False Positive Reduction Research Articles

3D deep learning for detecting pulmonary nodules in CT scans.

To demonstrate and test the validity of a novel deep-learning-based system for the automated detection of pulmonary nodules. The proposed system uses 2 3D deep learning models, 1 for each of the essential tasks of computer-aided nodule detection: candidate generation and false positive reduction. A total of 888 scans from the LIDC-IDRI dataset were used for training and evaluation. Results for candidate generation on the test data indicated a detection rate of 94.77% with 30.39 false positives per scan, while the test results for false positive reduction exhibited a sensitivity of 94.21% with 1.789 false positives per scan. The overall system detection rate on the test data was 89.29% with 1.789 false positives per scan. An extensive and rigorous validation was conducted to assess the performance of the proposed system. The system demonstrated a novel combination of 3D deep neural network architectures and demonstrates the use of deep learning for both candidate generation and false positive reduction to be evaluated with a substantial test dataset. The results strongly support the ability of deep learning pulmonary nodule detection systems to generalize to unseen data. The source code and trained model weights have been made available. A novel deep-neural-network-based pulmonary nodule detection system is demonstrated and validated. The results provide comparison of the proposed deep-learning-based system over other similar systems based on performance.

FEATHER: Automated Analysis of Force Spectroscopy Unbinding and Unfolding Data via a Bayesian Algorithm

Single-molecule force spectroscopy (SMFS) provides a powerful tool to explore the dynamics and energetics of individual proteins, protein-ligand interactions, and nucleic acid structures. In the canonical assay, a force probe is retracted at constant velocity to induce a mechanical unfolding/unbinding event. Next, two energy landscape parameters, the zero-force dissociation rate constant (ko) and the distance to the transition state (Δx‡), are deduced by analyzing the most probable rupture force as a function of the loading rate, the rate of change in force. Analyzing the shape of the rupture force distribution reveals additional biophysical information, such as the height of the energy barrier (ΔG‡). Accurately quantifying such distributions requires high-precision characterization of the unfolding events and significantly larger data sets. Yet, identifying events in SMFS data is often done in a manual or semiautomated manner and is obscured by the presence of noise. Here, we introduce, to our knowledge, a new algorithm, FEATHER (force extension analysis using a testable hypothesis for event recognition), to automatically identify the locations of unfolding/unbinding events in SMFS records and thereby deduce the corresponding rupture force and loading rate. FEATHER requires no knowledge of the system under study, does not bias data interpretation toward the dominant behavior of the data, and has two easy-to-interpret, user-defined parameters. Moreover, it is a linear algorithm, so it scales well for large data sets. When analyzing a data set from a polyprotein containing both mechanically labile and robust domains, FEATHER featured a 30-fold improvement in event location precision, an eightfold improvement in a measure of the accuracy of the loading rate and rupture force distributions, and a threefold reduction of false positives in comparison to two representative reference algorithms. We anticipate FEATHER being leveraged in more complex analysis schemes, such as the segmentation of complex force-extension curves for fitting to worm-like chain models and extended in future work to data sets containing both unfolding and refolding transitions.

Automated pulmonary nodule detection in CT images using deep convolutional neural networks

Lung cancer is one of the leading causes of cancer-related death worldwide. Early diagnosis can effectively reduce the mortality, and computer-aided diagnosis (CAD) as an important way to assist doctors has developed rapidly. In particular, automated pulmonary nodule detection in computed tomography (CT) images is crucial to CAD. It is a challenging task to quickly locate the exact positions of lung nodules. In this paper, a novel automated pulmonary nodule detection framework with 2D convolutional neural network (CNN) is proposed to assist the CT reading process. Firstly, we adjust the structure of Faster R-CNN with two region proposal networks and a deconvolutional layer to detect nodule candidates, and then three models are trained for three kinds of slices for later result fusion. Secondly, a boosting architecture based on 2D CNN is designed for false positive reduction, which is a classifier to distinguish true nodules from the candidates. The misclassified samples are still kept for retraining a model which boosts the sensitivity for pulmonary nodule detection. Finally, the results of these networks are fused to vote out the final classification results. Extensive experiments are conducted on LUNA16, and the sensitivity of nodule candidate detection achieves 86.42%. For the false positive reduction, the sensitivity reaches 73.4% and 74.4% at 1/8 and 1/4 FPs/scan, respectively. It illustrates that the proposed method can obviously achieve accurate pulmonary nodule detection.

