RESUMEN

PURPOSE: The purpose of this study was to report procedures developed to annotate abdominal computed tomography (CT) images from subjects without pancreatic disease that will be used as the input for deep convolutional neural networks (DNN) for development of deep learning algorithms for automatic recognition of a normal pancreas. MATERIALS AND METHODS: Dual-phase contrast-enhanced volumetric CT acquired from 2005 to 2009 from potential kidney donors were retrospectively assessed. Four trained human annotators manually and sequentially annotated 22 structures in each datasets, then expert radiologists confirmed the annotation. For efficient annotation and data management, a commercial software package that supports three-dimensional segmentation was used. RESULTS: A total of 1150 dual-phase CT datasets from 575 subjects were annotated. There were 229 men and 346 women (mean age: 45±12years; range: 18-79years). The mean intra-observer intra-subject dual-phase CT volume difference of all annotated structures was 4.27mL (7.65%). The deep network prediction for multi-organ segmentation showed high fidelity with 89.4% and 1.29mm in terms of mean Dice similarity coefficients and mean surface distances, respectively. CONCLUSIONS: A reliable data collection/annotation process for abdominal structures was developed. This process can be used to generate large datasets appropriate for deep learning.

Asunto(s)

RESUMEN

BACKGROUND: Jejunoileitis (JI) is an unusual manifestation of Crohn's disease (CD) that has been associated with high morbidity and the frequent need for surgical intervention. Although the disease has been well-described in adults, the true prevalence and clinical phenotype in children is unknown. AIM: To compare the clinical course and nutritional impact of CD in children with and without proximal small bowel involvement. METHODS: Patients with either Crohn's jejunitis or JI with or without colonic involvement were identified through a clinical database (1996--2002). All radiologic studies were reviewed by an experienced radiologist blinded to the clinical diagnosis. Thirty-six patients with CD without histologic or radiologic signs of proximal small bowel involvement were used for comparison. All medical, surgical, and hematologic parameters were compared in both disease groups. RESULTS: Among the 134 patients with CD, 23 (17%) had radiologic signs of JI, including intestinal fold thickening (57%), luminal narrowing (31%), and skip lesions (13%). Enteric fistula (6%) and strictures (6%) were less common. Patients with JI were likely to be stunted at the time of diagnosis, require surgical intervention (P < 0.03) and nutritional therapy in the form of nasogastric tube feeds (P < 0.03). Nutritional therapy was also associated with an improvement in height in patients with proximal small bowel disease (OR:5.87). DISCUSSION: JI is a relatively common disease phenotype in children with CD that requires aggressive nutritional and surgical intervention. Future studies are required to determine if the early detection and use of immune modulators may lessen the morbidity associated with proximal small bowel disease.

Asunto(s)

Asunto(s)

Asunto(s)

Asunto(s)

RESUMEN

Helical CT evaluation of the abdomen and pelvis can be challenging in the cancer patient. This pictorial essay illustrates the important sites of neoplastic dissemination in the abdomen and pelvis. We will emphasize the classic CT appearance of several selected sites and indicate, where appropriate, the primary lesions usually associated with each.

Asunto(s)

RESUMEN

Mesenteric ischemia is a complicated disorder whose prevalence in the United States is increasing as the population ages. It is often difficult to diagnose, both clinically and radiologically. In the past, computed tomography (CT) has allowed only limited success in the early detection of ischemia. However, with the introduction of multi-detector row CT and three-dimensional (3D) imaging, it is now possible to perform a detailed CT examination of the small bowel and mesenteric vessels. Multi-detector row CT allows routine studies to be performed much faster than with single-detector CT scanners and makes available new applications, especially in the field of CT angiography. Its increased speed and narrower collimation, coupled with the use of water as an oral contrast agent, improve visualization of the bowel wall and mesenteric vasculature. Multi-detector row CT with 3D reformatting may improve the ability to make an early diagnosis and identify the cause of disease in patients with suspected acute or chronic mesenteric ischemia. In many cases, this examination has eliminated the need for additional imaging studies such as Doppler ultrasonography or angiography. Further investigation will be needed to determine the scope of the utility of multi-detector row CT in this clinical setting.

