Non-medical Images Research Articles

Medical image encryption algorithm based on Fresnel zone formula, differential neural networks, and pixel-guided perturbation techniques

This paper proposes an image encryption technique that integrates deep pixel substitution, data-dependent and chaotic pixel perturbation, and differential neural networks. The pixels of the image are manipulated using a deep pixel substitution operation that is based on the Fresnel Zone equation to eliminate the correlations to the input plain image. Additionally, a perturbation process based on pixel values and chaotic noise is applied to further scramble the image. The resulting image is then subjected to a second round of deep substitution. The differential neural network generates blurring codes by incorporating plain pixel blocks and an encryption key, which are subsequently added to the processed image to produce the final ciphered image. The proposed technique’s effectiveness was evaluated on a large dataset that included both medical and non-medical images. Simulation results indicated that the proposed technique was not only efficient but also effective for both medical and non-medical images, and it outperformed state-of-the-art encryption methods in both security properties and computational efficiency.

From Research on TCM Images to Historiography of TCM Images——Comments on History of Chinese Image Culture: Medical Image Volume

Abstract Zhong Hua Tu Xiang Wen Hua Shi Yi Yao Tu Xiang Juan (《中华图像文化史•医药图像卷》History of Chinese Image Culture: Medical Image Volume) is a history of images of Chinese medicine. This work meticulously traces the evolution of Chinese medicine from the pre-Qin and Han dynasties through to the late Qing dynasty and the Republican era, utilizing a rich array of medical and non-medical images. Through detailed analysis and interpretation of these images, the volume reconstructs the developmental trajectory of traditional Chinese medicine (TCM) theories, while also revealing the impact of ideological, cultural, and societal contexts on medical culture and theoretical advancement. Zhong Hua Tu Xiang Wen Hua Shi Yi Yao Tu Xiang Juan has pushed the research of Chinese medical images from purely literary studies to a broader historical context, broadened the research object of Chinese medical history, and constructed a basic framework for the new research field of image medical history. The book not only facilitates a deeper understanding of TCM's historical and contemporary developments, but also holds significant value for scholars engaged in historical literature research or clinical practice, aiming to integrate historical insights.

Enhancing diagnostic deep learning via self-supervised pretraining on large-scale, unlabeled non-medical images

BackgroundPretraining labeled datasets, like ImageNet, have become a technical standard in advanced medical image analysis. However, the emergence of self-supervised learning (SSL), which leverages unlabeled data to learn robust features, presents an opportunity to bypass the intensive labeling process. In this study, we explored if SSL for pretraining on non-medical images can be applied to chest radiographs and how it compares to supervised pretraining on non-medical images and on medical images.MethodsWe utilized a vision transformer and initialized its weights based on the following: (i) SSL pretraining on non-medical images (DINOv2), (ii) supervised learning (SL) pretraining on non-medical images (ImageNet dataset), and (iii) SL pretraining on chest radiographs from the MIMIC-CXR database, the largest labeled public dataset of chest radiographs to date. We tested our approach on over 800,000 chest radiographs from 6 large global datasets, diagnosing more than 20 different imaging findings. Performance was quantified using the area under the receiver operating characteristic curve and evaluated for statistical significance using bootstrapping.ResultsSSL pretraining on non-medical images not only outperformed ImageNet-based pretraining (p < 0.001 for all datasets) but, in certain cases, also exceeded SL on the MIMIC-CXR dataset. Our findings suggest that selecting the right pretraining strategy, especially with SSL, can be pivotal for improving diagnostic accuracy of artificial intelligence in medical imaging.ConclusionsBy demonstrating the promise of SSL in chest radiograph analysis, we underline a transformative shift towards more efficient and accurate AI models in medical imaging.Relevance statementSelf-supervised learning highlights a paradigm shift towards the enhancement of AI-driven accuracy and efficiency in medical imaging. Given its promise, the broader application of self-supervised learning in medical imaging calls for deeper exploration, particularly in contexts where comprehensive annotated datasets are limited.Graphical

Uncovering the effects of model initialization on deep model generalization: A study with adult and pediatric chest X-ray images.