Application of Bayes' Theorem in Valuating Depression Tests Performance.

The validity of clinical diagnoses is a fundamental topic in clinical psychology, because now there are some political administrations, as the IOM or the U.K. government, which are focusing on best evidence-based practice in clinical psychology. The most problematic issue in clinical psychology is to avoid wrong diagnoses which can have negative consequences on individual life and on the utility of clinical treatments. In the case of diagnoses based on self-report tests, the diagnostic decision about individual health is based on the comparison between its score and the cutoff, according to the frequentist approach to probability. However, the frequentist approach underestimates the possible risks of incorrect diagnoses based on cutoffs only. The Bayesian approach is a valid alternative to make diagnoses on the basis of the scores from psychological tests. The Bayes' theorem estimates the posterior probability of the presence of a pathology on the basis of the knowledge about the diffusion of this pathology (prior probability) and of the knowledge of sensitivity and specificity values of the test. With all this information, it is possible to estimate the diagnostic accuracy of some self-report tests used for assessing depression. We analyzed the diagnostic accuracy of the most used psychological tests of depression (Zung's Self-Rating Depression Scale, Hamilton Rating Scale for Depression, Center for Epidemiological Studies for Depression and the Beck Depression Inventory), together with a new scale (Teate Depression Inventory) developed with the IRT procedure, by analyzing the published works in which data about sensitivity and specificity of these scales are reported. Except the TDI, none of these scales can reach a satisfactory level of diagnostic accuracy, probably for the absence of an optimal procedure to select test items and subjects with clearly defined pathological symptoms which could allow the reduction of false positives in test scoring.

Of daemons and men: reducing false positive rate in intrusion detection systems with file system footprint analysis

In this work, we propose a methodology for reducing false alarms in file system intrusion detection systems, by taking into account the daemon’s file system footprint. More specifically, we experimentally show that sequences of outliers can serve as a distinguishing characteristic between true and false positives, and we show how analysing sequences of outliers can lead to lower false positive rates, while maintaining high detection rates. Based on this analysis, we developed an anomaly detection filter that learns outlier sequences using k-nearest neighbours with normalised longest common subsequence. Outlier sequences are then used as a filter to reduce false positives on the $$FI^2DS$$ file system intrusion detection system. This filter is evaluated on both overlapping and non-overlapping sequences of outliers. In both cases, experiments performed on three real-world web servers and a honeynet show that our approach achieves significant false positive reduction rates (up to 50 times), without any degradation of the corresponding true positive detection rates.

Semi-Supervised Multi-Task Learning for Lung Cancer Diagnosis.

Early detection of lung nodules is of great importance in lung cancer screening. Existing research recognizes the critical role played by CAD systems in early detection and diagnosis of lung nodules. However, many CAD systems, which are used as cancer detection tools, produce a lot of false positives (FP) and require a further FP reduction step. Furthermore, guidelines for early diagnosis and treatment of lung cancer are consist of different shape and volume measurements of abnormalities. Segmentation is at the heart of our understanding of nodules morphology making it a major area of interest within the field of computer aided diagnosis systems. This study set out to test the hypothesis that joint learning of false positive (FP) nodule reduction and nodule segmentation can improve the computer aided diagnosis (CAD) systems' performance on both tasks. To support this hypothesis we propose a 3D deep multi-task CNN to tackle these two problems jointly. We tested our system on LUNA16 dataset and achieved an average dice similarity coefficient (DSC) of 91% as segmentation accuracy and a score of nearly 92% for FP reduction. As a proof of our hypothesis, we showed improvements of segmentation and FP reduction tasks over two baselines. Our results support that joint training of these two tasks through a multi-task learning approach improves system performance on both. We also showed that a semi-supervised approach can be used to overcome the limitation of lack of labeled data for the 3D segmentation task.

