Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
16 results
Search Results
Item Ensemble deep neural models for automated abnormality detection and classification in precision care applications(Elsevier, 2023) Karthik, K.; Mayya, V.; Kamath S․, S.Radiological imaging is one of the most relied upon modalities in the clinical diagnosis and treatment planning process. Conventional diagnosis involves the manual analysis of radiology images by experienced radiologists, which is often a time-consuming and labor-intensive process. The scarcity of experienced radiologists and necessity of large-scale X-rays image analysis given the huge diagnosis workload at most hospitals stresses the need for automated clinical diagnosis systems capable of fast and accurate identification of abnormalities, disease characteristic identification, disease classification, and others. Such automated methods are thus a fundamental requirement in clinical workflow management applications. In this work, we present an approach for multitask clinical objectives such as disease classification and detection of abnormalities. The proposed model leverages the predictive power of deep neural models for enabling evidence-based diagnosis. During validation experiments, the model achieved an accuracy of 89.58% along with sensitivity and specificity of 85.83% and 90.83%, respectively, with an AUC (area under the ROC curve) of 95.84% for normal/no findings versus COVID-19 chest radiograph classification and an accuracy of 73.19% for upper extremity musculoskeletal images. The performance of the model for the classification and abnormality identification tasks, when benchmarked over multiple standard datasets, emphasizes its suitability and adaptability in real-world clinical settings, with significant improvements in radiology-based diagnosis workflow and patient care. © 2023 Elsevier Inc. All rights reserved.Item GAN-Based Encoder-Decoder Model for Multi-Label Diagnostic Scan Classification and Automated Radiology Report Generation(CRC Press, 2024) Kumar, R.; Karthik, K.; Kamath S․, S.S.X-ray imaging is one of the most popular diagnostic imaging techniques and plays a critical role in the diagnosis and treatment process. Given the huge volume of patients and scans performed in most hospitals each day, the current practice of manual analysis of such scan images by experienced radiologists is a time-consuming and often error-prone process, worsened by the cognitive burden experienced by the radiologists. Conventional diagnostic reports written by radiologists after radiological image capture contain radiography-specific keywords (tags), observations of different body parts in the image (findings), and the actual diagnosis (impression). Automated multi-label classification of X-ray scans for disease prediction, and generation of an associated textual diagnostic scan report can ease the burden for radiologists, while also enabling fast, localized, and explanatory analysis. In this work, GAN-MLC, a CNN-LSTM description generator model trained in the adversarial setup, is proposed for the multi-label classification of X-ray images and improved feature learning for capturing disease-specific findings. Experiments performed on the NIH Chest X-ray Dataset revealed that the proposed GAN-MLC outperformed CNN-based models by a significant margin of more than seven percent. For the text diagnostic report generation task, the GAN-MLC achieved promising BLEU scores and was more robust against overfitting issues. © 2024 selection and editorial matter, Bhanu Chander, Koppala Guravaiah, B. Anoop, and G. Kumaravelan; individual chapters, the contributors.Item Influence of welding process parameters on microstructure and mechanical properties of friction stir welded aluminium matrix composite(Trans Tech Publications Ltd ttp@transtec.ch, 2017) Prabhu B, S.; Shettigar, A.K.; Karthik, K.; Rao, S.S.; Herbert, M.In this study, the effect of process parameters on microstructure and mechanical properties of friction stir welded aluminium matrix composites(AMC) have been explored. The results indicated that the recrystallized grain size at the bottom of the weld region is smaller than that at the top region due to difference in the heat transfer at the weld region. The joint strength of AMCs depends on proper selection of process parameters like tool rotational speed and welding speed. If process parameter values are beyond the optimal value, the joint strength decreases due to formation of defects. Maximum tensile strength is obtained for rotational speed of 1000 rpm and welding speed of 80mm/min. © 2017 Trans Tech Publications, Switzerland.Item A Hybrid Feature Modeling Approach for Content-Based Medical Image Retrieval(Institute of Electrical and Electronics Engineers Inc., 2018) Karthik, K.; Kamath S․, S.With the proliferation of various imaging based diagnostic procedures in the healthcare field, patient-specific scan images constitute huge volumes of data that needs to be well-organized and managed for supporting clinical decision support applications. One such crucial application with a significant impact on point-of-care treatment quality is a Content Based Medical Image Retrieval (CBMIR) system that can assist doctors in disease diagnosis based on similar image retrieval. Medical images are multi-dimensional and often contain manifold information, due to which efficient techniques for optimal feature extraction from large-scale image collections are the need of the