Georgios Z. Papadakis

Holistic classification of CT attenuation patterns for interstitial lung diseases via deep convolutional neural networks

Interstitial lung diseases (ILD) involve several abnormal imaging patterns observed in computed tomography (CT) images. Accurate classification of these patterns plays a significant role in precise clinical decision making of the extent and nature of the diseases. Therefore, it is important for developing automated pulmonary computer-aided detection systems. Conventionally, this task relies on experts’ manual identification of regions of interest (ROIs) as a prerequisite to diagnose potential diseases. This protocol is time consuming and inhibits fully automatic assessment. In this paper, we present a new method to classify ILD imaging patterns on CT images. The main difference is that the proposed algorithm uses the entire image as a holistic input. By circumventing the prerequisite of manual input ROIs, our problem set-up is significantly more difficult than previous work but can better address the clinical workflow. Qualitative and quantitative results using a publicly available ILD database demonstrate state-of-the-art classification accuracy under the patch-based classification and shows the potential of predicting the ILD type using holistic image.

Segmentation and Image Analysis of Abnormal Lungs at CT: Current Approaches, Challenges, and Future Trends.

The computer-based process of identifying the boundaries of lung from surrounding thoracic tissue on computed tomographic (CT) images, which is called segmentation, is a vital first step in radiologic pulmonary image analysis. Many algorithms and software platforms provide image segmentation routines for quantification of lung abnormalities; however, nearly all of the current image segmentation approaches apply well only if the lungs exhibit minimal or no pathologic conditions. When moderate to high amounts of disease or abnormalities with a challenging shape or appearance exist in the lungs, computer-aided detection systems may be highly likely to fail to depict those abnormal regions because of inaccurate segmentation methods. In particular, abnormalities such as pleural effusions, consolidations, and masses often cause inaccurate lung segmentation, which greatly limits the use of image processing methods in clinical and research contexts. In this review, a critical summary of the current methods for lung segmentation on CT images is provided, with special emphasis on the accuracy and performance of the methods in cases with abnormalities and cases with exemplary pathologic findings. The currently available segmentation methods can be divided into five major classes: (a) thresholding-based, (b) region-based, (c) shape-based, (d) neighboring anatomy-guided, and (e) machine learning-based methods. The feasibility of each class and its shortcomings are explained and illustrated with the most common lung abnormalities observed on CT images. In an overview, practical applications and evolving technologies combining the presented approaches for the practicing radiologist are detailed.

Ectopic Adrenocorticotropic Hormone and Corticotropin-Releasing Hormone Co-Secreting Tumors in Children and Adolescents Causing Cushing Syndrome: A Diagnostic Dilemma and How to Solve It

CONTEXT Ectopic ACTH/CRH syndrome is a rare cause of Cushing syndrome (CS), especially in children. The localization, work-up, and management of ACTH/CRH-secreting tumors are discussed. SETTING A retrospective study was conducted of patients under 21 years of age evaluated at the National Institutes of Health (NIH) for CS and diagnosed with ectopic ACTH/CRH-secreting tumors during the period 2009-2014. PATIENTS Seven patients with ectopic ACTH/CRH CS are included in this study with a median age 13.6 years (range 1-21), and 3 are female. MEASUREMENTS Clinical, biochemical, radiological features, treatment, and histological findings are described. RESULTS Seven patients were found to have ACTH/CRH-secreting tumors, all with neuroendocrine features. The site of the primary lesion varied: pancreas (3), thymus (2), liver (1), right lower pulmonary lobe (1). PATIENTS underwent biochemical evaluation for CS, including diurnal serum cortisol and ACTH levels, urinary free cortisol levels (UFC), and CRH stimulation tests. All patients underwent radiological investigations including MRI, CT, and PET scan; imaging with octreotide and 68 gallium DOTATATE scans were performed in individual cases. Five patients underwent inferior petrosal sinus sampling; 4 patients had sampling for ACTH and CRH levels from additional sites. Three patients underwent trans-sphenoidal surgery (TSS), and 3 patients required bilateral adrenalectomy. Three patients (43%) died due to metastatic disease, demonstrating the high mortality rate. One of the unique findings in these seven patients is that in each case, their neuroendocrine tumors were ultimately proven to be co-secreting ACTH and CRH. This explains the enigmatic presentation, in which 3 patients initially thought to have Cushings disease (CD) with corresponding pituitary hyperplasia underwent TSS prior to the correct localization of the causative tumor. CONCLUSIONS Ectopic ACTH/CRH co-secreting tumors are extremely rare in children and adolescents. The diagnosis of this condition is frequently missed and is sometimes confused with CD due to the effect of CRH on the pituitary.