Abstract 3417: Detection and analysis of oncovirus integration sites in FFPE-derived human tumor samples using hybrid capture and massively parallel sequencing

Abstract Viral genome integration into the host's genetic material can provide cells with selective advantages that make critical contributions to unregulated growth, cell division, and genomic instability. Current clinical tests include in situ hybridization, immunohistochemical, and real-time PCR assays that are unable to identify viral integration points in relation to oncongenic somatic mutations. This precludes a complete picture of how such mutations work together to promote, maintain, and disperse tumor progression. Precision cancer medicine employs genomic technologies (e.g., massively parallel sequencing) for high-throughput genomic profiling to molecularly define patient tumors, allowing identification of clinically actionable mutations. We created a custom probe set for targeted hybrid capture enrichment of several oncoviruses and have tested it in combination with our clinical tumor DNA profiling probe set, OncoPanel v3. Selected viruses were chosen due to their serving as causative cancer agents that disrupt tumor suppressors, facilitate genomic instability, and increase oncogene expression. The full viral genomes of hepatitis B virus and high-risk human papilloma virus (HPV) strains, 16, 18, 33, and 45 were targeted. Additionally, regions coding for E6 and E7 oncoproteins were targeted for several of the low-risk HPV strains, as was the LANA region of Kaposi's sarcoma-associated virus. FFPE-derived human tumor samples of known infection status were fragmented to 250 bp and converted into Illumina libraries. Pooled libraries underwent hybrid capture with custom, Agilent-designed OncoPanel and viral probes, and resulting captures were sequenced on an Illumina HiSeq2500 sequencer. SvABA (Structural variation and indel analysis by assembly) was used to perform de novo sequence assembly on soft-clipped and discordant aligned reads, and viral integration sites were identified by realigning contigs to the host genome to obtain breakpoint coordinates. Reduction of false positives was facilitated using dual-matched sample barcodes (IDT), which virtually eliminate barcode cross-talk, allowing for confident detection of low allele frequency events. Furthermore, inclusion of unique molecular identifiers (UMI) permitted discrimination of PCR duplicates and errors. Our results demonstrate high concordance with clinical samples of known infection status and provide viral integration break point locations in association with genomic tumor mutations. Viral integration sites are often associated with deletions or amplifications of the host's flanking genomic regions. Furthermore, viral integration may disrupt or enhance expression of tumor suppressors and oncogenes, respectively. Coupled with tumor molecular profiling, this information will better inform patient treatment. Citation Format: Michael K. Slevin, Winslow T. Powers, Robert T. Burns, Bruce M. Wollison, Haley A. Coleman, Amanda L. Paskavitz, Anwesha Nag, Danielle K. Manning, Elizabeth Garcia, Matthew D. Ducar, Aaron R. Thorner, Laura E. MacConaill. Detection and analysis of oncovirus integration sites in FFPE-derived human tumor samples using hybrid capture and massively parallel sequencing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3417.

Removing contaminants from databases of draft genomes.

Metagenomic sequencing of patient samples is a very promising method for the diagnosis of human infections. Sequencing has the ability to capture all the DNA or RNA from pathogenic organisms in a human sample. However, complete and accurate characterization of the sequence, including identification of any pathogens, depends on the availability and quality of genomes for comparison. Thousands of genomes are now available, and as these numbers grow, the power of metagenomic sequencing for diagnosis should increase. However, recent studies have exposed the presence of contamination in published genomes, which when used for diagnosis increases the risk of falsely identifying the wrong pathogen. To address this problem, we have developed a bioinformatics system for eliminating contamination as well as low-complexity genomic sequences in the draft genomes of eukaryotic pathogens. We applied this software to identify and remove human, bacterial, archaeal, and viral sequences present in a comprehensive database of all sequenced eukaryotic pathogen genomes. We also removed low-complexity genomic sequences, another source of false positives. Using this pipeline, we have produced a database of “clean” eukaryotic pathogen genomes for use with bioinformatics classification and analysis tools. We demonstrate that when attempting to find eukaryotic pathogens in metagenomic samples, the new database provides better sensitivity than one using the original genomes while offering a dramatic reduction in false positives.

