False Positive Regions Research Articles

A content-based late fusion approach applied to pedestrian detection

The diversity of pedestrians detectors proposed in recent years has encouraged some works to fuse them to achieve a more accurate detection. The intuition behind it is to combine the detectors based on its spatial consensus. The hypothesis is that a location pointed by multiple detectors has a high probability of actually belonging to a pedestrian, while false positive regions have little consensus among detectors (small support) which allows discarding the false positives in these regions. We proposed a novel method called Content-Based Spatial Consensus (CSBC), which, in addition to relying on spatial consensus, considers the content of the detection windows to learn a weighted-fusion of pedestrian detectors. The result is a reduction in false alarms and an enhancement in the detection. In this work, we also demonstrated that there is small influence of the feature used to learn the contents of the windows of each detector, which enables our method to be efficient even employing simple features. The CSBC overcomes state-of-the-art fusion methods in the ETH dataset and the Caltech dataset. Particularly, our method is also more efficient, since fewer detectors are necessary to achieve expressive results.

An Uncertainty-Driven GCN Refinement Strategy for Organ Segmentation

Organ segmentation in CT volumes is an important pre-processing step in many computer assisted intervention and diagnosis methods. In recent years, convolutional neural networks have dominated the state of the art in this task. However, since this problem presents a challenging environment due to high variability in the organ’s shape and similarity between tissues, the generation of false negative and false positive regions in the output segmentation is a common issue. Recent works have shown that the uncertainty analysis of the model can provide us with useful information about potential errors in the segmentation. In this context, we proposed a segmentation refinement method based on uncertainty analysis and graph convolutional networks. We employ the uncertainty levels of the convolutional network in a particular input volume to formulate a semi-supervised graph learning problem that is solved by training a graph convolutional network. To test our method we refine the initial output of a 2D U-Net. We validate our framework with the NIH pancreas dataset and the spleen dataset of the medical segmentation decathlon. We show that our method outperforms the state-of-the-art CRF refinement method by improving the dice score by 1% for the pancreas and 2% for spleen, with respect to the original U-Net’s prediction. Finally, we perform a sensitivity analysis on the parameters of our proposal and discuss the applicability to other CNN architectures, the results, and current limitations of the model for future work in this research direction. For reproducibility purposes, we make our code publicly available at https://github.com/rodsom22/gcn_refinement

3D Inception U-net with Asymmetric Loss for Cancer Detection in Automated Breast Ultrasound.

Breast cancer is the most common cancer and the leading cause of cancer-related deaths for women all over the world. Recently, automated breast ultrasound (ABUS) has become a new and promising screening modality for whole breast examination. However, reviewing volumetric ABUS is time-consuming and lesions could be missed during the examination. Therefore, computer-aided cancer detection in ABUS volume is extremely expected to help clinician for the breast cancer screening. We develop a novel end-to-end 3D convolutional network for automated cancer detection in ABUS volume, in order to accelerate reviewing and meanwhile to provide high detection sensitivity with low false positives (FPs). Specifically, an efficient 3D Inception Unet-style architecture with fusion deep supervision mechanism is proposed to attain decent detection performance. In addition, a novel asymmetric loss is designed to help the network balancing false positive and false negative regions, thus improving detection sensitivity for small cancerous lesions. The efficacy of our network was extensively validated on a dataset including 196 patients with 661 cancer regions. Our network obtained a detection sensitivity of 95.1% with 3.0 FPs per ABUS volume. Furthermore, the average inference time of the network was 0.1 second per volume, which largely shortens the conventional reviewing time. The proposed network provides efficient and accurate cancer detection scheme using ABUS volume, and may assist clinicians for more efficient breast cancer screening.

3D tumor detection in automated breast ultrasound using deep convolutional neural network.