Asunto(s)

RESUMEN

Renal cell carcinoma is the most common primary tumor of the kidney, with more than 30,000 new cases diagnosed in the United States each year. With the widespread use of cross-sectional imaging, many tumors are detected incidentally. Single- and multidetector computed tomography (CT) have helped refine the diagnostic work-up of renal masses by allowing image acquisition in various phases of renal enhancement after intravenous administration of a single bolus of contrast material. The scanning protocol should include unenhanced CT followed by imaging during the corticomedullary and nephrographic phases of enhancement. The nephrographic phase is the most sensitive for tumoral detection, while the corticomedullary phase is essential for imaging the renal veins for possible tumoral extension and the parenchymal organs for potential metastases. Knowledge of the tumoral stage at the time of diagnosis is essential for prognosis and surgical planning. The accuracy of CT for staging has been reported to reach 91%, with most staging errors related to the diagnosis of perinephric extension of tumor. Three-dimensional CT provides the urologist with an interactive road map of the relationships among the tumor, the major vessels, and the collecting system. This information is particularly critical if the tumor extends into the inferior vena cava and if nephron-sparing surgery is being planned.

Asunto(s)

RESUMEN

BACKGROUND: We investigated whether the left adrenal vein could be consistently localized on three-dimensional (3D) real-time volume-rendering computed tomographic angiographic (CTA) mapping in a group of living renal donors before laparoscopic nephrectomy. METHODS: Sixty-six consecutive potential renal donors were referred for CTA vascular mapping before laparoscopic donor nephrectomy. Thirty-three patients were examined on a single-detector helical CT scanner and the other 33 were examined on a multidetector unit. In each patient, arterial phase and venous phase volumetric data sets were acquired after the intravenous injection of 150 cc of non-ionic contrast material. Three radiologists reviewed the data sets at a free-standing workstation after the application of 3D volume-rendering software and reached a consensus on whether the left adrenal vein was visualized and, when seen, its position relative to the abdominal aorta. RESULTS: The left adrenal vein was found in 92.5% of the 66 donors (91% in the single-detector group and 94% in the multidetector group). The junction of the left adrenal vein and left renal vein averaged 5.2 mm (range = 0-13 mm) from the left lateral wall of the abdominal aorta. CONCLUSION: Three-dimensional real-time volume-rendering CTA with single-detector and multidetector scanners permits consistent localization of the left adrenal vein in more than 90 % of prospective living renal donors.

Asunto(s)

RESUMEN

Important sites of metastasis from intrathoracic and extrathoracic primary neoplasms are depicted on helical CT of the neck and thorax. This pictorial essay reviews several of these key sites of dissemination. For each, we illustrate the typical or classic appearance on helical CT and note the commonly associated primary malignancies.

Asunto(s)

Asunto(s)

Asunto(s)

Asunto(s)

RESUMEN

Over the past two decades, there have been significant technical advancements in computed tomography (CT). This has allowed CT to remain the gold standard for the evaluation of pancreatic pathology despite the advent of other imaging modalities, including MRI, PET and endoscopic ultrasound. Initially, CT scanners could only obtain 10-mm-thick slices at a rate of 4 slices per minute. Today, the current state of the art is multidetector CT (MDCT) technology, which allows the entire pancreas to be imaged by 1-mm slices in under 20 s. In addition, these new scanners allow true volume acquisition. The resultant data sets can be displayed not only as axial slices but also as a three-dimensional (3D) volume. The detail of these reconstructions when performed with volume rendering and maximum intensity projection techniques allows a detailed vascular mapping with accuracy that may exceed classic angiography. The use of thin collimation and dual-phase acquisition also improves the detection of hepatic metastasis as well as other sites of extrapancreatic disease. This article reviews the current state of the art of pancreatic imaging with specific emphasis on the use of MDCT, volume acquisitions and 3D arterial- and venous-phase vascular mapping. The advantages of these techniques and their impact on diagnosis and patient management are also addressed.