False Positive Reduction by an Annular Model as a Set of Few Features for Microcalcification Detection to Assist Early Diagnosis of Breast Cancer.

Early automatic breast cancer detection from mammograms is based on the extraction of lesions, known as microcalcifications (MCs). This paper proposes a new and simple system for microcalcification detection to assist in early breast cancer detection. This work uses the two most recognized public mammogram databases, MIAS and DDSM. We are introducing a MC detection method based on (1) Beucher gradient for detection of regions of interest (ROIs), (2) an annulus model for extraction of few and effective features from candidates to MCs, and (3) one classification stage with two different classifiers, k Nearest Neighbor (KNN) and Support Vector Machine (SVM). For dense mammograms in the MIAS database, the performance metrics achieved are sensitivity of 0.9835, false alarm rate of 0.0083, accuracy of 0.9835, and area under the ROC curve of 0.9980 with a KNN classifier. The proposed MC detection method, based on a KNN classifier, achieves, a sensitivity, false positive rate, accuracy and area under the ROC curve of 0.9813, 0.0224, 0.9795 and 0.9974 for the MIAS database; and 0.9035, 0.0439, 0.9298 and 0.9759 for the DDSM database. By slightly reducing the true positive rate the method achieves three instances with false positive rate of 0: 2 on fatty mammograms with KNN and SVM, and one on dense with SVM. The proposed method gives better results than those from state of the art literature, when the mammograms are classified in fatty, fatty-glandular, and dense.

Fusion of Random Projection, Multi-Resolution Features and Distance Weighted K Nearest Neighbor for Masses Detection in Mammographic Images

<p>Breast cancer is the top cancer in women both in the developed and the developing world. For early detection of the disease, mammography is still the most effective method beside ultrasound and magnetic resonance imaging. Computer Aided Detection systems have been developed to aid radiologists in diagnosing breast cancer. Different methods were proposed to overcome the main drawback of producing large number of False Positives. In this paper, we presented a novel method for masses detection in mammograms. To describe masses, multi-resolution features were utilized. In feature extraction step, we calculated multi-resolution Block Difference Inverse Probability features and multi-resolution statistical features. Once the descriptors were extracted, we deployed random projection and distance weighted K Nearest Neighbor to classify the detected masses. The result is quite sanguine with sensitivity, false positive reduction and time for carrying out the algorithm</p>

Development of an NGS assay to detect MRD in hematologic malignancies.