Automated breast ultrasound (ABUS) has drawn attention in breast disease detection and diagnosis applications. Reviewing hundreds of slices produced by ABUS is time-consuming. In this paper, a tumor detection method for ABUS image based on convolutional neural network is proposed. First, integrating multitask learning with YOLOv3, an improved YOLOv3 detection network is designed to detect tumor candidate in two-dimensional (2D) slices. Two-dimensional detection separately treats each slice, leading to larger differences of position and score for tumor candidate in adjacent slices. Due to the influence of artifact, noise, and mammary tissues, 2D detection may include many false positive regions. To alleviate these problems, a rescoring processing algorithm is first designed. Then three-dimensional volume forming and FP reduction scheme are built. This method was tested on 340 volumes (124 patients, 181 tumors) with fivefold cross validation. It achieved sensitivities of 90%, 85%, 80%, 75%, and 70% at 7.42, 3.31, 1.62, 1.23, and 0.88 false positives per volume. Compared with existing ABUS tumor detection methods, our method gets a promising result.

A novel CAD system to automatically detect cancerous lung nodules using wavelet transform and SVM

A novel cancerous nodules detection algorithm for computed tomography images (CT-images) is presented in this paper. CT-images are large size images with high resolution. In some cases, number of cancerous lung nodule lesions may missed by the radiologist due to fatigue. A CAD system that is proposed in this paper can help the radiologist in detecting cancerous nodules in CT- images. The proposed algorithm is divided to four stages. In the first stage, an enhancement algorithm is implement to highlight the suspicious regions. Then in the second stage, the region of interest will be detected. The adaptive SVM and wavelet transform techniques are used to reduce the detected false positive regions. This algorithm is evaluated using 60 cases (normal and cancerous cases), and it shows a high sensitivity in detecting the cancerous lung nodules with TP ration 94.5% and with FP ratio 7 cluster/image.

Automated detection of cribriform growth patterns in prostate histology images

Cribriform growth patterns in prostate carcinoma are associated with poor prognosis. We aimed to introduce a deep learning method to detect such patterns automatically. To do so, convolutional neural network was trained to detect cribriform growth patterns on 128 prostate needle biopsies. Ensemble learning taking into account other tumor growth patterns during training was used to cope with heterogeneous and limited tumor tissue occurrences. ROC and FROC analyses were applied to assess network performance regarding detection of biopsies harboring cribriform growth pattern. The ROC analysis yielded a mean area under the curve up to 0.81. FROC analysis demonstrated a sensitivity of 0.9 for regions larger than {0.0150},hbox {mm}^{2} with on average 7.5 false positives. To benchmark method performance for intra-observer annotation variability, false positive and negative detections were re-evaluated by the pathologists. Pathologists considered 9% of the false positive regions as cribriform, and 11% as possibly cribriform; 44% of the false negative regions were not annotated as cribriform. As a final experiment, the network was also applied on a dataset of 60 biopsy regions annotated by 23 pathologists. With the cut-off reaching highest sensitivity, all images annotated as cribriform by at least 7/23 of the pathologists, were all detected as cribriform by the network and 9/60 of the images were detected as cribriform whereas no pathologist labelled them as such. In conclusion, the proposed deep learning method has high sensitivity for detecting cribriform growth patterns at the expense of a limited number of false positives. It can detect cribriform regions that are labelled as such by at least a minority of pathologists. Therefore, it could assist clinical decision making by suggesting suspicious regions.

Multicontext multitask learning networks for mass detection in mammogram.

In this paper, for the purpose of accurate and efficient mass detection, we propose a new deep learning framework, including two major stages: Suspicious region localization (SRL) and Multicontext Multitask Learning (MCMTL). In the first stage, SRL focuses on finding suspicious regions [regions of interest (ROIs)] and extracting multisize patches of these suspicious regions. A set of bounding boxes with different size is used to extract multisize patches, which aim to capture diverse context information. In the second stage, MCMTL networks integrate features from multisize patches of suspicious regions for classification and segmentation simultaneously, where the purpose of this stage is to keep the true positive suspicious regions and to reduce the false positive suspicious regions. According to the experimental results on two public datasets (i.e., CBIS-DDSM and INBreast), our method achieves the overall performance of 0.812 TPR@2.53 FPI and 0.919 TPR@0.12 FPI on test sets, respectively. Our proposed method suggests comparable performance to the state-of-the-art methods.

