, 34% remained stable, and 7.5% enlarged), and the 5-year actuarial tumor control rate was 87% (95% confidence interval [CI] 76-98%). Analysis revealed that a radiation dose cutpoint of 10 Gy compared with more than 10 Gy prescribed to the tumor margin yielded the greatest relative difference in tumor growth risk (relative risk 2.4, 95% CI 0.6-9.3), although this difference was not statistically significant (p = 0.207).Five points must be noted. 1) Radiosurgery is a safe, effective treatment for small VSs. 2) Reduction in the radiation dose has played the most important role in reducing the complications associated with VS radiosurgery. 3) The dose to the brainstem is a more informative predictor of postradiosurgical cranial neuropathy than the length of the nerve that is irradiated. 4) Prior resection increases the risk of late cranial neuropathies after radiosurgery. 5) A prescription dose of 12.5 Gy to the tumor margin resulted in the best combination of maximum tumor control and minimum complications in this series.