e24177 Background: There are significant prognostic and therapeutic benefits in monitoring minimal/measurable residual disease (MRD) in hematologic malignancies. However, monitoring of MRD requires the detection of minute cell populations with mutations present at very low allele frequencies (AFs), and is currently performed using low throughput flow cytometry or qPCR/ddPCR methodologies which are limited in their ability to monitor multiple mutations. An approach that can assay mutations in multiple genes may provide a broader understanding of MRD and patient prognosis. We have developed a Next Generation Sequencing (NGS) MRD assay using the Illumina NextSeq platform to simultaneously monitor a number of genes associated with hematologic malignancies (e.g. NPM1, RUNX1, FLT3). Here we present studies demonstrating detection of mutations at the low AFs required for the monitoring of MRD in hematologic malignancies. Methods: A custom target capture NGS assay was developed using Unique Molecular Identifier-based error correction enabling mutation detection of AFs down to 0.01%–0.1%. This assay incorporates barcoded duplex adapters and xGen Lockdown Probes (Integrated DNA Technologies) to enrich for relevant genes. Following sequencing on the NextSeq500, a custom bioinformatics pipeline was utilized for the error correction and false-positive reduction necessary to accurately detect mutations at the low AFs required for MRD detection. Results: Initial studies utilized Horizon Discovery’s TruQ Reference standards with mutations characterized at 1.0%–1.3% by ddPCR. These samples were diluted with a wildtype standard to demonstrate detection at even lower AFs. Subsequent studies were performed on peripheral blood samples from patients diagnosed with myelodysplastic syndromes and acute myeloid leukemia. Conclusions: This NGS MRD assay can accurately detect low AF mutations in both control and clinical MDS and AML samples. These results show that the NGS MRD assay described can be effectively utilized to monitor and study the course of disease in hematologic malignancies. Subsequent studies are underway to further increase the sensitivity of the assay and demonstrate its utility in a broad range of high-sensitivity applications.

Mitigating false alarms using accumulator rule and dynamic sliding window in wireless body area

Faulty measurements of sensors in critical applications like remote patient monitoring systems may sometimes lead to misdiagnose or unnecessary interventions from healthcare professionals. The main motivation of this paper is to mitigate false alarms thereby improving the accuracy of the system. In this paper, we propose an approach for detecting anomalous sensor measurements. This paper utilizes dynamic sliding window, Weighted Moving Average for prediction purposes and accumulator rule for improving the accuracy in identifying true medical conditions. Finally we validate the performance of the proposed approach using a publicly available dataset and has been compared with existing approaches using statistical metrics. We achieved 37.40% reduction in False Positive Rate when compared with existing approaches.

Particle swarm optimization method for small retinal vessels detection on multiresolution fundus images

Retinal vessel segmentation plays an important role in the diagnosis of eye diseases and is considered as one of the most challenging tasks in computer-aided diagnosis (CAD) systems. The main goal of this study was to propose a method for blood-vessel segmentation that could deal with the problem of detecting vessels of varying diameters in high- and low-resolution fundus images. We proposed to use the particle swarm optimization (PSO) algorithm to improve the multiscale line detection (MSLD) method. The PSO algorithm was applied to find the best arrangement of scales in the MSLD method and to handle the problem of multiscale response recombination. The performance of the proposed method was evaluated on two low-resolution (DRIVE and STARE) and one high-resolution fundus (HRF) image datasets. The data include healthy (H) and diabetic retinopathy (DR) cases. The proposed approach improved the sensitivity rate against the MSLD by 4.7% for the DRIVE dataset and by 1.8% for the STARE dataset. For the high-resolution dataset, the proposed approach achieved 87.09% sensitivity rate, whereas the MSLD method achieves 82.58% sensitivity rate at the same specificity level. When only the smallest vessels were considered, the proposed approach improved the sensitivity rate by 11.02% and by 4.42% for the healthy and the diabetic cases, respectively. Integrating the proposed method in a comprehensive CAD system for DR screening would allow the reduction of false positives due to missed small vessels, misclassified as red lesions.

Convolutional neural network-based PSO for lung nodule false positive reduction on CT images

Background and objectiveDetection of lung nodules is critical in CAD systems; this is because of their similar contrast with other structures and low density, which result in the generation of numerous false positives (FPs). Therefore, this study proposes a methodology to reduce the FP number using a deep learning technique in conjunction with an evolutionary technique. MethodThe particle swarm optimization (PSO) algorithm was used to optimize the network hyperparameters in the convolutional neural network (CNN) in order to enhance the network performance and eliminate the requirement of manual search. ResultsThe methodology was tested on computed tomography (CT) scans from the Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) with the highest accuracy of 97.62%, sensitivity of 92.20%, specificity of 98.64%, and area under the receiver operating characteristic (ROC) curve of 0.955. ConclusionThe results demonstrate the high performance-potential of the PSO algorithm in the identification of optimal CNN hyperparameters for lung nodule candidate classification into nodules and non-nodules, increasing the sensitivity rates in the FP reduction step of CAD systems.