day. In this paper, an efficient CBMIR model is proposed that is built on multi-level feature sets extracted from medical images. Four different feature extraction techniques are used to optimally represent images in a multi-dimensional feature space, for facilitating classification using supervised machine learning algorithms and top-k similar image retrieval. Experimental validation of proposed model on the standard ImageCLEF 2009 dataset containing 12,560 X-ray images across 116 classes showed promising results in terms of classification accuracy of 85.91%. © 2018 IEEE.Item Automatic Quality Enhancement of Medical Diagnostic Scans with Deep Neural Image Super-Resolution Models(Institute of Electrical and Electronics Engineers Inc., 2020) Karthik, K.; Kamath S․, S.; Kamath, S.U.In modern healthcare, diagnostic imaging is an essential component for diagnosing ailments and delivering quality healthcare. Given the variety in medical scanning techniques, a recurring issue across different modalities is that the scan quality is often affected by artifacts introduced by hardware and software faults in the imaging equipment. Significant challenges in the 3D Imaging Techniques include low quality/low-resolution scan images or the addition of unwanted artifacts due to patient movement. Researchers have put forth solutions ranging from machine learning algorithms like Gradient Descent to more complex Deep CNN models for rectifying these faults. In this paper, we aim to benchmark deep learning models for improving the quality of diagnostic images, through Super-resolution, for enabling faster and easier detection of anomalies that may be missed otherwise. Super-resolution CNN and Deep CNN architectures were employed for up-sampling medical scans for enhancing their quality. The CNN models were trained to learn motion artifact characteristics that are a result of patient movement and negate its effects in the super-resolved output. We present comparative results of six super-resolution models on a standard dataset and metrics. During the experimental evaluation, it was observed that the ResNet SRCNN model outperformed all other models used for comparison by a large margin, with an improvement of 4.87 to 8.68% over the other state-of-the-art models. © 2020 IEEE.Item Analysis and Prediction of Fantasy Cricket Contest Winners Using Machine Learning Techniques(Springer Science and Business Media Deutschland GmbH info@springer-sbm.com, 2021) Karthik, K.; S. Krishnan, G.S.; Shetty, S.; Bankapur, S.; Kolkar, R.; Ashwin, T.S.; Vanahalli, M.K.Cricket is one of the well-known sports across the world. The increasing interest of cricket in recent years resulted in different forms like T20, T10 from test and one day format. The craze of all these formats of cricket matches today has come into online fantasy cricket league games. Dream11 is one such app that is most popular in this context, along with many similar apps. Creating a dream team of 11 players from playing 11 of both teams involves skills, ideas and luck. Predicting a winner among all the joined contestants based on the previous historical data is a challenging task. In this paper, we used a feed-forward deep neural network (DNN) classifier for predicting the winning contestant for the top three positions in a fantasy league cricket contest. The performance of the DNN approach was compared against that of state-of-the-art machine learning approaches like k-nearest neighbours (KNN), logistic regression (LR), Naive Bayes (NB), random forest (RF), support vector machines (SVM) and in predicting the fantasy cricket contest winners. Among the methods used, DNN showed the best results for all three positions, showing its consistency in predicting the winners and outperforms the state-of-the-art machine learning classifiers by 13%, 8% and 9%, respectively, for first, second and third winning positions, respectively. © 2021, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.Item COVIDDX: AI-based clinical decision support system for learning COVID-19 disease representations from multimodal patient data(SciTePress, 2021) Mayya, V.; Karthik, K.; Kamath S․, S.; Karadka, K.; Jeganathan, J.The COVID-19 pandemic has affected the world on a global scale, infecting nearly 68 million people across the world, with over 1.5 million fatalities as of December 2020. A cost-effective early-screening strategy is crucial to prevent new outbreaks and to curtail the rapid spread. Chest X-ray images have been widely used to diagnose various lung conditions such as pneumonia, emphysema, broken ribs and cancer. In this work, we explore the utility of chest X-ray images and available expert-written diagnosis reports, for training neural network models to learn disease representations for diagnosis of COVID-19. A manually curated dataset consisting of 450 chest X-rays of COVID-19 patients and 2,000 non-COVID cases, along with their diagnosis reports were collected from reputed online sources. Convolutional neural network models were trained on this multimodal dataset, for prediction of COVID-19 induced pneumonia. A comprehensive clinical decision support system powered by ensemble deep learning models (CADNN) is designed and deployed on the web. The system also provides a relevance feedback mechanism through which it learns multimodal COVID-19 representations for supporting clinical decisions. © © 2021 by SCITEPRESS – Science and Technology Publications, Lda. All rights reservedItem Machining Parameters