Deep learning, reusable and problem-based architectures for detection of consolidation on chest X-ray images.

BACKGROUND AND OBJECTIVE: In most patients presenting with respiratory symptoms, the findings of chest radiography play a key role in the diagnosis, management, and follow-up of the disease. Consolidation is a common term in radiology, which indicates focally increased lung density. When the alveolar structures become filled with pus, fluid, blood cells or protein subsequent to a pulmonary pathological process, it may result in different types of lung opacity in chest radiograph. This study aims at detecting consolidations in chest x-ray radiographs, with a certain precision, using artificial intelligence and especially Deep Convolutional Neural Networks to assist radiologist for better diagnosis. METHODS: Medical image datasets usually are relatively small to be used for training a Deep Convolutional Neural Network (DCNN), so transfer learning technique with well-known DCNNs pre-trained with ImageNet dataset are used to improve the accuracy of the models. ImageNet feature space is different from medical images and in the other side, the well-known DCNNs are designed to achieve the best performance on ImageNet. Therefore, they cannot show their best performance on medical images. To overcome this problem, we designed a problem-based architecture which preserves the information of images for detecting consolidation in Pediatric Chest X-ray dataset. We proposed a three-step pre-processing approach to enhance generalization of the models. To demonstrate the correctness of numerical results, an occlusion test is applied to visualize outputs of the model and localize the detected appropriate area. A different dataset as an extra validation is used in order to investigate the generalization of the proposed model. RESULTS: The best accuracy to detect consolidation is 94.67% obtained by our problem based architecture for the understudy dataset which outperforms the previous works and the other architectures. CONCLUSIONS: The designed models can be employed as computer aided diagnosis tools in real practice. We critically discussed the datasets and the previous works based on them and show that without some considerations the results of them may be misleading. We believe, the output of AI should be only interpreted as focal consolidation. The clinical significance of the finding can not be interpreted without integration of clinical data.

Management of Primary Osseous Spinal Tumors with PET.

Knowledge of the PET imaging findings of osseous spinal neoplasms is essential, because they are common incidental findings on PET scans done for staging of unrelated primary malignancies. Additionally, PET can help differentiate lesions that are not clearly defined by anatomic modalities alone. PET can also be used for follow-up of aggressive tumors to assess response to treatment, often proving superior to CT or MR imaging alone for this purpose. This review discusses the role of PET/CT and PET/MR imaging in the diagnosis and management of primary benign and malignant osseous tumors of the spine.

Clinical Nononcologic Applications of PET/CT and PET/MRI in Musculoskeletal, Orthopedic, and Rheumatologic Imaging.

OBJECTIVE: With improvements in PET/CT and PET/MRI over the last decade, as well as increased understanding of the pathophysiology of musculoskeletal diseases, there is an emerging potential for PET as a primary or complementary modality in the management of rheumatologic and orthopedic conditions. CONCLUSION: We discuss the role of PET/CT and PET/MRI in nononcologic musculoskeletal disorders, including inflammatory and infectious conditions and postoperative complications. There is great potential for an increased role for PET to serve as a primary or complementary modality in the management of orthopedic and rheumatologic disorders.

Diagnosis of non-osseous spinal metastatic disease: the role of PET/CT and PET/MRI.

The spine is the third most common site for distant metastasis in cancer patients with approximately 70% of patients with metastatic cancer having spinal involvement. Positron emission tomography (PET), combined with computed tomography (CT) or magnetic resonance imaging (MRI), has been deeply integrated in modern clinical oncology as a pivotal component of the diagnostic work-up of patients with cancer. PET is able to diagnose several neoplastic processes before any detectable morphological changes can be identified by anatomic imaging modalities alone. In this review, we discuss the role of PET/CT and PET/MRI in the diagnostic management of non-osseous metastatic disease of the spinal canal. While sometimes subtle, recognizing such disease on FDG PET/CT and PET/MRI imaging done routinely in cancer patients can guide treatment strategies to potentially prevent irreversible neurological damage.

Non-oncologic Applications of PET/CT and PET/MR in Musculoskeletal, Orthopedic, and Rheumatologic Imaging: General Considerations, Techniques, and Radiopharmaceuticals.

Positron Emission Tomography (PET) is often underutilized in the field of musculoskeletal imaging, with key reasons including the excellent performance of conventional musculoskeletal MRI, the limited spatial resolution of PET, and the lack of reimbursement for PET for non-oncologic musculoskeletal indications. However, with improvements in PET/CT and PET/MR imaging over the last decade as well as an increased understanding of the pathophysiology of musculoskeletal diseases, there is an emerging potential for PET as a primary or complementary modality in the management of rheumatologic and orthopedic patients. Specific advantages of PET include the convenience of whole body imaging in a single session, the relative resilience of the modality in the imaging of metallic implants compared to CT and MRI, the ability to evaluate deep joints not amenable to palpation, and the potential for improved specificity of diagnosis with novel radiopharmaceuticals. In this review, we discuss multiple radiopharmaceuticals and technical consideration of PET/CT and PET/MRI that can be employed in imaging of non-tumoral bone and soft tissue disorders. Both PET/CT and PET/MR hold significant promise in the field of musculoskeletal imaging, and with further radiopharmaceutical development and clinical research, these hybrid modalities can potentially transform the current management of patients with orthopedic and rheumatologic disease.