A deep CNN based transfer learning method for false positive reduction

A low false positive (FP) rate is of great importance for the use of a Computer Aided Detection (CAD) system to detect pulmonary nodules in thoracic Computed Tomography (CT). However, due to the variations of nodules in appear and size, it is still a very challenging task to obtain a low FP rate. In this paper, we propose a deep Convolutional Neural Network (CNN) based transfer learning method for FP reduction in pulmonary nodule detection on CT slices. We utilized one of the state-of-the-art CNN models, VGG-16 [4], as a feature extractor to obtain nodule features, and used a support vector machine (SVM) for nodule classification. Firstly we transferred all the layers from a pre-trained VGG-16 model in ImageNet to our target networks. Then, we tuned the last fully connected layers to adjust the computer-vision-task-trained CNN model to pulmonary nodule classification task. The initial CNN filter weights were then optimized using the training data, i.e., the pulmonary nodule patch images and corresponding labels through back-propagation so that they better reflected the modalities in the pulmonary nodule image dataset. Finally, features learned in the fine-tuned CNN were used to train a SVM classifier. The output of the trained SVM was used for final classification. Experimental results show that the overall sensitivity of the proposed method was 87.2% with 0.39 FPs per scan, which is higher than 85.4% with 4 FPs per scan obtained by other state of art method.

Computer-Aided Detection of Mammographic Masses Using Hybrid Region Growing Controlled by Multilevel Thresholding

Masses are one of the common signs of nonpalpable breast cancer visible in mammograms. However, due to its irregular and obscured margin, variability in size, and occlusion within dense breast tissue, a mass may be missed during screening. In this paper, we propose a novel approach for automatic detection of mammographic masses using an iterative method of multilevel high-to-low intensity thresholding, followed by region growing and reduction of false positives, in which an image is considered as a 3D topographic map with intensity as the third dimension. At each iteration, first, the focal regions of masses are obtained by thresholding, and then potential sites of masses are extracted from the focal regions with a newly developed region growing technique. Finally, false positives are reduced using contrast and distance between two potential mass regions, and by using a classifier after the extraction of shape- and orientation-based features. The performance of the method is evaluated with 120 scanned-film images, including 55 images with 57 masses and 65 normal images from the mini-MIAS database; 555 scanned-film images, including 355 images with 370 masses and 200 normal images from the DDSM; and 219 digital radiography (DR) images, including 99 images with 120 masses and 120 normal images from a local database. For the mini-MIAS, DDSM, and DR images 90% sensitivity is achieved at a rate of 4.4, 0.99, and 1.0 false positive per images, respectively.

Prehospital Acute ST-Elevation Myocardial Infarction Identification in San Diego: A Retrospective Analysis of the Effect of a New Software Algorithm

BackgroundA significant increase in false positive ST-elevation myocardial infarction (STEMI) electrocardiogram interpretations was noted after replacement of all of the City of San Diego's 110 monitor-defibrillator units with a new brand. These concerns were brought to the manufacturer and a revised interpretive algorithm was implemented. ObjectivesThis study evaluated the effects of a revised interpretation algorithm to identify STEMI when used by San Diego paramedics. MethodsData were reviewed 6 months before and 6 months after the introduction of a revised interpretation algorithm. True-positive and false-positive interpretations were identified. Factors contributing to an incorrect interpretation were assessed and patient demographics were collected. ResultsA total of 372 (234 preimplementation, 138 postimplementation) cases met inclusion criteria. There was a significant reduction in false positive STEMI (150 preimplementation, 40 postimplementation; p < 0.001) after implementation. The most common factors resulting in false positive before implementation were right bundle branch block, left bundle branch block, and atrial fibrillation. The new algorithm corrected for these misinterpretations with most postimplementation false positives attributed to benign early repolarization and poor data quality. Subsequent follow-up at 10 months showed maintenance of the observed reduction in false positives. ConclusionsThis study shows that introducing a revised 12-lead interpretive algorithm resulted in a significant reduction in the number of false positive STEMI electrocardiogram interpretations in a large urban emergency medical services system. Rigorous testing and standardization of new interpretative software is recommended before introduction into a clinical setting to prevent issues resulting from inappropriate cardiac catheterization laboratory activations.

