Quantitative Evaluation Metrics Research Articles

Bidirectional Mapping Generative Adversarial Networks for Brain MR to PET Synthesis.

Fusing multi-modality medical images, such as magnetic resonance (MR) imaging and positron emission tomography (PET), can provide various anatomical and functional information about the human body. However, PET data is not always available for several reasons, such as high cost, radiation hazard, and other limitations. This paper proposes a 3D end-to-end synthesis network called Bidirectional Mapping Generative Adversarial Networks (BMGAN). Image contexts and latent vectors are effectively used for brain MR-to-PET synthesis. Specifically, a bidirectional mapping mechanism is designed to embed the semantic information of PET images into the high-dimensional latent space. Moreover, the 3D Dense-UNet generator architecture and the hybrid loss functions are further constructed to improve the visual quality of cross-modality synthetic images. The most appealing part is that the proposed method can synthesize perceptually realistic PET images while preserving the diverse brain structures of different subjects. Experimental results demonstrate that the performance of the proposed method outperforms other competitive methods in terms of quantitative measures, qualitative displays, and evaluation metrics for classification.

Weather Radar Image Superresolution Using a Nonlocal Residual Network

Accurate and high-resolution weather radar images reflecting detailed structure information of radar echo are vital for analysis and forecast of extreme weather. Typically, this is performed by using interpolation schemes, which only use several neighboring data values for computational approximation to get the estimated value regardless of the large-scale context feature of weather radar images. Inspired by the striking performance of the convolutional neural network (CNN) applied in feature extraction and nonlocal self-similarity of weather radar images, we proposed a nonlocal residual network (NLRN) on the basis of CNN. The proposed network mainly consists of several nonlocal residual blocks (NLRB), which combine short skip connection (SSC) and nonlocal operation to train the deep network and capture large-scale context information. In addition, long skip connection (LSC) added in the network avoids learning low-frequency information, making the network focus on high-level features. Extensive experiments of ×2 and ×4 super-resolution reconstruction demonstrate that NLRN achieves superior performance in terms of both quantitative evaluation metrics and visual quality, especially for the reconstruction of the edge and detailed information of the weather radar echo.

AUTOENCODER PERFORMANCE ANALYSIS OF SKIN LESION DETECTION

Melanoma is a rapidly pervasive and deathly type of skin cancer that is responsible for most deaths from this kind of disease. It can quickly prevail in other organs if not handled early. Fortunately, the symptoms of skin cancer become visible to the sick, which creates a chance to detect it at an early stage. Because people know so little about their specific symptoms and because of a shortage of expert doctors, automated skin cancer detection has become an important public health issue. Many computer-aided diagnostic methods have been suggested so far. Besides traditional techniques based on image processing, researchers have recently used deep learning successfully for many different purposes. Deep neural networks are widely used in segmentation, classification, detection, etc. In this paper, we check the applicability of deep learning approaches to the segmentation of skin lesions to detect lesion boundaries by evaluating five architectures: U-NET, RESU-NET, VGG16UNET, DENSENET121, and EfficientNet-B0 by presenting a comparative view of those approaches. The five architectures were trained on three different data sets: ISIC 2016, ISIC 2018, and PH2, each set consisting of skin lesion images and the ground truth for their segmentation, and then used pre-processed on three datasets. Quantitative evaluation metrics were used for evaluating the performance of the studied architectures. Among these five architectures, the DENSENET121 architecture showed the best precision rate in all training datasets. We obtained an above 95% precision score in the PH2 dataset. In addition, the pre-processing steps were beneficial for the results.

Deep learning enables confocal laser-scanning microscopy with enhanced resolution.