Model initialization techniques are vital for improving the performance and reliability of deep learning models in medical computer vision applications. While much literature exists on non-medical images, the impacts on medical images, particularly chest X-rays (CXRs) are less understood. Addressing this gap, our study explores three deep model initialization techniques: Cold-start, Warm-start, and Shrink and Perturb start, focusing on adult and pediatric populations. We specifically focus on scenarios with periodically arriving data for training, thereby embracing the real-world scenarios of ongoing data influx and the need for model updates. We evaluate these models for generalizability against external adult and pediatric CXR datasets. We also propose novel ensemble methods: F-score-weighted Sequential Least-Squares Quadratic Programming (F-SLSQP) and Attention-Guided Ensembles with Learnable Fuzzy Softmax to aggregate weight parameters from multiple models to capitalize on their collective knowledge and complementary representations. We perform statistical significance tests with 95% confidence intervals and p-values to analyze model performance. Our evaluations indicate models initialized with ImageNet-pretrained weights demonstrate superior generalizability over randomly initialized counterparts, contradicting some findings for non-medical images. Notably, ImageNet-pretrained models exhibit consistent performance during internal and external testing across different training scenarios. Weight-level ensembles of these models show significantly higher recall (p<0.05) during testing compared to individual models. Thus, our study accentuates the benefits of ImageNet-pretrained weight initialization, especially when used with weight-level ensembles, for creating robust and generalizable deep learning solutions.

Transfer Learning from Non-Medical Images to Medical Images Using Deep Learning Algorithms

Introduction: Machine learning, especially deep convolutional neural networks (DCNNs), is a popular method for computerizing medical image analysis. This study aimed to develop DCNN models for histopathology image classification utilizing transfer learning.Material and Methods: We utilized 16 different pre-trained DCNNs to analyze the histopathology images from the animal diagnostic laboratory (ADL) database. During the image preprocessing stage, we applied two methods. The first method involved subtracting the mean of ImageNet images from all images. The second method involved subtracting the mean of histopathology training images from all images. Next, in the 16 pre-trained networks, feature extraction was done from their final six layers, and the features extracted from each layer were fed separately into the linear and non-linear support vector machine (SVM) for classification.Results: The results obtained from the ADL database show that the classification rate in lung tissue images is much better than that of the kidney and spleen. For example, the lowest detection rate in non-linear SVM for lung tissue is 14.96%, almost close to the highest accuracy in kidney and spleen tissue. The classification accuracy of the spleen images is better than that of the kidneys, with only a slight difference. In linear SVM on lung images, ResNet101 obtained the most accurate result with a value of 99.56%, followed by ResNet50, ResNet152, VGG_16, and VGG_19. In non-linear SVM on lung tissue images, the ResNet101 network with 99.65% and ResNet50 with 99.21%, followed by ResNet152, VGG_16, and VGG_19 obtained the highest detection rate.Conclusion: The classification results obtained from different methods on the ADL (including kidney, spleen, and lung histopathology images) database, confirmed the validity of transferring knowledge between non-medical and medical histopathology images. Additionally, it demonstrates the success of combining classifiers trained on deep features. This research obtained higher accuracy in the ADL database than the works done.