Refining background subtraction using consistent motion detection in adverse weather

Most background subtraction algorithms developed to detect moving objects are potentially problematic in that they experience performance degradation when weather conditions are adverse. We solve this problem by proposing a refinement method using a consistent motion detection method, the performance of which is robust to weather related changes in video images captured by a static camera. The proposed algorithm reduces the number of false-positive regions and fills parts that are missing as a result of the nature of the background subtraction methods. We show the extent of the improvement afforded by our algorithm in the handling of moving object detection in adverse weather conditions.

Detected text‐based image retrieval approach for textual images

This work addresses the problem of searching and retrieving similar textual images based on the detected text and opens the new directions for textual image retrieval. For image retrieval, several methods have been proposed to extract visual features and social tags; however, to extract embedded and scene text within images and use that text as automatic keywords/tags is still a young research field for text-based and content-based image retrieval applications. The automatic text detection retrieval is an emerging technology for robotics and artificial intelligence. In this study, the authors have proposed a novel approach to detect the text in an image and exploit it as keywords and tags for automatic text-based image retrieval. First, text regions are detected using maximally stable extremal region algorithm. Second, unwanted false positive text regions are eliminated based on geometric properties and stroke width transform. Next, the true text regions are proceeded into optical character recognition for recognition. Third, keywords are formed using a neural probabilistic language model. Finally, the textual images are indexed and retrieved based on the detected keywords. The experimental results on two benchmark datasets show the dominancy of text is efficient and valuable for image retrieval specifically for textual images.

A combined linear and nonlinear ultrasound time-domain approach for impact damage detection in composite structures using a constructive nonlinear array technique

Discovery and evaluation concerns of barely visible impact damage in composite materials is a well-known issue in industries using these materials. This work proposes a frequency sweep method where damage assessment is conducted with respect to the time domain. Firstly, a combined linear and nonlinear ultrasound imaging technique is proposed, which focuses on the excitation of damage/defect regions using a frequency sweep methodology from multiple transducer locations. Secondly, the method deconstructs time domain signals, which allows for the visualisation of linear and nonlinear ultrasound components independently. While, a filtering and frequency band separation method was used to exploit defect responses over different frequency ranges and provide time domain visualisation at the damage region. Finally, image segmentation was employed to automate the damage sizing procedure, while a binary imaging method was used to remove false positive damage regions produced by material vibration mode excitation (fundamental frequency responses) by using the nonlinear responses as a baseline-free tool.The results showed that the combined linear and nonlinear results provided more accurate results than a purely linear or nonlinear approach, furthermore the results were shown to be equivalent to those of a standard phased array system. The ability of the method to visualise nonlinear outputs in time can improve the understanding of nonlinear ultrasound mechanisms while provides a clear argument that a complete approach, incorporating both linear and nonlinear methods should be regarded as the future of NDT/E systems.

Lymphoma images analysis using morphological and non-morphological descriptors for classification

Mantle cell lymphoma, follicular lymphoma and chronic lymphocytic leukemia are the principle subtypes of the non-Hodgkin lymphomas. The diversity of clinical presentations and the histopathological features have made diagnosis of such lymphomas a complex task for specialists. Computer aided diagnosis systems employ computational vision and image processing techniques, which contribute toward the detection, diagnosis and prognosis of digitised images of histological samples. Studies aimed at improving the understanding of morphological and non-morphological features for classification of lymphoma remain a challenge in this area. This work presents a new approach for the classification of information extracted from morphological and non-morphological features of nuclei of lymphoma images. We extracted data of the RGB model of the image and employed contrast limited adaptive histogram equalisation and 2D order-statistics filter methods to enhance the contrast and remove noise. The regions of interest were identified by the global thresholding method. The flood-fill and watershed techniques were used to remove the small false positive regions. The area, extent, perimeter, convex area, solidity, eccentricity, equivalent diameter, minor axis and major axis measurements were computed for the regions detected in the nuclei. In the feature selection stage, we applied the ANOVA, Ansari-Bradley and Wilcoxon rank sum methods. Finally, we employed the polynomial, support vector machine, random forest and decision tree classifiers in order to evaluate the performance of the proposed approach. The non-morphological features achieved higher AUC and AC values for all cases: the obtained rates were between 95% and 100%. These results are relevant, especially when we consider the difficulties of clinical practice in distinguishing the studied groups. The proposed approach is useful as an automated protocol for the diagnosis of lymphoma histological tissue.