Theoretical resolution enhancement of confocal laser-scanning microscopy (CLSM) is sacrificed for the best compromise between optical sectioning and the signal-to-noise ratio (SNR). The pixel reassignment reconstruction algorithm can improve the effective spatial resolution of CLSM to its theoretical limit. However, current implementations are not versatile and are time-consuming or technically complex. Here we present a parameter-free post-processing strategy for laser-scanning microscopy based on deep learning, which enables a spatial resolution enhancement by a factor of ∼1.3, compared to conventional CLSM. To speed up the training process for experimental data, transfer learning, combined with a hybrid dataset consisting of simulated synthetic and experimental images, is employed. The overall resolution and SNR improvement, validated by quantitative evaluation metrics, allowed us to correctly infer the fine structures of real experimental images.

Noise reduction by adaptive-SIN filtering for retinal OCT images

Optical coherence tomography (OCT) images is widely used in ophthalmic examination, but their qualities are often affected by noises. Shearlet transform has shown its effectiveness in removing image noises because of its edge-preserving property and directional sensitivity. In the paper, we propose an adaptive denoising algorithm for OCT images. The OCT noise is closer to the Poisson distribution than the Gaussian distribution, and shearlet transform assumes additive white Gaussian noise. We hence propose a square-root transform to redistribute the OCT noise. Different manufacturers and differences between imaging objects may influence the observed noise characteristics, which make predefined thresholding scheme ineffective. We propose an adaptive 3D shearlet image filter with noise-redistribution (adaptive-SIN) scheme for OCT images. The proposed adaptive-SIN is evaluated on three benchmark datasets using quantitative evaluation metrics and subjective visual inspection. Compared with other algorithms, the proposed algorithm better removes noise in OCT images and better preserves image details, significantly outperforming in terms of both quantitative evaluation and visual inspection. The proposed algorithm effectively transforms the Poisson noise to Gaussian noise so that the subsequent shearlet transform could optimally remove the noise. The proposed adaptive thresholding scheme optimally adapts to various noise conditions and hence better remove the noise. The comparison experimental results on three benchmark datasets against 8 compared algorithms demonstrate the effectiveness of the proposed approach in removing OCT noise.

Feature Purity: A Quantitative Evaluation Metric for Feature Extraction in Convolutional Neural Networks

The quantitative analysis of specific convolutional layer feature information in Convolutional Neural Networks (CNN) has received considerable critical attention in the field of deep learning. In this paper, we proposed an innovative measuring method based on the principle of information entropy to quantitatively analyze the performance of feature extraction in CNN. In the method, the feature purity was defined as the normalized entropy of activation histogram, which was calculated from all kernels in feature layer of CNN. Feature purities were evaluated respectively in different layers of CNN as the inspection for internal structure in specific classification models. At the same time, general degrees of purities among different models were compared with classification performances. The experiments were performed in models of AlexNet, VGG, ResNet, and SENet with different epochs trained by cifar10 and ImageNet1000 datasets. Additionally, a visualization evaluation was performed by grad-CAM method in the manners of the intra-model and inter-model purities. Experimental results showed a strong relationship between feature purity and the classification accuracy in each model. Additionally, the visualized feature map also demonstrated that salient region extracted in specified feature layer had good consistency with purity value. As a result, the proposed feature purity can serve as a quantitative metric representing the degree of feature extraction in CNN without the dependency of label or specific network structure; and has significant interpretability and versatility across models.

Restoration and enhancement of breast ultrasound images using extended complex diffusion based unsharp masking.

Ultrasound is a well-known imaging modality for the interpretation of breast cancer. It is playing very important role for breast cancer detection that are missed by mammograms. The image acquisition is usually affected by the presence of noise, artifacts, and distortion. To overcome such type of issues, there is a need of image restoration and enhancement to improve the quality of image. This paper proposes a single framework for denoising and enhancement of ultrasound images, where a smoothing filter is replaced with an extended complex diffusion-based filter in an unsharp masking technique. The performance evaluation of the proposed method is tested on real ultrasound breast cancer images database and synthetic ultrasound image. The performance evaluation comprises qualitative and quantitative evaluation along with comparative analysis of pre-existing and proposed method. The quantitative evaluation metrics are mean squared error, peak-signal-to-noise ratio, correlation parameter, normalized absolute error, universal quality index, similarity structure index, edge preservation index, a measure of enhancement, a measure of enhancement by entropy, and second derivative like measurement. The result specifies that the proposed method is better suited approach for the removal of speckle noise which follows Rayleigh distribution, restoration of information, enhancement of abnormalities, and proper edge preservation.