A Pediatric Tumor Found Frequently in the Adult Population: A Case of Anaplastic Astroblastoma in an Elderly Patient and Review of the Literature.

Astroblastomas are rare, potentially curable primary brain tumors which can be difficult to diagnose. We present the case of astroblastoma in a 73-year-old male, an atypical age for this tumor, more classically found in pediatric and young adult populations. Through our case and review of the literature, we note that this tumor is frequently reported in adult populations and the presentation of this tumor in the elderly is well described. This tumor is an important consideration in the differential diagnosis when managing both pediatric and adult patients of any age who present with the imaging findings characteristic of this rare tumor.

Primary Splenic Angiosarcoma: Clinical and Imaging Manifestations of This Rare Aggressive Neoplasm.

Primary splenic angiosarcoma is a rare and fatal neoplasm arising from vascular endothelial cells within the spleen. With an incidence of 2 cases per 10 million people worldwide, the diagnosis and treatment of this rare entity is unfamiliar and challenging. We describe the case of a previously healthy 45-year-old woman who presented with vague upper-abdominal pain and was found to have a splenic mass on computed tomography. The patient underwent laparoscopic splenectomy and was found to have splenic angiosarcoma on microscopic evaluation. Although specific radiologic diagnosis is not possible, bringing the possibility of primary splenic angiosarcoma to the ordering clinician's attention has the potential to hasten treatment and improve patient outcomes. This case highlights the importance for radiologists to be aware of this rare neoplasm and to consider it in the differential when encountering a heterogeneously enhancing splenic mass.

The global cancer divide: relationships between national healthcare resources and cancer outcomes in high-income vs. middle- and low-income countries.

BACKGROUND: Cancer continues to rise as a contributor to premature death in the developing world. Despite this, little is known about whether cancer outcomes are related to a country's income level, and what aspects of national healthcare systems are associated with improved cancer outcomes. METHODS: The most recent estimates of cancer incidence and mortality were used to calculate mortality-to-incidence ratio (MIR) for the 85 countries with reliable data. Countries were categorized according to high-income (Gross Domestic Product (GDP)>$15,000) or middle/low-income (GDP<$15,000), and a multivariate linear regression model was used to determine the association between healthcare system indicators and cancer MIR. Indicators study included per capita GDP, overall total healthcare expenditure (THE), THE as a proportion of GDP, total external beam radiotherapy devices (TEBD) per capita, physician density, and the year 2000 WHO healthcare system rankings. RESULTS: Cancer MIR in high-income countries (0.47) was significantly lower than that of middle/low-income countries (0.64), with a p<0.001. In high-income countries, GDP, health expenditure and TEBD showed significant inverse correlations with overall cancer MIR. A $3040 increase in GDP (p=0.004), a $379 increase in THE (p<0.001), or an increase of 0.59 TEBD per 100,000 population (p=0.027) were all associated with a 0.01 decrease in cancer MIR. In middle/low-income countries, only WHO scores correlated with decreased cancer MIR (p=0.022); 12 specific cancer types also showed similar significant correlations (p<0.05) as overall cancer MIR. CONCLUSIONS: The analysis of this study suggested that cancer MIR is greater in middle/low-income countries. Furthermore, the WHO healthcare score was associated with improved cancer outcomes in middle/low-income countries while absolute levels of financial resources and infrastructure played a more important role in high-income countries.

Malignant melanoma and radiotherapy: past myths, excellent local control in 146 studied lesions at Georgetown University, and improving future management.

INTRODUCTION: Once thought to be radioresistant, emerging cellular and clinical evidence now suggests melanoma can respond to large radiation doses per fraction. MATERIALS AND METHODS: We conducted a retrospective study of all patients treated with stereotactic radiosurgery and stereotactic body radiotherapy at Georgetown University Hospital from May 2002 through November 2008 and studied the classic extrapolated total dose corrected for volume (ETD(vol)) model for predicting melanoma tumor response. Region-specific tumor outcomes were categorized by RECIST criteria and local control curves were estimated and analyzed when stratified by ETD(vol) thresholds by use of the Kaplan-Meier method. RESULTS: Follow-up information was available for 78 lesions (49 intracranial, 8 spinal, and 21 body) with mean follow-up period of 9.2 (range, 2-36) months. 1-year local control rates for intracranial, spinal, and body tumors were 84, 100, and 72%, respectively. Treatments in general were well-tolerated. Increased ETD(vol) (p < 0.001) among intracranial sites resulted from larger (p < 0.001) doses per fraction combined with smaller (p < 0.001) tumor diameters. Intracranial 6-, 12-, and 24-month local control rates when treated above ETD(vol) threshold of 230 Gy were all 90 vs. 89, 80, and 53% below this threshold. Body 6- and 12-month local control rates when treated above ETD(vol) threshold of 100 Gy were 100 and 80% vs. 74 and 59% below this threshold. DISCUSSION: By tailoring to melanoma's unique radiobiology with large radiation doses per fraction, favorable local control was safely achieved. The ETD(vol) model combines the important factor of dose per fraction in melanoma treatment with a volume correction factor to predict tumor response. Although limited sample size may have prevented reaching statistical significance for local control improvements using ETD(vol) thresholds, optimal thresholds may exist to improve future tumor responses and further research is required.