Effectiveness of a Novel Specimen Collection System in Reducing Blood Culture Contamination Rates

ProblemFalse-positive blood-culture results due to skin contamination of samples remain a persistent problem for health care providers. Our health system recognized that our rates of contamination across the 4 emergency department campuses were above the national average. MethodsA unique specimen collection system was implemented throughout the 4 emergency departments and became the mandatory way to collect adult blood cultures. The microbiology laboratory reported contamination rates weekly to manage potential problems; 7 months of data are presented here. ResultsThere was an 82.8% reduction in false positives with the unique specimen collection system compared with the standard method (chi-squared test with Yates correction, 2-tailed, P = 0.0001). Based on the historical 3.52% rate of blood-culture contamination for our health facilities, 2.92 false positives were prevented for every 100 blood cultures drawn, resulting from adoption of the unique specimen collection system as the standard of care. ConclusionThis unique collection system can reduce the risk of blood culture contamination significantly and is designed to augment, rather than replace, the standard phlebotomy protocol already in use in most health care settings.Unlabelled Box

Detection of white matter lesion regions in MRI using SLIC0 and convolutional neural network

Background and Objective: White matter lesions are non-static brain lesions that have a prevalence rate up to 98% in the elderly population. Because they may be associated with several brain diseases, it is important that they are detected as soon as possible. Magnetic Resonance Imaging (MRI) provides three-dimensional data with the possibility to detect and emphasize contrast differences in soft tissues, providing rich information about the human soft tissue anatomy. However, the amount of data provided for these images is far too much for manual analysis/interpretation, representing a difficult and time-consuming task for specialists. This work presents a computational methodology capable of detecting regions of white matter lesions of the brain in MRI of FLAIR modality. The techniques highlighted in this methodology are SLIC0 clustering for candidate segmentation and convolutional neural networks for candidate classification. Methods: The methodology proposed here consists of four steps: (1) images acquisition, (2) images preprocessing, (3) candidates segmentation and (4) candidates classification. Results: The methodology was applied on 91 magnetic resonance images provided by DASA, and achieved an accuracy of 98.73%, specificity of 98.77% and sensitivity of 78.79% with 0.005 of false positives, without any false positives reduction technique, in detection of white matter lesion regions. Conclusions: It is demonstrated the feasibility of the analysis of brain MRI using SLIC0 and convolutional neural network techniques to achieve success in detection of white matter lesions regions.

Deep learning aided decision support for pulmonary nodules diagnosing: a review.

Deep learning techniques have recently emerged as promising decision supporting approaches to automatically analyze medical images for different clinical diagnosing purposes. Diagnosing of pulmonary nodules by using computer-assisted diagnosing has received considerable theoretical, computational, and empirical research work, and considerable methods have been developed for detection and classification of pulmonary nodules on different formats of images including chest radiographs, computed tomography (CT), and positron emission tomography in the past five decades. The recent remarkable and significant progress in deep learning for pulmonary nodules achieved in both academia and the industry has demonstrated that deep learning techniques seem to be promising alternative decision support schemes to effectively tackle the central issues in pulmonary nodules diagnosing, including feature extraction, nodule detection, false-positive reduction, and benign-malignant classification for the huge volume of chest scan data. The main goal of this investigation is to provide a comprehensive state-of-the-art review of the deep learning aided decision support for pulmonary nodules diagnosing. As far as the authors know, this is the first time that a review is devoted exclusively to deep learning techniques for pulmonary nodules diagnosing.