NIRS-ICA: A MATLAB Toolbox for Independent Component Analysis Applied in fNIRS Studies.

Independent component analysis (ICA) is a multivariate approach that has been widely used in analyzing brain imaging data. In the field of functional near-infrared spectroscopy (fNIRS), its promising effectiveness has been shown in both removing noise and extracting neuronal activity-related sources. The application of ICA remains challenging due to its complexity in usage, and an easy-to-use toolbox dedicated to ICA processing is still lacking in the fNIRS community. In this study, we propose NIRS-ICA, an open-source MATLAB toolbox to ease the difficulty of ICA application for fNIRS studies. NIRS-ICA incorporates commonly used ICA algorithms for source separation, user-friendly GUI, and quantitative evaluation metrics assisting source selection, which facilitate both removing noise and extracting neuronal activity-related sources. The options used in the processing can also be reported easily, which promotes using ICA in a more reproducible way. The proposed toolbox is validated and demonstrated based on both simulative and real fNIRS datasets. We expect the release of the toolbox will extent the application for ICA in the fNIRS community.

Extraction of media adventitia and luminal intima borders by reconstructing intravascular ultrasound image sequences with vascular structural continuity.

Most published methods directly achieve vessel membrane border detection on cross-sectional intravascular ultrasound (IVUS) images. The vascular structural continuity that exists in entire IVUS image sequences has been overlooked. However, this continuity can have a helpful role in the delineation of vessel membrane contours. To achieve the vessel membrane segmentation more effectively through employing this continuity, a strategy, referred to as multiangle reconstruction, segmentation, and recovery (RSR), is proposed in this paper. Four main steps are contained in the multiangle-RSR: first, a combination of sampling and interpolation is employed to reconstruct long-axis-model IVUS frames, in which continuity information becomes available. Second, a clustering algorithm is conducted on long-axis-model IVUS frames to roughly extract the media-adventitia (MA) and lumen-intima (LI) boundaries. Third, the segmentation results of cross-sectional IVUS frames are recovered based on the rough results of long-axis-model IVUS frames, and an optimization process that combines downsampling, fitting and smoothing is designed to reduce the interference of bifurcation and side vessels. Multiangle-RSR is tested on a public dataset, and the Hausdorff distance (HD), Jaccard measure (JM), and percentage of area difference (PAD) are utilized as quantitative evaluation metrics. Mean HDs of 0.34 and 0.29mm are obtained for MA border detection and LI border detection, respectively, which decrease by 43.3% and 9.4%, respectively, compared with their counterparts in previously published approaches. Furthermore, the mean JM is 0.87 for both MA border detection and LI border detection. The mean PADs of the MA contour extraction and the LI contour extraction are 0.10 and 0.11, respectively. The results indicate that the proposed strategy effectively introduces vascular structural continuity by reconstructing long-axis-model IVUS frames and achieves more precise extraction of MA and LI borders.

MoDL-QSM: Model-based deep learning for quantitative susceptibility mapping

Quantitative susceptibility mapping (QSM) has demonstrated great potential in quantifying tissue susceptibility in various brain diseases. However, the intrinsic ill-posed inverse problem relating the tissue phase to the underlying susceptibility distribution affects the accuracy for quantifying tissue susceptibility. Recently, deep learning has shown promising results to improve accuracy by reducing the streaking artifacts. However, there exists a mismatch between the observed phase and the theoretical forward phase estimated by the susceptibility label. In this study, we proposed a model-based deep learning architecture that followed the STI (susceptibility tensor imaging) physical model, referred to as MoDL-QSM. Specifically, MoDL-QSM accounts for the relationship between STI-derived phase contrast induced by the susceptibility tensor terms (χ13, χ23 and χ33) and the acquired single-orientation phase. The convolutional neural networks are embedded into the physical model to learn a regularization term containing prior information. χ33 and phase induced by χ13 and χ23 terms were used as the labels for network training. Quantitative evaluation metrics were compared with recently developed deep learning QSM methods. The results showed that MoDL-QSM achieved superior performance, demonstrating its potential for future applications.