Applying a new computer-aided detection scheme generated imaging marker to predict short-term breast cancer risk

This study aims to investigate the feasibility of identifying a new quantitative imaging marker based on false-positives generated by a computer-aided detection (CAD) scheme to help predict short-term breast cancer risk. An image dataset including four view mammograms acquired from 1044 women was retrospectively assembled. All mammograms were originally interpreted as negative by radiologists. In the next subsequent mammography screening, 402 women were diagnosed with breast cancer and 642 remained negative. An existing CAD scheme was applied ‘as is’ to process each image. From CAD-generated results, four detection features including the total number of (1) initial detection seeds and (2) the final detected false-positive regions, (3) average and (4) sum of detection scores, were computed from each image. Then, by combining the features computed from two bilateral images of left and right breasts from either craniocaudal or mediolateral oblique view, two logistic regression models were trained and tested using a leave-one-case-out cross-validation method to predict the likelihood of each testing case being positive in the next subsequent screening. The new prediction model yielded the maximum prediction accuracy with an area under a ROC curve of AUC = 0.65 ± 0.017 and the maximum adjusted odds ratio of 4.49 with a 95% confidence interval of (2.95, 6.83). The results also showed an increasing trend in the adjusted odds ratio and risk prediction scores (p < 0.01). Thus, this study demonstrated that CAD-generated false-positives might include valuable information, which needs to be further explored for identifying and/or developing more effective imaging markers for predicting short-term breast cancer risk.

Computer-aided diagnosis of prostate cancer using multi-parametric MRI: comparison between PUN and Tofts models

Computer-aided diagnosis (CAD) systems are increasingly being used in clinical settings to report multi-parametric magnetic resonance imaging (mp-MRI) of the prostate. Usually, CAD systems automatically highlight cancer-suspicious regions to the radiologist, reducing reader variability and interpretation errors. Nevertheless, implementing this software requires the selection of which mp-MRI parameters can best discriminate between malignant and non-malignant regions. To exploit functional information, some parameters are derived from dynamic contrast-enhanced (DCE) acquisitions. In particular, much CAD software employs pharmacokinetic features, such as Ktrans and kep, derived from the Tofts model, to estimate a likelihood map of malignancy. However, non-pharmacokinetic models can be also used to describe DCE-MRI curves, without any requirement for prior knowledge or measurement of the arterial input function, which could potentially lead to large errors in parameter estimation. In this work, we implemented an empirical function derived from the phenomenological universalities (PUN) class to fit DCE-MRI. The parameters of the PUN model are used in combination with T2-weighted and diffusion-weighted acquisitions to feed a support vector machine classifier to produce a voxel-wise malignancy likelihood map of the prostate. The results were all compared to those for a CAD system based on Tofts pharmacokinetic features to describe DCE-MRI curves, using different quality aspects of image segmentation, while also evaluating the number and size of false positive (FP) candidate regions. This study included 61 patients with 70 biopsy-proven prostate cancers (PCa). The metrics used to evaluate segmentation quality between the two CAD systems were not statistically different, although the PUN-based CAD reported a lower number of FP, with reduced size compared to the Tofts-based CAD. In conclusion, the CAD software based on PUN parameters is a feasible means with which to detect PCa, without affecting segmentation quality, and hence it could be successfully applied in clinical settings, improving the automated diagnosis process and reducing computational complexity.