Blind watermarking scheme for medical and non-medical images copyright protection using the QZ algorithm

The widely used watermarking technique is to hide the watermark information through adding process, which means the watermark is added with the host image to realize embedding. The constraints of these addition-based watermarking systems mainly are non-blind, they are vulnerable to forgery attacks and have the false-positive problem. To solve these problems, a novel blind watermarking scheme based on the QZ algorithm for medical and normal images is first presented in this paper. In the presented scheme, we take the watermark and the host image as two inputs of the QZ algorithm to realize embedding, rather than the adding process. The outputs of the QZ algorithm include four parts: two upper quasitriangular matrices and two unitary matrices. Then the three parts of the outputs are synthesized by the vector synthesis and inverse discrete wavelet transform-singular value decomposition (DWT-SVD) to obtain the watermarked host image. Due to the characteristics of the QZ algorithm, our scheme is blind. The performance is tested and compared by applying our scheme to both medical and non-medical images. The experimental results show that our scheme is robust against various image-processing attacks, forgery attacks, and false-positive problem. And from the comparison results, the proposed scheme outperforms the existing addition-based schemes in terms of robustness against the Median filter, Low-pass filter, Gaussian blur attacks, and forgery attacks, while maintaining good invisibility.

Analysis of Specimen Mammography with Artificial Intelligence to Predict Margin Status.

Intraoperative specimen mammography is a valuable tool in breast cancer surgery, providing immediate assessment of margins for a resected tumor. However, the accuracy of specimen mammography in detecting microscopic margin positivity is low. We sought to develop an artificial intelligence model to predict the pathologic margin status of resected breast tumors using specimen mammography. A dataset of specimen mammography images matched with pathologic margin status was collected from our institution from 2017 to 2020. The dataset was randomly split into training, validation, and test sets. Specimen mammography models pretrained on radiologic images were developed and compared with models pretrained on nonmedical images. Model performance was assessed using sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC). The dataset included 821 images, and 53% had positive margins. For three out of four model architectures tested, models pretrained on radiologic images outperformed nonmedical models. The highest performing model, InceptionV3, showed sensitivity of 84%, specificity of 42%, and AUROC of 0.71. Model performance was better among patients with invasive cancers, less dense breasts, and non-white race. This study developed and internally validated artificial intelligence models that predict pathologic margins status for partial mastectomy from specimen mammograms. The models' accuracy compares favorably with published literature on surgeon and radiologist interpretation of specimen mammography. With further development, these models could more precisely guide the extent of resection, potentially improving cosmesis and reducing reoperations.

Region based medical image encryption using advanced zigzag transform and 2D logistic sine map (2DLSM)

A large number of medical images are generated for diagnostic purposes, disease monitoring, research and education, quality control in health services, and so on. The secure transmission and storage of them demand a significant effort. Most of the available encryption schemes are designed for non-medical images, whereas medical images need a higher level of security and robust authentication. Additionally, in certain cases, only a specific part of the image, which may be separated into the region of interest and the region of background, medical images can be divided into these two regions. A region-based medical image encryption using a 2D logistic sine map (2DLSM) and an advanced zig zag transform is used to secure medical images. First, the Region of Interest (ROI) is extracted from the original medical image using basic morphological techniques, including edge detection, dilation, and erosion. Secondly, the ROI is encrypted using a complex zigzag transform and a 2D logistic sine map (2DLSM). Advanced zigzag transform that crosses in both directions while beginning at random points to jumble the image. This new zigzag transform method is more complex than existing zigzag transform techniques because the number of sequence types is equal to the number of pixels in the plaintext image. The confused image is diffused using a random sequence created using the 2D logistic sine map approach after numerous iterations of an advanced zigzag transformation. In order to save time and computational resources, the background region pixels are eliminated during encryption. Experiments and security analyses show that the suggested approach is strong in defending against diverse assaults and can effectively secure ROI of different types and sizes of medical photos.