Infrared small target detection based on region proposal and CNN classifier

Infrared small target detection has been a challenging task due to the weak radiation intensity of targets and the complexity of the background. Traditional methods using hand-designed features are usually effective for specific background and have the problems of low detection rate and high false alarm rate in complex infrared scene. In order to fully exploit the features of infrared image, this paper proposes an infrared small target detection method based on region proposal and convolution neural network. Firstly, the small target intensity is enhanced according to the local intensity characteristics. Then, potential target regions are proposed by corner detection to ensure high detection rate of the method. Finally, the potential target regions are fed into the classifier based on convolutional neural network to eliminate the non-target regions, which can effectively suppress the complex background clutter. Extensive experiments demonstrate that the proposed method can effectively reduce the false alarm rate, and outperform other state-of-the-art methods in terms of subjective visual impression and quantitative evaluation metrics.

Quantitative Evaluation of Faculty Research Productivity

A fair and objective evaluation of faculty research productivity is essential for faculty development and department success. At the University of Texas at San Antonio, we recently implemented a new system to quantify faculty performance for annual evaluations (Han, Wan, and Wang 2020). It provides a framework for quantitative performance evaluation using objective expectations and flexible weight among key measures. Annual research expenditures, proposals and new research grants, scholarly work (journal papers, books, book chapters, patents, conference presentations, seminar talks), graduate student support, and paper peer citations were used as key measures for research performance. Among these quantitative measures, the peer-reviewed journal publication is the core of scholarly work. It is widely accepted that both quality and quantity should be considered in evaluating journal publications, so it is imperative to incorporate journal paper quality into quantitative evaluations. This article proposes a quantitative method to weigh paper quality when quantifying journal publications. Although it can be a challenge to evaluate the quality and impact of a good journal paper, there are multiple ways to quantify the quality of such work, including using peer evaluation scoring, peer citation number, and journal impact factor. Authorship is also a way to assess faculty contribution and productivity. Peer citation is often used as an indicator of the long-term impact of a paper, but the number varies with different fields, and it cannot be used to evaluate the quality of newly published papers in the evaluation year. The quality/ranking of the journal in which a paper is published is frequently used as the surrogate for the quality of the paper. Some suggest using the impact factor (IF) in weighing the influence of a paper (Wiegers et al. 2015). The IF of a journal is based on the history of the peer citations of the papers published by the journal, and it varies with different fields. Journal quantile ranking (Q1, Q2, Q3, Q4) is the relative ranking of journals in their specific fields and thus advantageous over journal impact factor when diverse scientific fields are involved (such as in a mechanical engineering department), so we weigh the quality of the papers based on the journal ranking quantile (see table 1). Authorship is also factored into the evaluation. More weight is given to the first author and senior/corresponding authors than to coauthors. Previously, a report from the University of Athens Medical School described that they gave “100%, 50%, and 33% of the impact points and citation of an original or review article to the first or last author, second author, and third author, respectively; any other authorship position corresponds to 25% of the impact points and citations” (Mentzelopoulos and Zakynthinos 2017, 1222). A report from France describes that they assign “an index of 1 to the first or last author, an index of 0.5 for the second and the next to last author, and 0.25 for all other author positions” (Sahel 2011). An alternative is using fractional contributions. Harmonic models were proposed to give the i-th author a credit of 1/i (normalized based on the number of total authors) (Hagen 2010). Another fractional distribution model was proposed as 40 percent for the first and last authors and 20 percent to be divided among the other authors (or 30 percent for the first and last authors, 15 percent for the second and second to the last authors, and the remaining 10 percent divided among the other authors) (Abramo, Cicero, and D'Angelo 2013). Our faculty voted to give the first author and the senior author a weight of 1 and other coauthors a weight of 0.6 (see table 2). To encourage collaboration, we also accept co-first authorship and co-corresponding authorship. Although the approach is generally applicable, the specific weights are adjustable, as chairs and their faculty may select the weights that best fit their department. These quantitative factors can be incorporated into quantitative evaluation metrics to better evaluate the research productivity of faculty members (Han et al. 2020). There is a limitation to these weight distributions. Journal papers in the fields of mathematics, economics, and finance often list authors by alphabetical order as a common practice. Therefore, for papers in these areas, the credit should be evenly distributed among all authors (or use alternative approaches). There is also a limitation in the use of journal quantile. Some journal papers can be extremely high impact (such as some papers in science and nature), which is not reflected by the quantile; we use bonus points to alleviate this limitation (Han et al. 2020). The author thanks the faculty in the Department of Mechanical Engineering at the University of Texas at San Antonio for their input and support. Hai-Chao Han is professor and chair of the Department of Mechanical Engineering at the University of Texas at San Antonio. Email: [email protected]

