Suitable Region Of Interest Research Articles

Altered diffusion tensor imaging (DTI) parameters in brain parenchyma after traumatic brain injury

Aim of studyTo evaluate the role of diffuse tensor imaging (DTI) in detecting structural alterations following TBI with persistent symptoms without findings on conventional structural imaging.MethodsThis was a prospective study that enrolled a total of 30 individuals (12 females and 18 males), 6 control with no history of trauma and 24 with previous history of trauma. All subjects underwent conventional brain MRI and DTI. The eigenvalues of the genue and splenium of corpus callosum (CC), superior longitudinal fasciculus (SLF), inferior longitudinal fasciculus (ILF), corticospinal tract (CS), arcuate fasciculus (AF) were measured bilaterally in all cases using suitable region of interest and the fraction anisotropy (FA), mean diffusivity (MD), were calculated. FA and MD values were compared between patients and control subjects. Diagnostic accuracy was calculated for FA, MD and combined. Sensitivity and Specificity were determined from the curve and PPV, NPV and accuracy underwent calculation through cross-tabulation.ResultAs compared with controls, FA was significantly decreased and MD was significantly increased in genue and splenium of CC and CS. For the Combined FA & MD in differentiating cases with trauma there was statistically significant difference in SLF, AF, genue, splenium, ILF and CS as p value < 0.05.ConclusionsDTI parameters is a promising tool in the detection of subtle brain parenchymal changes in patients with TBI who have persistent symptoms and no evident abnormalities on conventional MRI. Furthermore combined use of FA and MD shows better diagnostic accuracy in differentiation between control and trauma patient.

Deep learning-based dental implant recognition using synthetic X-ray images.

A novel algorithm for generating artificial training samples from triangulated three-dimensional (3D) surface models within the context of dental implant recognition is proposed. The proposed algorithm is based on the calculation of two-dimensional (2D) projections (from a number of different angles) of 3D volumetric representations of computer-aided design (CAD) surface models. A fully convolutional network (FCN) is subsequently trained on the artificially generated X-ray images for the purpose of automatically identifying the connection type associated with a specific dental implant in an actual X-ray image. Semi-automated and fully automated systems are proposed for segmenting questioned dental implants from the background in actual X-ray images. Within the context of the semi-automated system, suitable regions of interest (ROIs), which contain the dental implants, are manually specified. However, as part of the fully automated system, suitable ROIs are automatically detected. It is demonstrated that a segmentation/detection accuracy of 94.0% and a classification/recognition accuracy of 71.7% are attainable within the context of the proposed fully automated system. Since the proposed systems utilise an ensemble of techniques that has not been employed for the purpose of dental implant classification/recognition on any previous occasion, the above-mentioned results are very encouraging.Graphical abstract

Object Detection in High Resolution Remote Sensing Imagery Based on Convolutional Neural Networks With Suitable Object Scale Features

Object detection in high spatial resolution remote sensing images (HSRIs) is an important part of image information automatic extraction, analysis, and understanding. The region of interest (ROI) scale of object detection and the object feature representation are two vital factors in HSRI object detection. With respect to these two issues, this article presents a novel HSRI object detection method based on convolutional neural networks (CNNs) with suitable object scale features. First, the suitable ROI scale of object detection is obtained by compiling statistics for the scale range of objects in HSRIs. Then, a CNN framework for object detection in HSRIs is designed using a suitable ROI scale of object detection. The object features obtained using a CNN have good universality and robustness. Finally, a CNN framework with a suitable ROI scale of object detection is trained and tested. Using the WHU-RSONE data set, the proposed method is compared with the faster region-based CNN (Faster-RCNN) framework. The experimental results show that the proposed method outperforms the Faster-RCNN framework and provides good object detection results in HSRIs.

Improvement of region of interest extraction and scanning method of computer-aided diagnosis system for osteoporosis using panoramic radiographs.