Optimization of AA6061 Using Response Surface Methodology and Particle Swarm Optimization(SpringerOpen, 2018) Lmalghan, R.; Karthik, K.; Shettigar, A.; Rao, S.; Herbert, M.The influence of cutting parameters on the responses in face milling has been examined. Spindle speed, feed rate and depth of cut have been considered as the influential factors. In accordance with the design of experiments (DOE) a series of experiments have been carried out. The paper exemplifies on the optimizing the process parameters in milling through the application of Response surface methodology (RSM), RSM-based Particle Swarm Optimization (PSO) technique and Desirability approach. These aforesaid techniques have been applied to experimentally establish data of AA6061 aluminium material to study the effect of process parameters on the responses such as cutting force, surface roughness and power consumption. By adopting the multiple regression techniques, the interaction between the process parameters are acquired. The optimal parameters have been found by adopting the multi-response optimization techniques, i.e. desirability approach and PSO. The performance capability of PSO and desirability approach is investigated and found that the values obtained from PSO are comparable with the values of desirability approach. © 2018, Korean Society for Precision Engineering and Springer-Verlag GmbH Germany, part of Springer Nature.Item A deep neural network model for content-based medical image retrieval with multi-view classification(Springer Science and Business Media Deutschland GmbH, 2021) Karthik, K.; Kamath S?, S.S.In medical applications, retrieving similar images from repositories is most essential for supporting diagnostic imaging-based clinical analysis and decision support systems. However, this is a challenging task, due to the multi-modal and multi-dimensional nature of medical images. In practical scenarios, the availability of large and balanced datasets that can be used for developing intelligent systems for efficient medical image management is quite limited. Traditional models often fail to capture the latent characteristics of images and have achieved limited accuracy when applied to medical images. For addressing these issues, a deep neural network-based approach for view classification and content-based image retrieval is proposed and its application for efficient medical image retrieval is demonstrated. We also designed an approach for body part orientation view classification labels, intending to reduce the variance that occurs in different types of scans. The learned features are used first to predict class labels and later used to model the feature space for similarity computation for the retrieval task. The outcome of this approach is measured in terms of error score. When benchmarked against 12 state-of-the-art works, the model achieved the lowest error score of 132.45, with 9.62–63.14% improvement over other works, thus highlighting its suitability for real-world applications. © 2020, Springer-Verlag GmbH Germany, part of Springer Nature.Item Deep neural models for automated multi-task diagnostic scan management-quality enhancement, view classification and report generation(NLM (Medline), 2021) Karthik, K.; Kamath S?, S.The detailed physiological perspectives captured by medical imaging provides actionable insights to doctors to manage comprehensive care of patients. However, the quality of such diagnostic image modalities is often affected by mismanagement of the image capturing process by poorly trained technicians and older/poorly maintained imaging equipment. Further, a patient is often subjected to scanning at different orientations to capture the frontal, lateral and sagittal views of the affected areas. Due to the large volume of diagnostic scans performed at a modern hospital, adequate documentation of such additional perspectives is mostly overlooked, which is also an essential key element of quality diagnostic systems and predictive analytics systems. Another crucial challenge affecting effective medical image data management is that the diagnostic scans are essentially stored as unstructured data, lacking a well-defined processing methodology for enabling intelligent image data management for supporting applications like similar patient retrieval , automated disease prediction etc. One solution is to incorporate automated diagnostic image descriptions of the observation/findings by leveraging computer vision and natural language processing. In this work, we present multi-task neural models capable of addressing these critical challenges. We propose ESRGAN, an image enhancement technique for improving the quality and visualization of medical chest x-ray images, thereby substantially improving the potential for accurate diagnosis, automatic detection and region-of-interest segmentation. We also propose a CNN-based model called ViewNet for predicting the view orientation of the x-ray image and generating a medical report using Xception net, thus facilitating a robust medical image management system for intelligent diagnosis applications. Experimental results are demonstrated using standard metrics like BRISQUE, PIQE and BLEU scores, indicating that the proposed models achieved excellent performance. Further, the proposed deep learning approaches enable diagnosis in a lesser time and their hybrid architecture shows significant potential for supporting many intelligent diagnosis applications. © 2021 IOP Publishing Ltd.