Automatic semantic segmentation of breast tumors in ultrasound images based on combining fuzzy logic and deep learning-A feasibility study.

Computer aided diagnosis (CAD) of biomedical images assists physicians for a fast facilitated tissue characterization. A scheme based on combining fuzzy logic (FL) and deep learning (DL) for automatic semantic segmentation (SS) of tumors in breast ultrasound (BUS) images is proposed. The proposed scheme consists of two steps: the first is a FL based preprocessing, and the second is a Convolutional neural network (CNN) based SS. Eight well-known CNN based SS models have been utilized in the study. Studying the scheme was by a dataset of 400 cancerous BUS images and their corresponding 400 ground truth images. SS process has been applied in two modes: batch and one by one image processing. Three quantitative performance evaluation metrics have been utilized: global accuracy (GA), mean Jaccard Index (mean intersection over union (IoU)), and mean BF (Boundary F1) Score. In the batch processing mode: quantitative metrics' average results over the eight utilized CNNs based SS models over the 400 cancerous BUS images were: 95.45% GA instead of 86.08% without applying fuzzy preprocessing step, 78.70% mean IoU instead of 49.61%, and 68.08% mean BF score instead of 42.63%. Moreover, the resulted segmented images could show tumors' regions more accurate than with only CNN based SS. While, in one by one image processing mode: there has been no enhancement neither qualitatively nor quantitatively. So, only when a batch processing is needed, utilizing the proposed scheme may be helpful in enhancing automatic ss of tumors in BUS images. Otherwise applying the proposed approach on a one-by-one image mode will disrupt segmentation's efficiency. The proposed batch processing scheme may be generalized for an enhanced CNN based SS of a targeted region of interest (ROI) in any batch of digital images. A modified small dataset is available: https://www.kaggle.com/mohammedtgadallah/mt-small-dataset (S1 Data).

Correlation Evaluation of Functional Corticomuscular Coupling With Abnormal Muscle Synergy After Stroke.

While neuroplasticity and functional reorganization during motor recovery can be indirectly reflected and evaluated by functional corticomuscular coupling (fCMC), little work has been published regarding the cortical origin of abnormal muscle synergy and compensatory mechanism in the separation movement of stroke patients. In this study, we proposed to use extended partial directed coherence (ePDC) combined with an optimal spatial filtering approach to estimate fCMC in stroke patients and healthy controls, and further established muscle synergy model (MSM) to jointly explore the modulation mechanism between cortex and muscles. Compared to healthy controls, stroke patients had significantly reduced coupling strength in both descending and ascending pathway. Moreover, the MSM were abnormal with high variability and low similarity in the separation stage of stroke patients. Further exploration of the positive relationship between fCMC characteristics and MSM parameters proved the possibility of using fCMC-MSM-based correlation indicator to evaluate abnormality of the cortical related synergy movement as well as the rehabilitation level of stroke patients. We developed a computational procedure to evaluate the correlation between fCMC and MSM in stroke patients. This article provides a quantitative evaluation metrics based on fCMC to reveal the deficits during poststroke motor restoration and a promising approach to help patients correct abnormal movement habits, paving the way for neurophysiological assessment of neuromuscular control in conjunction with clinical scores.