Patients undergoing osteoporosis treatment benefit greatly from early detection. We previously developed a computer-aided diagnosis (CAD) system to identify osteoporosis using panoramic radiographs. However, the region of interest (ROI) was relatively small, and the method to select suitable ROIs was labor-intensive. This study aimed to expand the ROI and perform semi-automatized extraction of ROIs. The diagnostic performance and operating time were also assessed. We used panoramic radiographs and skeletal bone mineral density data of 200 postmenopausal women. Using the reference point that we defined by averaging 100 panoramic images as the lower mandibular border under the mental foramen, a 400 × 100-pixel ROI was automatically extracted and divided into four 100 × 100-pixel blocks. Valid blocks were analyzed using program 1, which examined each block separately, and program 2, which divided the blocks into smaller segments and performed scans/analyses across blocks. Diagnostic performance was evaluated using another set of 100 panoramic images. Most ROIs (97.0%) were correctly extracted. The operation time decreased to 51.4% for program 1 and to 69.3% for program 2. The sensitivity, specificity, and accuracy for identifying osteoporosis were 84.0, 68.0, and 72.0% for program 1 and 92.0, 62.7, and 70.0% for program 2, respectively. Compared with the previous conventional system, program 2 recorded a slightly higher sensitivity, although it occasionally also elicited false positives. Patients at risk for osteoporosis can be identified more rapidly using this new CAD system, which may contribute to earlier detection and intervention and improved medical care.

Influence of region-of-interest designs on quantitative measurement of multimodal imaging of MR non-enhancing gliomas.

A number of studies have revealed the usefulness of multimodal imaging in gliomas. Although the results have been heavily affected by the method used for region of interest (ROI) design, the most discriminatory method for setting the ROI remains unclear. The aim of the present study was to determine the most suitable ROI design for 18F-fluorodeoxyglucose (FDG) and 11C-methionine (MET) positron emission tomography (PET), apparent diffusion coefficient (ADC), and fractional anisotropy (FA) obtained by diffusion tensor imaging (DTI) from the viewpoint of grades of non-enhancing gliomas. A total of 31 consecutive patients with newly diagnosed, histologically confirmed magnetic resonance (MR) non-enhancing gliomas who underwent FDG-PET, MET-PET and DTI were retrospectively investigated. Quantitative measurements were performed using four different ROIs; hotspot/tumor center and whole tumor, constructed in either two-dimensional (2D) or three-dimensional (3D). Histopathological grading of the tumor was considered as empirical truth and the quantitative measurements obtained from each ROI was correlated with the grade of the tumor. The most discriminating ROI for non-enhancing glioma grading was different according to the different imaging modalities. 2D-hotspot/center ROI was most discriminating for FDG-PET (P=0.087), ADC map (P=0.0083), and FA map (P=0.25), whereas 3D-whole tumor ROI was best for MET-PET (P=0.0050). In the majority of scenarios, 2D-ROIs performed better than 3D-ROIs. Results from the image analysis using FDG-PET, MET-PET, ADC and FA may be affected by ROI design and the most discriminating ROI for non-enhancing glioma grading was different according to the imaging modality.

Non-contact remote estimation of cardiovascular parameters

Abstract Cardiovascular disease is a serious threat to human health. It is crucial to monitor the cardiovascular parameters reliably and conveniently. Non-contact measurement has been widely studied. However, there are some inevitable factors that limit the use of the platform and even lead to inaccuracy estimation. Hence, a novel non-contact method that estimates cardiovascular parameters under ambient light is proposed. The most suitable region of interest (ROI) is determined by a colormap, which is a map consists of Fast Fourier Transform (FFT) peak amplitude of every pixel. Comparisons suggest that the region includes cheeks and nose is the most appropriate, followed by the forehead. To remove the motion artifacts caused by body movement, face detection and tracking algorithms are performed. Further, a multi-step signal process that combines independent component analysis (ICA) and wavelet de-noising is utilized to extract clean signal from the noise-corrupted raw waveform. Moreover, various distance between subject and camera and the change of ambient light intensity are considered, where statistics results reveal that this non-contact methodology is blind to these factors. Compared with the gold standard pulse oximeter, the proposed method shows a high accuracy even in the presence of motion artifacts.