Asunto(s)

RESUMEN

Multidetector CT (MDCT) provides unparalleled capabilities for combining narrow scan collimation with rapid data acquisition protocols. When combined with CT angiographic techniques and 3D-volume rendering we are able to create unique displays for evaluating a range of clinical pathologies. In this pictorial review we present the potential advantages of using MDCT angiography for the evaluation of pancreatic cancer and its role in the accurate staging of these patients. The use of dual-phase CT scanning in both the arterial phase and portal phase is addressed with the role of 3D CT angiography clearly defined. Numerous case studies are presented to show the advantages of these techniques over simple axial CT imaging.

Asunto(s)

RESUMEN

PURPOSE: To use radiologic-histopathologic correlation in an animal model to distinguish normal postoperative findings from evidence of residual tumor after cryoablation of malignant hepatic tumors. MATERIALS AND METHODS: Hepatic cryoablation was performed in 12 rabbits with VX2 tumors and in two healthy rabbits. Nonenhanced and dynamic contrast material-enhanced computed tomography (CT) and magnetic resonance (MR) imaging and power and color Doppler flow ultrasonography (US) were performed 7-8 days after cryoablation. Histopathologic findings were correlated with imaging findings. RESULTS: Twenty tumors of 5-20 mm (mean, 10 mm) and seven areas of normal liver were treated with cryolesions of 11-21 mm (mean, 15 mm). All cryolesions exhibited arterial phase rim enhancement at CT and MR imaging, and 13 (57%) of 23 lesions demonstrated peripheral flow at US because of granulation tissue. There was macroscopic recurrence in 15 (75%) of 20 treated tumors; 14 (93%) appeared as peripheral nodularity with low-grade enhancement. Necrotic tissue did not enhance. Intact vessels extended up to 6 mm inside cryolesion margins and caused focal internal enhancement and Doppler flow. Areas of high signal intensity on T2-weighted MR images correlated with liquefaction necrosis, granulation tissue, and tumor. CONCLUSION: In this animal model, recurrent tumor typically appeared as focal nodules at the cryolesion periphery. Rim and central foci of enhancement, Doppler flow, and increased signal intensity on T2-weighted MR images can be normal findings after hepatic cryoablation.

Asunto(s)

Asunto(s)

Asunto(s)

RESUMEN

OBJECTIVE: The purpose of this study was to determine whether CT can be used to distinguish serous cystadenomas from mucinous cystadenomas or cystadenocarcinomas of the pancreas and play an enhanced role in patient triage and treatment. MATERIALS AND METHODS: A blinded retrospective analysis of CT scans from 50 patients with pathologically proven primary cystic pancreatic neoplasms was performed independently by three radiologists. Using classic CT criteria as reported in the literature, each tumor was categorized as definitely serous, mucinous, or indeterminate. Tumor location, size, presence of calcification, and size of largest cyst were recorded. Data for each reviewer were analyzed independently. Consensus data were then subjected to multivariate logistic regression analysis. RESULTS: The ability of our reviewers to diagnose serous neoplasms ranged from 23% to 41%. Eight mucinous neoplasms were mistaken for serous tumors by two of the three reviewers. When consensus between at least two of the three reviewers was used for diagnosis, 27% of serous neoplasms were correctly diagnosed and all of the mucinous tumors were correctly identified as uncertain or mucinous, yielding the same clinical end point. For multivariate logistic regression analysis, a cyst smaller than 2 cm had a statistically significant association (p = 0.005) with serous tumors, and the presence of peripheral tumoral calcification had a statistically significant association (p = 0.01) with mucinous tumors. CONCLUSION: There is significant variability in the CT appearance of serous and mucinous neoplasms of the pancreas, making CT an insensitive tool for differentiating these tumors. All tumors with peripheral calcifications were identified as mucinous neoplasms.

Asunto(s)