Acoustic detection of DNA conformation in genetic assays combined with PCR

Application of PCR to multiplexing assays is not trivial; it requires multiple fluorescent labels for amplicon detection and sophisticated software for data interpretation. Alternative PCR-free methods exploiting new concepts in nanotechnology exhibit high sensitivities but require multiple labeling and/or amplification steps. Here, we propose to simplify the problem of simultaneous analysis of multiple targets in genetic assays by detecting directly the conformation, rather than mass, of target amplicons produced in the same PCR reaction. The new methodology exploits acoustic wave devices which are shown to be able to characterize in a fully quantitative manner multiple double stranded DNAs of various lengths. The generic nature of the combined acoustic/PCR platform is shown using real samples and, specifically, during the detection of SNP genotyping in Anopheles gambiae and gene expression quantification in treated mice. The method possesses significant advantages to TaqMan assay and real-time PCR regarding multiplexing capability, speed, simplicity and cost.

Somatic USP8 Gene Mutations Are a Common Cause of Pediatric Cushing Disease

Context: Somatic mutations in the ubiquitin-specific protease 8 (USP8) gene have been recently identified as the most common genetic alteration in patients with Cushing disease (CD). However, the frequency of these mutations in the pediatric population has not been extensively assessed. Objective: We investigated the status of the USP8 gene at the somatic level in a cohort of pediatric patients with corticotroph adenomas. Design and Methods: The USP8 gene was fully sequenced in both germline and tumor DNA samples from 42 pediatric patients with CD. Clinical, biochemical, and imaging data were compared between patients with and without somatic USP8 mutations. Results: Five different USP8 mutations (three missense, one frameshift, and one in-frame deletion) were identified in 13 patients (31%), all of them located in exon 14 at the previously described mutational hotspot, affecting the 14-3-3 binding motif of the protein. Patients with somatic mutations were older at disease presentation [mean 5.1 ± 2.1 standard deviation (SD) vs 13.1 ± 3.6 years, P = 0.03]. Levels of urinary free cortisol, midnight serum cortisol, and adrenocorticotropic hormone, as well as tumor size and frequency of invasion of the cavernous sinus, were not significantly different between the two groups. However, patients harboring somatic USP8 mutations had a higher likelihood of recurrence compared with patients without mutations (46.2% vs 10.3%, P = 0.009). Conclusion: Somatic USP8 gene mutations are a common cause of pediatric CD. Patients harboring a somatic mutation had a higher likelihood of tumor recurrence, highlighting the potential importance of this molecular defect for the disease prognosis and the development of targeted therapeutic options.

Ectopic ACTH and CRH Co-secreting Tumor Localized by 68Ga-DOTA-TATE PET/CT.

Diagnosis of ectopic adrenocorticotropic hormone (ACTH) and corticotropin-releasing hormone (CRH) co-secreting tumors causing Cushing syndrome (CS) is challenging because these tumors are rare and their diagnosis is frequently confused with Cushing disease (CD), caused by the effect of CRH on the pituitary. We report a case of a 21-year-old male patient who was referred to our institution with persistent hypercortisolemia and CS after undergoing unnecessary transsphenoidal surgery (TSS). ⁶⁸Ga-DOTA-TATE PET/CT revealed increased tracer uptake in the thymus, which was histologically proven to be a neuroendocrine tumor (NET) that stained positive for ACTH and CRH. Imaging with ¹⁸F-FDG PET/CT was not diagnostic.

Neurobehavioral Abnormalities in the HIV-1 Transgenic Rat Do Not Correspond to Neuronal Hypometabolism on 18F-FDG-PET

Motor and behavioral abnormalities are common presentations among individuals with HIV-1 associated neurocognitive disorders (HAND). We investigated whether longitudinal motor and behavioral performance in the HIV-1 transgenic rat (Tg), a commonly used neuro-HIV model, corresponded to in vivo neuronal death/dysfunction, by using rotarod and open field testing in parallel to [18F] 2-fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET). We demonstrated that age-matched non-Tg wild type (WT) rats outperformed the HIV-1 Tg rats at most time points on rotarod testing. Habituation to rotarod occurred at 8 weeks of age (fifth weekly testing session) in the WT rats but it never occurred in the Tg rats, suggesting deficits in motor learning. Similarly, in open field testing, WT rats outperformed the Tg rats at most time points, suggesting defective exploratory/motor behavior and increased emotionality in the Tg rat. Despite the neurobehavioral abnormalities, there were no concomitant deficits in 18F-FDG uptake in Tg rats on PET compared to age-matched WT rats and no significant longitudinal loss of FDG uptake in either group. The negative PET findings were confirmed using 14C- Deoxy-D-glucose autoradiography in 32 week-old Tg and WT rats. We believe that the neuropathology in the HIV-1 Tg rat is more likely a consequence of neuronal dysfunction rather than overt neurodegeneration/neuronal cell death, similar to what is seen in HIV-positive patients in the post-ART era.