Diagnostic value of whole-body MRI with diffusion-weighted sequence for detection of peritoneal metastases in colorectal malignancy.

Objective:To assess the diagnostic accuracy of whole-body MRI using diffusion-weighted sequence (WB-DWI) to determine the peritoneal cancer index (PCI) in correlation with surgical and histopathological findings.Methods:Twenty-seven patients underwent preoperative WB-MRI, followed by cytoreductive surgery for primary tumors of the appendix (n = 15), colorectum (n = 12), and associated peritoneal disease. A total of 351 regions were retrospectively reviewed. The sensitivity, specificity, and accuracy were calculated at 13 anatomical sites. The WB-DWI PCI and PCI type were compared with surgical and histopathological findings. Results:No statistical difference was found between the WB-DWI PCI and surgical PCI (P = 0.574). WB-DWI correctly predicted the PCI type in 24 of 27 patients with high accuracy (88.9%), including 10 of 10 patients with small-volume tumor, 12 of 14 with moderate-volume tumor, and 2 of 3 with large-volume tumor. WB-DWI correctly depicted tumors in 163 of 203 regions, with 40 false-negative and 23 false-positive regions. The overall sensitivity, specificity, and accuracy of WB-DWI for the detection of peritoneal tumors were 80.3%, 84.5%, and 82.1%, respectively. For lesions < 0.5 cm in diameter, WB-DWI demonstrated good sensitivity (69.4%). Conclusions:WB-DWI accurately predicted PCI before surgery in patients undergoing evaluation for cytoreductive surgery.

Using wavelet sub-band and fuzzy 2-partition entropy to segment chronic lymphocytic leukemia images

Histological images analysis is an important procedure to diagnose different types of cancer. One of them is the chronic lymphocytic leukemia (CLL), which can be identified by applying image segmentation techniques. This study presents an unsupervised method to segment neoplastic nuclei in CLL images. Firstly, deconvolution, histogram equalization and mean filter were applied to enhance nuclear regions. Then, a segmentation technique based on a combination of wavelet transform, fuzzy 2-partition entropy and genetic algorithm was used, followed by removal of false positive regions, and application of valley-emphasis and morphological operations. In order to evaluate the proposed algorithm H&E-stained histological images were used. In the accuracy metric, the proposed method attained more than 80%, which can surpass similar methods. This proposal presents spatial distribution that has a good consistency with a manual segmentation and lower overlapping rate than other techniques in the literature.

영상 클러스터링과 HSV 컬러 모델을 이용한 차선 검출 전처리 기법

Among the technologies for implementing autonomous vehicles, advanced driver assistance system is a key technology to support driver's safe driving. In the technology using the vision sensor having a high utility, various preprocessing methods are used prior to feature extraction for lane detection. However, in the existing methods, the unnecessary lane candidates such as cars, lawns, and road separator in the road area are false positive. In addition, there are cases where the lane candidate itself can not be extracted in the area under the overpass, the lane within the dark shadow, the center lane of yellow, and weak lane. In this paper, we propose an efficient preprocessing method using k-means clustering for image division and the HSV color model. When the proposed preprocessing method is applied, the true positive region is maximally maintained during the lane detection and many false positive regions are removed.

Variant-aware saturating mutagenesis using multiple Cas9 nucleases identifies regulatory elements at trait-associated loci.

Cas9-mediated, high-throughput, saturating in situ mutagenesis permits fine-mapping of function across genomic segments. Disease- and trait-associated variants from genome-wide association studies largely cluster in regulatory DNA. Here we demonstrate the use of multiple designer nucleases and variant-aware library design to interrogate trait-associated regulatory DNA at high resolution. We developed a computational tool for the creation of saturating mutagenesis libraries with single or combinatorial nucleases with incorporation of variants. We applied this methodology to the HBS1L-MYB intergenic region, a locus associated with red blood cell traits, including fetal hemoglobin levels. This approach identified putative regulatory elements that control MYB expression. Analysis of genomic copy number highlighted potential false positive regions, which emphasizes the importance of off-target analysis in design of saturating mutagenesis experiments. Taken together, these data establish a widely applicable high-throughput and high-resolution methodology to reliably identify minimal functional sequences within large regions of disease- and trait-associated DNA.