Comparison and Optimization of 3.0 T Breast Images Quality of Diffusion-Weighted Imaging with Multiple B-Values

Breast 3.0 T magnetic resonance diffusion-weighted imaging (MR-DWI) of benign and malignant lesions were obtained to measure and calculate the signal-to-noise ratio (SNR), signal intensity ratio (SIR), and contrast-to-noise ratio (CNR) of lesions at different b-values. The variation patterns of SNR and SIR were analyzed with different b-values and the images of DWI were compared at four different b-values with higher image quality. The effect of SIR on the differential diagnostic efficiency of benign and malignant lesions was compared using receiver operating characteristic curves to provide a reference for selecting the optimal b-value. A total of 96 qualified patients with 112 lesions and 14 patients with their contralateral 14 normal breasts were included in this study. The single-shot echo planar imaging sequence was used to perform the DWI and a total of 13 b-values were used: 0, 50, 100, 200, 400, 600, 800, 1000, 1200, 1500, 1800, 2000, and 2500 s/mm2. On DWI, the suitable regions of interest were selected. The SNRs of normal breasts (SNRnormal), SNRlesions, SIR, and CNR of benign and malignant lesions were measured on DWI with different b-values and calculated. The variation patterns of SNR, SIR, and CNR values on DWI for normal breasts, benign lesions, and malignant lesions with different b-values were analyzed by using Pearson correlation analysis. The SNR and SIR of benign and malignant lesions with the same b-values were compared using t-tests. The diagnostic efficiencies of SIR with different b-values for benign and malignant lesions were evaluated using receiver operating characteristic curves. Breast DWI had higher CNR for b-values ranging from 600 to 1200 s/mm2. It had the best CNR at b = 1000 s/mm2 for the benign lesions and at b = 1200 s/mm2 for the malignant lesions. The signal intensity and SNR values of normal breasts decreased with increasing b-values, with a negative correlation (r = -0.945, P < 0.01). The mean SNR values of benign and malignant lesions were negatively correlated (r = -0.982 and -0.947, respectively, and P < 0.01), gradually decreasing with increasing b-values. The mean SIR value of benign lesions gradually decreased with increasing b-values, a negative correlation (r = -0.991, P < 0.01). The mean SIR values of malignant lesions gradually increased with increasing b-values between 0 and 1200 s/mm2, and gradually decreased with increasing b-values ≥ 1500 s/mm2. For b-values of 600, 800, 1000, and 1200 s/mm2, the sensitivity and specificity of SIR in identifying benign and malignant lesions gradually increased with increasing b-values, peaking at 1200 s/mm2. Breast DWI had higher image quality for b-values ranging from 600 to 1200 s/mm2, and was best for b-values ranging from 1000 to 1200 s/mm2. The SIR had the highest diagnostic efficiency in differentiating benign and malignant lesions for a b-value of 1200 s/mm2.

A Generative Model for Concurrent Image Retrieval and ROI Segmentation

This paper proposes a probabilistic generative model that concurrently tackles the problems of image retrieval and region-of-interest (ROI) segmentation. Specifically, the proposed model takes into account several properties of the matching process between two objects in different images, namely: objects undergoing a geometric transformation, typical spatial location of the region of interest, and visual similarity. In this manner, our approach improves the reliability of detected true matches between any pair of images. Furthermore, by taking advantage of the links to the ROI provided by the true matches, the proposed method is able to perform a suitable ROI segmentation. Finally, the proposed method is able to work when there is more than one ROI in the query image. Our experiments on two challenging image retrieval datasets proved that our approach clearly outperforms the most prevalent approach for geometrically constrained matching and compares favorably to most of the state-of-the-art methods. Furthermore, the proposed technique concurrently provided very good segmentations of the ROI. Furthermore, the capability of the proposed method to take into account several objects-of-interest was also tested on three experiments: two of them concerning image segmentation and object detection in multi-object image retrieval tasks, and another concerning multiview image retrieval. These experiments proved the ability of our approach to handle scenarios in which more than one object of interest is present in the query.

