Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
4 results
Search Results
Item Osteosarcoma Bone Cancer Detection(Springer Science and Business Media Deutschland GmbH, 2025) Payani, C.A.; Gupta, C.; Vamsidhar, K.; Bhat, P.; Patil, N.Osteosarcoma is a type of bone cancer commonly found in the elongated bones found in the upper and lower limbs. The precise cause is unknown, but experts believe it’s linked to changes in the DNA of the bones, resulting in the growth of abnormal and harmful bone tissue. If caught early, osteosarcoma is treatable, with about 75% people cured when the cancer hasn’t spread to other body parts. Analyzing biopsy samples can be time-consuming, but there are advanced computer programs, known as supervised deep learning methods, that can help speed up the process and enhance the efficiency of the diagnosis. Previous studies have already evaluated the performance of deep learning models such as VGG16, VGG19, DenseNet201, and ResNet101, among which ResNet101 performed better with 90.36% accuracy. When it comes to understanding complex image features, previous models lack behind. We propose EfficientNetV2, Xception, and InceptionV3 models, among which Xception outperformed other models with 94.5% accuracy on the image dataset. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.Item Automated Colorization of Grayscale Images Using Superpixels and K-Means Clustering(Springer Science and Business Media Deutschland GmbH, 2025) Kulkarni, B.C.; Teja, B.; Hegde, A.R.; Bhat, P.; Patil, N.The process of transforming grayscale photos into aesthetically pleasing color images is called colorization. Convincing the audience of the realism of the outcome is the primary objective of colorization. Natural scenery makes up the majority of the grayscale photographs that require colorization. A broad range of colorization techniques have been created over the past 20 years; these vary from algorithmically basic procedures that need time and energy due to inevitable human participation to more complex ways that are also more automated. The complex field of automatic conversion mixes deep learning, machine learning, and art. Most of the earlier works which use deep learning, use every pixel values to train the models which is computationally expensive. We present a methodology for colorizing grayscale images using convolutional neural network (CNN), our method uses a combination of superpixel segmentation and K-Means clustering to significantly reduce number of pixel values. The process begins with the conversion of grayscale images to superpixels, which are perceptually uniform regions that aid in efficient colorization. Subsequently, K-Means clustering is applied within each superpixel to identify dominant color clusters, followed by quantization of color information to simplify representation. The prepared input, comprising grayscale images and quantized color information, is then fed into a CNN for colorization, leveraging spatial coherence and semantic context to predict plausible colors for grayscale pixels. The proposed methodology is evaluated on a diverse set of grayscale images, demonstrating its effectiveness in producing vibrant and visually appealing colorized outputs. Through experiments and analysis, we showcase the potential applications and benefits of the proposed approach in historical photograph restoration, movie colorization, and other domains requiring accurate and efficient grayscale image colorization. We use SSIM and PSNR as our evaluation metrics. SSIM is calculated based on the similarity of the luminance and brightness values of the considered and obtained rgb images for the grayscale images, and PSNR is calculated using Mean Squared Error (MSE) of the peak signal values within images. Our methodology’s SSIM and PSNR for the considered flower class is 81.5 and 25.6, respectively. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.Item Detection of heart arrhythmia with electrocardiography(Springer, 2024) Jat, T.; Patil, N.; Bhat, P.Early detection of cardiac arrhythmia, a prevalent form of cardiovascular disease (CVD) impacting millions globally, is heavily reliant on the accurate analysis of heartbeats. Physicians often recommend that patients wear Holter monitors for 24 h or longer to observe concerning cardiac issues, resulting in the collection of substantial amounts of electrocardiogram (ECG) data. Consequently, there is a need to automate the process of interpreting ECGs to detect cardiac abnormalities efficiently. Current state-of-the-art studies rely on handcrafted feature extraction, which may not effectively capture the intricate temporal relationships inherent in ECG signal data. To address this limitation and facilitate the diagnosis of cardiac diseases, this study proposes a technique that converts electrocardiogram signals into images, subsequently training a deep learning model on the generated images. Image encoding techniques such as Gramian Angular Difference Field (GADF), Gramian Angular Summation Field (GASF) and Markov Transition Field (MTF) are employed to translate the ECG signals into images. The highest accuracy, 96.71%, was achieved by training the Convolutional Neural Network (CNN) model using the concatenation of these three image encoding techniques. The proposed approach is assessed using ECG recordings from the MIT-BIH Arrhythmia Database to detect heart arrhythmia, demonstrating the efficacy of the approach. © The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2024.Item Class-Balanced Protein Interaction Site Prediction Using Global and Local Features with XGBoost and Deep Learning(Springer, 2025) Kulkarni, B.C.; Sai, B.S.; Kolagad, V.; Patil, N.; Bhat, P.Inter-protein interactions are critical in biological pathways. Determining the protein–protein interaction (PPI) sites is vital for comprehending protein behavior and designing medications. Traditional experimental protocols for pinpointing these sites are prolonged and costly, making computational approaches an efficient alternative. However, many computational methods fail to resolve the problem of class imbalance in PPI datasets and focus predominantly on local contextual features, ignoring global sequence information. In this work, we address class imbalance in PPI site prediction by applying a series of balancing techniques: selective thinning of the majority class, Tomek Links to remove noisy samples near the class boundary, and random augmentation of the minority class. We then further balance the data using Synthetic Minority Over-sampling Technique (SMOTE) and Generative Adversarial Networks (GANs), with GANs showing a slight edge in improving data quality and reducing noise. Our approach incorporates four key features: secondary structure, raw protein sequence, Position-Specific Scoring Matrix (PSSM), and Relative Solvent Accessibility (RSA). We use both nearby contextual and holistic sequence features for training two models: XGBoost and a Deep Neural Network (DNN). The performance of the models was assessed using accuracy, Matthews correlation coefficient (MCC), precision, recall, and F-score. We correlate the impact of using balanced versus unbalanced datasets and measure the share of global features in enhancing model performance. The findings demonstrate that class balancing significantly upgrades prediction performance. The XGBoost model realized an accuracy of 0.831 and precision of 0.417, outperforming the DNN in these metrics. The DNN model attained a higher recall of 0.723 and an F-score of 0.485, exemplifying its effectiveness in accurately detecting true PPI sites. Both models showcased a good MCC of 0.30, corroborating the effectiveness of the introduced balancing strategies and the assimilation of global features in robust PPI site prediction. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.