Purpose:To investigate the use of four-dimensional (4D) co-occurrence-based texture analysis to distinguish between nonmalignant and malignant tissues in dynamic contrast-enhanced (DCE) MR images. Materials and Methods:4D texture analysis was performed on DCE-MRI data sets of breast lesions. A modelfree neural network-based classification system assigned each voxel a "nonmalignant" or "malignant" label based on the textural features. The classification results were compared via receiver operating characteristic (ROC) curve analysis with the manual lesion segmentation produced by two radiologists (observers 1 and 2). Results:The mean sensitivity and specificity of the classifier agreed with the mean observer 2 performance when compared with segmentations by observer 1 for a 95% confidence interval, using a two-sided t-test with ␣ ϭ 0.05. The results show that an area under the ROC curve (A z ) of 0.99948, 0.99867, and 0.99957 can be achieved by comparing the classifier vs. observer 1, classifier vs. union of both observers, and classifier vs. intersection of both observers, respectively. Conclusion:This study shows that a neural network classifier based on 4D texture analysis inputs can achieve a performance comparable to that achieved by human observers, and that further research in this area is warranted.

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.

We describe a case of primary pulmonary synovial sarcoma. A 19-year-old man presented with low-grade fever, dyspnea, chest pain and left arm numbness. Chest radiographs revealed a large, well-circumscribed left perihilar mass and a small ipsilateral pleural effusion. Chest computed tomography (CT) revealed a large well-defined, heterogeneous lung mass. Magnetic resonance imaging (MRI) demonstrated a mass of heterogeneous signal intensity on T1-weighted and proton density images, and high signal intensity on short tau inversion recovery (STIR) images. Whole-body bone scintigraphy showed no evidence of skeletal involvement. Abdominal and pelvic CT showed no intra-abominal or pelvic metastases. A CT-guided biopsy provided the diagnosis of monophasic synovial sarcoma. Following four cycles of chemotherapy, integrated F-18 fluorodeoxyglucose positron emission tomography-computed tomography (18F FDG PET/CT) was performed, which demonstrated interval decrease in the size of the lesion and no significant metabolic activity. Surgical resection was then undertaken. Microscopically, the lesion was a high-grade spindle cell sarcoma consistent with monophasic synovial sarcoma. A variant X;18 translocation was identified by cytogenetic analysis and confirmed with metaphase in situ hybridization. The imaging and pathological features of this rare lesion are reviewed.

Purpose:To augment traditional visual data perception of complex multiparametric imaging data sets by adding auditory feedback to improve the delineation of regions of interest (ROIs) in tumor assessment in dynamic contrastenhanced (DCE) MRI. Materials and Methods:In addition to conventional display methodologies, we have created an application window which interfaces with audio output using dynamically loadable sound modules, providing goodness of fit (GF) information through auditory feedback. We have assessed effectiveness of conveying sound information with three independent readers on eight DCE-MR breast image data sets. The assessment was based on either conventional visual only mode or combined visual plus auditory mode. For statistical comparison between two sensory approaches, interobserver repeatability was measured with three different criteria.Results: Adding auditory feedback improves repeatability significantly (P Ͻ 0.01), and the enhanced sensory approach had higher repeatability than visual only mode in visually complex breast tumor cases. However, in easy and moderate cases, visual only mode was more reproducible than the combined mode with very high significance (P Ͻ 0.001). Conclusion:Adding auditory information to visual based image analysis for identifying tumor ROIs provides higher interobserver repeatability for analyzing complex multidimensional/multiparametric medical image data sets with visually difficult lesions to delineate. TODAY'S ADVANCED IMAGING scanner systems generate multiparametric and multimodality (computed tomography [CT], MR, positron emission tomography [PET], ultrasound [US]) information in a short amount of time.It is important to simultaneously access and to coherently communicate with all these different types of image information for better lesion characterization and for more accurate and reliable diagnostic image analysis in many clinical applications. As the complexity and combination of different types of acquired imaging data increase, conveying this information to a human observer in a meaningful way for diagnostic purpose becomes a bottleneck.The traditional method of analyzing medical images solely relies on the visual perception system, using grayscale values and color overlays to represent composite imaging data or showing different types of image data in a side by side display regarding which the observer must be trained to mentally align corresponding positions in the image data sets. Furthermore, the visual representation of complex medical images is often limited by screen size, two-dimensional (2D) display of three-dimensional (3D) spatial information, image resolution, and the capacity of the human visual system. Therefore, it is not easy to quickly access valuable information due to the complexity of multiparametric imaging data sets and the limits of visual perception system.Dynamic contrast-enhanced MRI (DCE-MRI) is widely used as a powerful noninvasive clinical tool for the diagnosis of tumors and for therapy monitoring of a variety of treatment...

Thymic hyperplasia occurs in a small proportion of patients receiving chemotherapy for various malignancies. It likely results from an immunologic rebound phenomenon. Fluorodeoxyglucose positron emission tomography-computed tomography is an important tool for staging malignant neoplasms. We report a case of rebound thymic hyperplasia manifesting as a hypermetabolic mass on fluorodeoxyglucose positron emission tomography-computed tomography after pneumonectomy and chemotherapy for primary pulmonary synovial sarcoma. We highlight the importance of recognizing the phenomenon of rebound thymic hyperplasia, as it can mimic residual or recurrent malignancy, especially in the setting of altered chest anatomy.