Analysis and effectiveness of deeper levels of SVD on performance of hybrid DWT and SVD watermarking

In this paper, an analysis on hybrid Discrete Wavelet Transform (DWT) and Singular Value Decomposition (SVD) for image watermarking is carried out to investigate the effect of a deeper level of the SVD on imperceptibility and robustness to resist common signal processing and geometric attacks. For this purpose, we have designed two hybrid watermarking schemes, the first one with DWT and first level of SVD, whereas, in the second scheme, the same design is employed with a second level of SVD. In this experiment, a comprehensive analysis is performed on the two designed schemes and the effect of robustness and imperceptibility is compared in the first and second levels of SVD in each DWT sub-band. Having analyzed more than 100 medical and non-medical images in standard datasets and real medical samples of patients, the experimental outcomes show a remarkable increase in both imperceptibility and robustness in the second level of SVD, in comparison to the first level. In addition, the achieved result shows that the SVD2 scheme offers the highest imperceptibility in the LL sub-band (more than 60 dB on average PSNR), with satisfactory robustness against noise attacks, but less persistence in some geometric attacks such as cropping. For the HH sub-band, strong robustness against all types of tested of attacks is obtained, though its imperceptibility is slightly lower than the achieved PSNR in the LL sub-band. In HH sub-band, an average growth of 5 dB in PSNR and 2% in NC can be observed from the second level of SVD in comparison to the first level. These results make SVD2 a good candidate for content protection, especially for medical images.

Maxillofacial Fracture Detection Using Transfer Learning Models : A Review

Early detection and treatment of face bone fractures reduce long-term problems. Fracture identification needs CT scan interpretation, but there aren't enough experts. To address these issues, researchers are classifying and identifying objects. Categorization-based studies can't pinpoint fractures. Proposed Study Convolutional neural networks with transfer learning may detect maxillofacial fractures. CT scans were utilized to retrain and fine-tune a convolutional neural network trained on non-medical images to categorize incoming CTs as "Positive" or "Negative." Model training employed maxillofacial fractogram data. If two successive slices had a 95% fracture risk, the patient had a fracture. In terms of sensitivity/person for facial fractures, the recommended strategy beat the machine learning model. The recommended approach may minimize physicians' effort identifying facial bone fractures in face CT. Even though technology can't fully replace a radiologist, the recommended technique may be helpful. It reduces human error, diagnostic delays, and hospitalization costs.

An overview of artificial intelligence in diabetic retinopathy and other ocular diseases.

Artificial intelligence (AI), also known as machine intelligence, is a branch of science that empowers machines using human intelligence. AI refers to the technology of rendering human intelligence through computer programs. From healthcare to the precise prevention, diagnosis, and management of diseases, AI is progressing rapidly in various interdisciplinary fields, including ophthalmology. Ophthalmology is at the forefront of AI in medicine because the diagnosis of ocular diseases heavy reliance on imaging. Recently, deep learning-based AI screening and prediction models have been applied to the most common visual impairment and blindness diseases, including glaucoma, cataract, age-related macular degeneration (ARMD), and diabetic retinopathy (DR). The success of AI in medicine is primarily attributed to the development of deep learning algorithms, which are computational models composed of multiple layers of simulated neurons. These models can learn the representations of data at multiple levels of abstraction. The Inception-v3 algorithm and transfer learning concept have been applied in DR and ARMD to reuse fundus image features learned from natural images (non-medical images) to train an AI system with a fraction of the commonly used training data (<1%). The trained AI system achieved performance comparable to that of human experts in classifying ARMD and diabetic macular edema on optical coherence tomography images. In this study, we highlight the fundamental concepts of AI and its application in these four major ocular diseases and further discuss the current challenges, as well as the prospects in ophthalmology.

An extensive review of state-of-the-art transfer learning techniques used in medical imaging: Open issues and challenges