Automatic Segmentation of Clinical Target Volumes for Post-Modified Radical Mastectomy Radiotherapy Using Convolutional Neural Networks.

BackgroundThis study aims to construct and validate a model based on convolutional neural networks (CNNs), which can fulfil the automatic segmentation of clinical target volumes (CTVs) of breast cancer for radiotherapy.MethodsIn this work, computed tomography (CT) scans of 110 patients who underwent modified radical mastectomies were collected. The CTV contours were confirmed by two experienced oncologists. A novel CNN was constructed to automatically delineate the CTV. Quantitative evaluation metrics were calculated, and a clinical evaluation was conducted to evaluate the performance of our model.ResultsThe mean Dice similarity coefficient (DSC) of the proposed model was 0.90, and the 95th percentile Hausdorff distance (95HD) was 5.65 mm. The evaluation results of the two clinicians showed that 99.3% of the chest wall CTV slices could be accepted by clinician A, and this number was 98.9% for clinician B. In addition, 9/10 of patients had all slices accepted by clinician A, while 7/10 could be accepted by clinician B. The score differences between the AI (artificial intelligence) group and the GT (ground truth) group showed no statistically significant difference for either clinician. However, the score differences in the AI group were significantly different between the two clinicians. The Kappa consistency index was 0.259. It took 3.45 s to delineate the chest wall CTV using the model.ConclusionOur model could automatically generate the CTVs for breast cancer. AI-generated structures of the proposed model showed a trend that was comparable, or was even better, than those of human-generated structures. Additional multicentre evaluations should be performed for adequate validation before the model can be completely applied in clinical practice.

Minibatch least-squares reverse time migration in a deep-learning framework

Migration techniques are an integral part of seismic imaging workflows. Least-squares reverse time migration (LSRTM) overcomes some of the shortcomings of conventional migration algorithms by compensating for illumination and removing sampling artifacts to increase spatial resolution. However, the computational cost associated with iterative LSRTM is high and convergence can be slow in complex media. We implement prestack LSRTM in a deep-learning framework and adopt strategies from the data science domain to accelerate convergence. Our hybrid framework leverages the existing physics-based models and machine-learning optimizers to achieve better and cheaper solutions. Using a time-domain formulation, we find that minibatch gradients can reduce the computation cost by using a subset of total shots for each iteration. The minibatch approach not only reduces source crosstalk, but it is also less memory-intensive. Combining minibatch gradients with deep-learning optimizers and loss functions can improve the efficiency of LSRTM. Deep-learning optimizers such as adaptive moment estimation are generally well-suited for noisy and sparse data. We compare different optimizers and determine their efficacy in mitigating migration artifacts. To accelerate the inversion, we adopt the regularized Huber loss function in conjunction. We apply these techniques to 2D Marmousi and 3D SEG/EAGE salt models and find improvements over conventional LSRTM baselines. Our approach achieves higher spatial resolution in less computation time measured by various qualitative and quantitative evaluation metrics.