Association between neuroendocrine tumors biomarkers and primary tumor site and disease type based on total 68Ga-DOTATATE-Avid tumor volume measurements

OBJECTIVE To determine the association between neuroendocrine tumor (NET) biomarker levels and the extent of disease as assessed by 68Ga DOTATATE PET/CT imaging. DESIGN A retrospective analysis of a prospective database of patients with NETs. METHODS Fasting plasma chromogranin A (CgA), neuron-specific enolase (NSE), gastrin, glucagon, vasoactive intestinal peptide (VIP) and pancreatic polypeptide (PP), and 24-h urinary 5-hydroxyindoleacetic acid (5-HIAA) levels were measured. Correlation between biomarkers and total 68Ga-DOTATATE-avid tumor volume (TV) was analyzed. RESULTS The analysis included 232 patients. In patients with pancreatic NETs (n = 112), 68Ga-DOTATATE TV correlated with CgA (r = 0.6, P = 0.001, Spearman). In patients with multiple endocrine neoplasia type 1 (n = 39), 68Ga-DOTATATE TV correlated with glucagon (r = 0.5, P = 0.01) and PP levels (r = 0.5, P = 0.049). In patients with von Hippel-Lindau (n = 24), plasma VIP (r = 0.5, P = 0.02) and PP levels (r = 0.7, P < 0.001) correlated with 68Ga-DOTATATE TV. In patients with small intestine NET (SINET, n = 74), 68Ga-DOTATATE TV correlated with CgA (r = 0.5, P = 0.02) and 5-HIAA levels (r = 0.7, P < 0.001), with 5-HIAA ≥8.1 mg/24 h associated with metastatic disease with high positive (81.8%) and negative (85.7%) predictive values (P = 0.001). 68Ga-DOTATATE TV in patients with NET of unknown primary (n = 16) and those with NET of other primary location (n = 30) correlated with 5-HIAA levels (r = 0.8, P = 0.002 and r = 0.7, P = 0.02 respectively). CONCLUSIONS Our data supports the use of specific NET biomarkers based on the site of the primary NET and the presence of hereditary syndrome-associated NET. High urinary 5-HIAA levels indicate the presence of metastatic disease in patients with SINET.

Schmorl Nodes Can Cause Increased 68Ga DOTATATE Activity on PET/CT, Mimicking Metastasis in Patients With Neuroendocrine Malignancy.

Schmorl node (SN) is the herniation of the nucleus pulposus through the cartilaginous and bony endplate into the adjacent vertebral body. It is documented that SNs produce areas of moderately increased F-FDG uptake. We present a case of a patient with history of neuroendocrine tumor, who underwent Ga DOTATATE PET/CT for follow-up, showing increased focal vertebral uptake suggestive of bone metastasis. CT revealed typical findings of an SN. The presented case indicates that SNs should be considered when encountering focally increased skeletal uptake in Ga DOTATATE PET/CT studies, which can mimic metastasis in patients with history of neuroendocrine tumors.

Optimized acoustic biochip integrated with microfluidics for biomarkers detection in molecular diagnostics

The development of integrated platforms incorporating an acoustic device as the detection element requires addressing simultaneously several challenges of technological and scientific nature. The present work was focused on the design of a microfluidic module, which, combined with a dual or array type Love wave acoustic chip could be applied to biomedical applications and molecular diagnostics. Based on a systematic study we optimized the mechanics of the flow cell attachment and the sealing material so that fluidic interfacing/encapsulation would impose minimal losses to the acoustic wave. We have also investigated combinations of operating frequencies with waveguide materials and thicknesses for maximum sensitivity during the detection of protein and DNA biomarkers. Within our investigations neutravidin was used as a model protein biomarker and unpurified PCR amplified Salmonella DNA as the model genetic target. Our results clearly indicate the need for experimental verification of the optimum engineering and analytical parameters, in order to develop commercially viable systems for integrated analysis. The good reproducibility of the signal together with the ability of the array biochip to detect multiple samples hold promise for the future use of the integrated system in a Lab-on-a-Chip platform for application to molecular diagnostics.