Abstract Deep learning techniques, which use a massive technology known as convolutional neural networks, have shown excellent results in a variety of areas, including image processing and interpretation. However, as the depth of these networks grows, so does the demand for a large amount of labeled data required to train these networks. In particular, the medical field suffers from a lack of images because the procedure for obtaining labeled medical images in the healthcare field is difficult, expensive, and requires specialized expertise to add labels to images. Moreover, the process may be prone to errors and time-consuming. Current research has revealed transfer learning as a viable solution to this problem. Transfer learning allows us to transfer knowledge gained from a previous process to improve and tackle a new problem. This study aims to conduct a comprehensive survey of recent studies that dealt with solving this problem and the most important metrics used to evaluate these methods. In addition, this study identifies problems in transfer learning techniques and highlights the problems of the medical dataset and potential problems that can be addressed in future research. According to our review, many researchers use pre-trained models on the Imagenet dataset (VGG16, ResNet, Inception v3) in many applications such as skin cancer, breast cancer, and diabetic retinopathy classification tasks. These techniques require further investigation of these models, due to training them on natural, non-medical images. In addition, many researchers use data augmentation techniques to expand their dataset and avoid overfitting. However, not enough studies have shown the effect of performance with or without data augmentation. Accuracy, recall, precision, F1 score, receiver operator characteristic curve, and area under the curve (AUC) were the most widely used measures in these studies. Furthermore, we identified problems in the datasets for melanoma and breast cancer and suggested corresponding solutions.

An application of non-dominated sorting genetic algorithm for reversible data hiding based on histogram shifting in neuroimages

This paper presents an application of multi-objective non-dominated sorting genetic algorithm with a modified chromosome encoding for histogram shifting-based multiple reversible data hiding scheme in neuroimages which aims to minimize distortion and maximize capacity. The modified chromosomes encoding scheme is designed according to the zero-bin characteristic of the intensity histogram of the structural magnetic resonance imaging scans of human brain. The experiments of this study is designed to assess the effect of non-dominated sorting for multi-objective optimization compared to Euclidian distance, the convenience of modified chromosome encoding scheme for medical images compared to non-medical images. The performance of the proposed method has been measured in terms of the peak signal-to-noise ratio (PSNR) for image quality and the bits per pixel (bpp) for capacity assessments. The experimental results show that the proposed method is better than its counterparts.

Automated assessment of transthoracic echocardiogram image quality using deep neural networks

BackgroundStandard views in two-dimensional echocardiography are well established but the qualities of acquired images are highly dependent on operator skills and are assessed subjectively. This study was aimed at providing an objective assessment pipeline for echocardiogram image quality by defining a new set of domain-specific quality indicators. Consequently, image quality assessment can thus be automated to enhance clinical measurements, interpretation, and real-time optimization. MethodsWe developed deep neural networks for the automated assessment of echocardiographic frames that were randomly sampled from 11,262 adult patients. The private echocardiography dataset consists of 33,784 frames, previously acquired between 2010 and 2020. Unlike non-medical images where full-reference metrics can be applied for image quality, echocardiogram's data are highly heterogeneous and requires blind-reference (IQA) metrics. Therefore, deep learning approaches were used to extract the spatiotemporal features and the image's quality indicators were evaluated against the mean absolute error. Our quality indicators encapsulate both anatomical and pathological elements to provide multivariate assessment scores for anatomical visibility, clarity, depth-gain and foreshortedness. ResultsThe model performance accuracy yielded 94.4%, 96.8%, 96.2%, 97.4% for anatomical visibility, clarity, depth-gain and foreshortedness, respectively. The mean model error of 0.375±0.0052 with computational speed of 2.52 ms per frame (real-time performance) was achieved. ConclusionThe novel approach offers new insight to the objective assessment of transthoracic echocardiogram image quality and clinical quantification in A4C and PLAX views. It also lays stronger foundations for the operator's guidance system which can leverage the learning curve for the acquisition of optimum quality images during the transthoracic examination.

