Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
3 results
Search Results
Item Leveraging Deep Learning for Fever Temperature Analysis and Pattern Recognition(Institute of Electrical and Electronics Engineers Inc., 2024) Prabhakaran, A.; Sumam David, S.; Vijayasenan, D.; Mahabala, C.; Dakappa, P.Tympanic temperature is one of the most fundamental indicators for the diagnosis of diseases. Due to its importance, using patients' temperature data to aid in the diagnostic process would be beneficial. This work uses temperature data collected from various patients to classify diseases. We consider dengue, tuberculosis, and non-infectious and non-tubercular bacterial diseases. Extracting essential features from the temperature data is necessary so that the downstream layers only have to consider important features, not miscellaneous information. This feature extraction is done using two methods - Convolution Neural Networks and Autoencoders. We introduce three models for Explainable Temperature Analysis - ExTemp-Conv-SM, ExTemp-Conv-LG and ExTemp-Auto. We achieve a classification accuracy of 70% over these four disease classes. We also use explainable AI tools, like GradCAM, to identify distinguishing patterns in temperature fluctuations that can characterize diseases. We generate such patterns for all four diseases under consideration. We note that the patterns generated for dengue and tuberculosis match the findings in biological observation studies. We hope that the methods in this paper can be leveraged for other diseases and used to aid the diagnostic process. © 2024 IEEE.Item Language Detection in Overlapping Multilingual Speech: A Focus on Indian Languages(Institute of Electrical and Electronics Engineers Inc., 2025) Kolsur, A.A.; Prajwal, K.; Vijayasenan, D.The growing demand for technology capable of recognizing spoken languages and extracting information from real-world audio, especially in scenarios with overlapping speech, has become a significant focus of research due to its essential role in improving global connectivity and accessibility. In our paper, we focus on identifying languages present in audio files that consist of overlapping speech. We have focused our research particularly on Indian languages, as there is limited research on identifying low-resource languages in overlapping speech. In this paper, we have synthesized a custom dataset from the VoxLingua107 dataset due to the lack of overlapping Indian speech data. Further, we have developed a novel solution that first separates the overlapped audio using a speaker separation model and then uses a language recognition model to detect the languages present in the separated audio. We have compared the results obtained through our method with the current state-of-the-art model, Whisper, and concluded that our solution significantly outperforms the Whisper model. The results highlight the potential for significant improvements in multilingual communication systems and speech processing applications, paving the way for more inclusive and accurate language recognition technologies. © 2025 IEEE.Item Semantic segmentation of low magnification effusion cytology images: A semi-supervised approach(Elsevier Ltd, 2022) Aboobacker, S.; Vijayasenan, D.; Sumam David, S.; Suresh, P.K.; Sreeram, S.Cytopathologists examine microscopic images obtained at various magnifications to identify malignancy in effusions. They locate the malignant cell clusters at a low magnification and then zoom in to investigate cell-level features at a high magnification. This study predicts the malignancy at low magnification levels such as 4X and 10X in effusion cytology images to reduce scanning time. However, the most challenging problem is annotating the low magnification images, particularly the 4X images. This paper extends two semi-supervised learning (SSL) models, MixMatch and FixMatch, for semantic segmentation. The original FixMatch and MixMatch algorithms are designed for classification tasks. While performing image augmentation, the generated pseudo labels are spatially altered. We introduce reverse augmentation to compensate for the effect of the spatial alterations. The extended models are trained using labelled 10X and unlabelled 4X images. The average F-score of benign and malignant pixels on the predictions of 4X images is improved approximately by 9% for both Extended MixMatch and Extended FixMatch respectively compared with the baseline model. In the Extended MixMatch, 62% sub-regions of low magnification images are eliminated from scanning at a higher magnification, thereby saving scanning time. © 2022 Elsevier Ltd