Fractal dimension evaluation of cone beam computed tomography in patients with bisphosphonate-associated osteonecrosis

The aim of this study was to (1) evaluate the fractal dimension (FD) in regions of the mandible on cone beam CT (CBCT) images of patients with bisphosphonate-associated osteonecrosis of the jaws (BP-ONJ) and (2) to select the most suitable region of interest (ROI) for further study on detection of bone alterations associated with bisphosphonates. CBCT images of patients with BP-ONJ were included with matched controls. Values of FD were compared between groups. Selected ROIs were: ROI-1 - below the mandibular foramen; ROI-2 - above the mandibular foramen; ROI-3 - anterior to the mental foramen; ROI-4 - above the mandibular canal. The area of bone exposure was included as ROI-5. The results were analysed using generalized estimating equations and conditional logistic regression. There were 36 patients (67% female) with a mean age of 60.7 years. The mean FDs were: ROI-1 - 1.678 for controls and 1.673 for patients (P = 0.81); ROI-2 - 1.657 for controls and 1.653 for patients (P = 0.78); ROI-3 - 1.661 for controls and 1.684 for patients (P = 0.17); and ROI-4 - 1.670 for controls and 1.698 for patients (P = 0.03). The value of the FD in the area of exposed bone was the highest (1.729). The odds of being a BP-ONJ patient vs being a control was six times as high for individuals with a higher FD score at ROI-4, although the confidence interval was quite wide owing to the small sample size. In this preliminary study, BP-ONJ patients had higher FD values than controls at regions close to the alveolar process. The results suggest that FD is a promising tool for detection of bone alterations associated with BP-ONJ.

TH-A-220-01: An Automated Optimal ROI Discovery Method for Real-Time 3D Surface Image Monitoring of Head Motion

Purpose: To increase image registration accuracy, and to remove the time‐ consuming manual region‐of‐interest (ROI) selection process for patient monitoring 3D surface imaging systems (AlignRT,VisionRT,UK), a method was developed to adaptively define an optimal ROI based on discovering topographically stable surfaces over time. Methods: Our method calculates the displacement and velocity of 3D surface points between a series of images captured in real‐time and the reference image. Using a set of selection criteria, the algorithm adaptively constructs a ROI by including only facial surfaces that move rigidly with the target and excludes surfaces that fluctuate locally such as the mandible regions and eyes. Data for five volunteers were acquired over various facial surface and motion types. A total of 27 data sets were analyzed, with 2–7 data sets for each volunteer and each data set contained consecutively captured 40–120 real‐time images. Results: The adaptive ROI selection algorithm correspondingly detected surfaces affected by different types of facial movement independent of target motions. It interpreted the eyes, eyebrows and mandible motions shown in the real‐time 3D images as independent of target motion even when AlignRT misinterpreted them, and excluded them from the final ROI. For 3D images taken with substantial facial motions, either most of the facial areas were excluded or only small areas with random patterns were included in the final ROI. The algorithm excluded the upper side and nose where the view of the cameras was blocked, and thus surface was not well‐ reconstructed in many of real‐time 3D images when couch was rotated greater than 40°. Conclusion: The method successfully located suitable ROI surfaces undergoing minimal displacement and temporal fluctuation. By automating the ROI selection process, the time and complexity of current ROI definition can be reduced, together with user‐dependent registration errors.