Multi-structure bone segmentation in pediatric MR images with combined regularization from shape priors and adversarial network

Morphological and diagnostic evaluation of pediatric musculoskeletal system is crucial in clinical practice. However, most segmentation models do not perform well on scarce pediatric imaging data. We propose a new pre-trained regularized convolutional encoder–decoder network for the challenging task of segmenting heterogeneous pediatric magnetic resonance (MR) images. To this end, we have conceived a novel optimization scheme for the segmentation network which comprises additional regularization terms to the loss function. In order to obtain globally consistent predictions, we incorporate a shape priors based regularization, derived from a non-linear shape representation learnt by an auto-encoder. Additionally, an adversarial regularization computed by a discriminator is integrated to encourage precise delineations. The proposed method is evaluated for the task of multi-bone segmentation on two scarce pediatric imaging datasets from ankle and shoulder joints, comprising pathological as well as healthy examinations. The proposed method performed either better or at par with previously proposed approaches for Dice, sensitivity, specificity, maximum symmetric surface distance, average symmetric surface distance, and relative absolute volume difference metrics. We illustrate that the proposed approach can be easily integrated into various bone segmentation strategies and can improve the prediction accuracy of models pre-trained on large non-medical images databases. The obtained results bring new perspectives for the management of pediatric musculoskeletal disorders.

A Multilevel Transfer Learning Technique and LSTM Framework for Generating Medical Captions for Limited CT and DBT Images.

Medical image captioning has been recently attracting the attention of the medical community. Also, generating captions for images involving multiple organs is an even more challenging task. Therefore, any attempt toward such medical image captioning becomes the need of the hour. In recent years, the rapid developments in deep learning approaches have made them an effective option for the analysis of medical images and automatic report generation. But analyzing medical images that are scarce and limited is hard, and it is difficult even with machine learning approaches. The concept of transfer learning can be employed in such applications that suffer from insufficient training data. This paper presents an approach to develop a medical image captioning model based on a deep recurrent architecture that combines Multi Level Transfer Learning (MLTL) framework with a Long Short-Term-Memory (LSTM) model. A basic MLTL framework with three models is designed to detect and classify very limited datasets, using the knowledge acquired from easily available datasets. The first model for the source domain uses the abundantly available non-medical images and learns the generalized features. The acquired knowledge is then transferred to the second model for the intermediate and auxiliary domain, which is related to the target domain. This information is then used for the final target domain, which consists of medical datasets that are very limited in nature. Therefore, the knowledge learned from a non-medical source domain is transferred to improve the learning in the target domain that deals with medical images. Then, a novel LSTM model, which is used for sequence generation and machine translation, is proposed to generate captions for the given medical image from the MLTL framework. To improve the captioning of the target sentence further, an enhanced multi-input Convolutional Neural Network (CNN) model along with feature extraction techniques is proposed. This enhanced multi-input CNN model extracts the most important features of an image that help in generating a more precise and detailed caption of the medical image. Experimental results show that the proposed model performs well with an accuracy of 96.90%, with BLEU score of 76.9%, even with very limited datasets, when compared to the work reported in literature.

A Simple Image Encryption Based on Binary Image Affine Transformation and Zigzag Process

In this paper, we propose a new and simple method for image encryption. It uses an external secret key of 128 bits long and an internal secret key. The novelties of the proposed encryption process are the methods used to extract an internal key to apply the zigzag process, affine transformation, and substitution-diffusion process. Initially, an original gray-scale image is converted into binary images. An internal secret key is extracted from binary images. The two keys are combined to compute the substitution-diffusion keys. The zigzag process is firstly applied on each binary image. Using an external key, every zigzag binary image is reflected or rotated and a new gray-scale image is reconstructed. The new image is divided into many nonoverlapping subblocks, and each subblock uses its own key to take out a substitution-diffusion process. We tested our algorithms on many biomedical and nonmedical images. It is seen from evaluation metrics that the proposed image encryption scheme provides good statistical and diffusion properties and can resist many kinds of attacks. It is an efficient and secure scheme for real-time encryption and transmission of biomedical images in telemedicine.

Hybrid Sharpening Transformation Approach for Multifocus Image Fusion Using Medical and Nonmedical Images.