An Adaptive CAD System to Detect Microcalcification in Compressed Mammogram Images

Microcalcifications (MC) in mammogram images are an early sign for breast cancer and their early detection is vital to improve its prognosis. Since MC appears as small dot in the mammogram image with size less than 1 mm and maybe easily overlooked by the radiologist, the Computer Aided Diagnosis (CAD) approach can assist the radiologist to improve their diagnostic accuracy. On the other hand, the mammogram images are a high resolution image with large image size which makes difficult the image transfer through the media. Therefore, in this paper, two image compressions techniques which are Discrete Cosine Transform (DCT) with entropy coding and Singular Value Decomposition (SVD) were investigated to reduce the mammogram image size. Then a novel adaptive CAD system is used to test the quality of the processed image based on true positive (TP) ratio and number of detected false positive (FP) regions in the mammogram image. The proposed adaptive CAD system used the visual appearance of MC in the mammogram to detect a potential MC regions. Then five texture features are implemented to reduce number of detected FP regions in the mammogram images. After implementing the adaptive CAD system on 100 mammogram images from USF and MIAS databases, it was found that the DCT can reduce the image size with a high quality since the ratio of TP is 87.6% with 11 FP/regions while in SVD the TP ratio is 79.1% with 26 FP/regions.

Automatic segmentation of airway tree based on local intensity filter and machine learning technique in 3D chest CT volume.

Airway segmentation plays an important role in analyzing chest computed tomography (CT) volumes for computerized lung cancer detection, emphysema diagnosis and pre- and intra-operative bronchoscope navigation. However, obtaining a complete 3D airway tree structure from a CT volume is quite a challenging task. Several researchers have proposed automated airway segmentation algorithms basically based on region growing and machine learning techniques. However, these methods fail to detect the peripheral bronchial branches, which results in a large amount of leakage. This paper presents a novel approach for more accurate extraction of the complex airway tree. This proposed segmentation method is composed of three steps. First, Hessian analysis is utilized to enhance the tube-like structure in CT volumes; then, an adaptive multiscale cavity enhancement filter is employed to detect the cavity-like structure with different radii. In the second step, support vector machine learning will be utilized to remove the false positive (FP) regions from the result obtained in the previous step. Finally, the graph-cut algorithm is used to refine the candidate voxels to form an integrated airway tree. A test dataset including 50 standard-dose chest CT volumes was used for evaluating our proposed method. The average extraction rate was about 79.1% with the significantly decreased FP rate. A new method of airway segmentation based on local intensity structure and machine learning technique was developed. The method was shown to be feasible for airway segmentation in a computer-aided diagnosis system for a lung and bronchoscope guidance system.

An automated framework for 3D serous pigment epithelium detachment segmentation in SD-OCT images.

Pigment epithelium detachment (PED) is an important clinical manifestation of multiple chorioretinal diseases, which can cause loss of central vision. In this paper, an automated framework is proposed to segment serous PED in SD-OCT images. The proposed framework consists of four main steps: first, a multi-scale graph search method is applied to segment abnormal retinal layers; second, an effective AdaBoost method is applied to refine the initial segmented regions based on 62 extracted features; third, a shape-constrained graph cut method is applied to segment serous PED, in which the foreground and background seeds are obtained automatically; finally, an adaptive structure elements based morphology method is applied to remove false positive segmented regions. The proposed framework was tested on 25 SD-OCT volumes from 25 patients diagnosed with serous PED. The average true positive volume fraction (TPVF), false positive volume fraction (FPVF), dice similarity coefficient (DSC) and positive predictive value (PPV) are 90.08%, 0.22%, 91.20% and 92.62%, respectively. The proposed framework can provide clinicians with accurate quantitative information, including shape, size and position of the PED region, which can assist clinical diagnosis and treatment.