TH-C-303A-05: Development of a Dynamic KV Collimator for Low Diagnostic Dose Real-Time 3D Motion Tracking During Radiation Therapy by Combined MV-KV Imaging

Purpose: Currently, real-time 3D MV-kV monitoring requires the use of continuous kV imaging throughout the treatment process leading to high diagnostic dose costs. For MV-kV tracking purposes the only needed kV image information are the projected images of the metallic fiducial markers. Generally these markers are small (3mm in length and 0.8mm in diameter), and for a standard 40cm × 30cm kV image comprise less than 1% of the total area. The proposed technique here uses a dynamic kV aperture to confine the kV exposure to a small region of interest (ROI) encompassing only the markers. As the internal markers move, the aperture is dynamically updated using feedback information provided by the last known marker positions. Method: A Varian Trilogy equipped with both an EPID and a kV imaging system was used. The kV collimator was mounted over the kV source and consists of four lead blades placed orthogonally on low friction linear guide rails. The position of each blade was controlled independently using a servomotor. MV-kV imaging was performed and software was used to calculate a suitable ROI that will be used as an input for the kV collimator. As the internal markers move, the aperture will be dynamically updated using the last known marker positions. Results: The combination of controller circuitry and chosen servomotors allows for blade travel speed of up to 11mm/s, depending on orientation, with an accuracy of 0.2mm in the collimator plane. This should be sufficient to keep the ROI properly centered on nearly any fiducial cluster. Conclusion: The technique proposed here would potentially lower the kV exposure by a factor of 50–500 depending on the speed, number, and spatial separation of the fiducials. The technology is directly applicable to any kV imaging system where only selective ROI information is required.

Radial Strain Gradient Across the Normal Myocardial Wall in Open-chest Pigs Measured with Doppler Strain Rate Imaging

One of the reasons for the large variation in radial strain measured with Doppler strain rate imaging in normal myocardium might be the different strain length (SL) used during analyses. The aim of this study was to evaluate the effect of different SL settings on strain recordings and the method's ability to detect transmural radial strain gradients. In 8 anesthetized pigs (mean weight 54 kg) epicardial echocardiography was performed. Strain analysis was carried out by defining the wall as a 1-, 2-, 3-, and 4-layer structure with suitable regions of interest. Peak ejection strain was measured with SL settings of 2 to 14 mm. The systolic (ejection) strain showed large variation with SL. Sampling in one layer gave no significant reduction in strain for increasing SL. The strain in the subepicardial layer was low and decreased when the wall was divided into several layers (15.9 +/- 4.8% [2 layers]-2.1 +/- 2.4% [4 layers]; both measurements with SL = 4 mm). The method could separate 4 layers with SL of 4 mm or less, 3 layers with SL of 6 mm or less, and 2 layers with SL of 8 mm or less. When measuring radial strain in the myocardial wall the SL must be low to evaluate transmural strain gradients. With correct settings of SL and region of interest, strain in 4 layers can be distinguished. As a rule of thumb the SL should be set to approximately half the systolic thickness of the wall or half the wall layer.

The value of local99mTc (Sn)-MDP bone to soft tissue uptake ratio in osteoporosis, before and during fluoride therapy

The reproducibility and diagnostic value of local bone to soft tissue uptake ratio of 99mTc(Sn)-MDP as a bone tracer was examined in a prospective study in 35 patients who were under investigation and/or treatment for postmenopausal osteoporosis. The ratio of tracer uptake in the second lumbar vertebra (L2) and both femoral shafts was calculated from the number of counts in suitable regions of interest. Results obtained with settings and calculations in the routine practice were compared to the results obtained by revision of all raw data in one run by one person. The results were compared to the serum alkaline phosphatase activity (AP) and to local bone mineral mass as determined by dual photon absorptiometry (DPA). In 15 patients serial measurements during fluoride therapy were also compared to serum osteocalcin values and to bone histomorphometric data. The precision error of the calculation of uptake ratios from raw counts (including selection of region of interest) was 13.9% for the femoral shaft and 14.7% for L2. The mean difference between left and right femoral shaft in individuals was not significant and its variance was small P greater than 0.1). There was a weak but significant linear correlation between local uptake ratio in the spine and AP in the total material (r = 0.328 P less than 0.01). However, changes in local uptake ratio during therapy with fluoride in 15 patients were too small to be of any value and did not correlate with changes in alkaline phosphatase or osteocalcin or trabecular surface covered with osteoblasts.(ABSTRACT TRUNCATED AT 250 WORDS)