In this study, we introduced a preprocessing novel transformation approach for multifocus image fusion. In the multifocus image, fusion has generated a high informative image by merging two source images with different areas or objects in focus. Acutely the preprocessing means sharpening performed on the images before applying fusion techniques. In this paper, along with the novel concept, a new sharpening technique, Laplacian filter + discrete Fourier transform (LF + DFT), is also proposed. The LF is used to recognize the meaningful discontinuities in an image. DFT recognizes that the rapid change in the image is like sudden changes in the frequencies, low-frequency to high-frequency in the images. The aim of image sharpening is to highlight the key features, identifying the minor details, and sharpen the edges while the previous methods are not so effective. To validate the effectiveness the proposed method, the fusion is performed by a couple of advanced techniques such as stationary wavelet transform (SWT) and discrete wavelet transform (DWT) with both types of images like grayscale and color image. The experiments are performed on nonmedical and medical (breast medical CT and MRI images) datasets. The experimental results demonstrate that the proposed method outperforms all evaluated qualitative and quantitative metrics. Quantitative assessment is performed by eight well-known metrics, and every metric described its own feature by which it is easily assumed that the proposed method is superior. The experimental results of the proposed technique SWT (LF + DFT) are summarized for evaluation matrices such as RMSE (5.6761), PFE (3.4378), MAE (0.4010), entropy (9.0121), SNR (26.8609), PSNR (40.1349), CC (0.9978), and ERGAS (2.2589) using clock dataset.

Image Super-Resolution Through Compressive Sensing-based Recovery.

The primary aim of image super-resolution techniques is to produce a high resolution (HR) image from a low resolution (LR) image efficiently. Deep learning algorithms are being extensively used to address the ill-posed problem of single image super-resolution which requires extremely large data-sets and high processing power. When one does not have access to large data-sets or have limited processing power, an alternative technique may be in order. In this study, we have developed a novel positive scale image resizing method inspired by compressive sensing (CS). We have considered the image super-resolution as a CS recovery problem in which a low resolution image is assumed as a compressed measurement and the required interpolated image is treated as output of the CS-based recovery. In the proposed HR recovery method, a deterministic binary block diagonal measurement matrix, (DBBD), is used as measurement matrix since it maintains the visual similarity between the low and high resolution images. Then along with a sparsification matrix, the sparse representation of HR image is first recovered and subsequently the dense HR image is obtained. The proposed method is applied to medical and non-medical images. The HR images obtained using the traditional proximal, bilinear and bi-cubic interpolation techniques are compared with those obtained using the proposed method. The proposed CS inspired method delivers superior HR images than the traditional techniques. The superiority of the proposed method is attributed to the unique usage of the DBBD matrix and the CS recovery algorithm to obtain a high resolution image without any prior training data-set.

Transfer Learning for an Automated Detection System of Fractures in Patients with Maxillofacial Trauma

An original maxillofacial fracture detection system (MFDS), based on convolutional neural networks and transfer learning, is proposed to detect traumatic fractures in patients. A convolutional neural network pre-trained on non-medical images was re-trained and fine-tuned using computed tomography (CT) scans to produce a model for the classification of future CTs as either “fracture” or “noFracture”. The model was trained on a total of 148 CTs (120 patients labeled with “fracture” and 28 patients labeled with “noFracture”). The validation dataset, used for statistical analysis, was characterized by 30 patients (5 with “noFracture” and 25 with “fracture”). An additional 30 CT scans, comprising 25 “fracture” and 5 “noFracture” images, were used as the test dataset for final testing. Tests were carried out both by considering the single slices and by grouping the slices for patients. A patient was categorized as fractured if two consecutive slices were classified with a fracture probability higher than 0.99. The patients’ results show that the model accuracy in classifying the maxillofacial fractures is 80%. Even if the MFDS model cannot replace the radiologist’s work, it can provide valuable assistive support, reducing the risk of human error, preventing patient harm by minimizing diagnostic delays, and reducing the incongruous burden of hospitalization.