Retrospective motion artifact correction of structural MRI images using deep learning improves the quality of cortical surface reconstructions

Head motion during MRI acquisition presents significant challenges for neuroimaging analyses. In this work, we present a retrospective motion correction framework built on a Fourier domain motion simulation model combined with established 3D convolutional neural network (CNN) architectures. Quantitative evaluation metrics were used to validate the method on three separate multi-site datasets. The 3D CNN was trained using motion-free images that were corrupted using simulated artifacts. CNN based correction successfully diminished the severity of artifacts on real motion affected data on a separate test dataset as measured by significant improvements in image quality metrics compared to a minimal motion reference image. On the test set of 13 image pairs, the mean peak signal-to-noise-ratio was improved from 31.7 to 33.3 dB. Furthermore, improvements in cortical surface reconstruction quality were demonstrated using a blinded manual quality assessment on the Parkinson's Progression Markers Initiative (PPMI) dataset. Upon applying the correction algorithm, out of a total of 617 images, the number of quality control failures was reduced from 61 to 38. On this same dataset, we investigated whether motion correction resulted in a more statistically significant relationship between cortical thickness and Parkinson's disease. Before correction, significant cortical thinning was found to be restricted to limited regions within the temporal and frontal lobes. After correction, there was found to be more widespread and significant cortical thinning bilaterally across the temporal lobes and frontal cortex. Our results highlight the utility of image domain motion correction for use in studies with a high prevalence of motion artifacts, such as studies of movement disorders as well as infant and pediatric subjects.

Multi-scale retinal vessel segmentation using encoder-decoder network with squeeze-and-excitation connection and atrous spatial pyramid pooling.

The segmentation of blood vessels in retinal images is crucial to the diagnosis of many diseases. We propose a deep learning method for vessel segmentation based on an encoder-decoder network combined with squeeze-and-excitation connection and atrous spatial pyramid pooling. In our implementation, the atrous spatial pyramid pooling allows the network to capture features at multiple scales, and the high-level semantic information is combined with low-level features through the encoder-decoder architecture to generate segmentations. Meanwhile, the squeeze-and-excitation connections in the proposed network can adaptively recalibrate features according to the relationship between different channels of features. The proposed network can achieve precise segmentation of retinal vessels without hand-crafted features or specific post-processing. The performance of our model is evaluated in terms of visual effects and quantitative evaluation metrics on two publicly available datasets of retinal images, the Digital Retinal Images for Vessel Extraction and Structured Analysis of the Retina datasets, with comparison to 12 representative methods. Furthermore, the proposed network is applied to vessel segmentation on local retinal images, which demonstrates promising application prospect in medical practices.

Visualization of Salient Object With Saliency Maps Using Residual Neural Networks

Visual saliency techniques based on Convolutional Neural Networks (CNNs) exhibit an excessive performance for saliency fixation in a scene, but it is harder to train a network in view of their complexity. The imparting Residual Network Model (ResNet) that is more capable to optimize features for predicting salient area in the form of saliency maps within the images. To get saliency maps, an amalgamated framework is presented that contains two streams of Residual Network Model (ResNet-50). Each stream of Reset-50 that is used to enhance the low-level and high-level semantics features and build a network of 99 layers at two different image scales for generating the normal saliency attention. This model is trained with transfer learning for initialization that is pretrained on ImageNet for object detection, and with some modifications to minimize prediction error. At the end, the two streams integrate the features by fusion at low and high scale dimensions of images. This model is fine-tuned on four commonly used datasets and examines both qualitative and quantitative evaluation metrics for state-of-the-art deep saliency model outcomes.

Automated Multimodal image fusion for brain tumor detection

In medical domain, various multimodalities such as Computer tomography (CT) and Magnetic resonance imaging (MRI) are integrated into a resultant fused image. Image fusion (IF) is a method by which vital information can be preserved by extracting all important information from the multiple images into the resultant fused image. The analytical and visual image quality can be enhanced by the integration of different images. In this paper, a new algorithm has been proposed on the basis of guided filter with new fusion rule for the fusion of different imaging modalities such as MRI and Fluorodeoxyglucose images of brain for the detection of tumor. The performance of the proposed method has been evaluated and compared with state-of-the-art image fusion techniques using various qualitative as well as quantitative evaluation metrics. From the results, it has been observed that more information has achieved on edges and content visibility is also high as compared to the other techniques which makes it more suitable for real applications. The experimental results are evaluated on the basis of with-reference and without-references metric such as standard deviation, entropy, peak signal to noise ratio, mutual information etc.