SMSurya Majumder

Surya Majumder

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Surya Majumder has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • DTCA ‐Net: A Dilated Transformer Channel Attention With External Attention‐Based Network for Oral Cancer Segmentation

    by Shivam Singh, Surya Majumder, Akash Halder, Mahantapas Kundu, Ram Sarkar

    Accurate segmentation of oral cancer in histopathological images is essential for early diagnosis and treatment. While U‐Net‐based architectures have shown promise, balancing high accuracy with low computational complexity remains a key challenge. This work aims to design an efficient model that enhances feature extraction without significantly increasing model size. We propose DTCA‐Net, an enhanced U‐Net variant that integrates three key modules: (1) a Dilated Convolution and Transformer Channel Attention (DTCA) block to capture global and local features, (2) an Inverted External Attention (IEA) block to model intersample relationships and dataset‐level context, and (3) an Attention Bridge (AB) module connecting encoder and decoder to improve multiscale feature fusion. The model was evaluated on two public datasets: Oral Cavity‐Derived Cancer (OCDC) and ORal Cancer Annotated (ORCA). DTCA‐Net achieved strong performance, with a Dice score of 91.45% and IoU of 84.44% on the OCDC dataset. On the ORCA dataset, it obtained a Dice score of 83.66% and IoU of 71.96%, surpassing existing baselines in both accuracy and efficiency. We effectively combine lightweight architecture with rich feature extraction capabilities, making DTCA‐Net a promising tool for automatic oral cancer segmentation in clinical workflows. The entire codes of our model are publicly available through this Github Link: https://github.com/shivamsingh‐gpu/dtca‐net .


  • RAU-Net: Richard’s Curve Based Attention-Aided U-Net for Medical Image Segmentation

    by Surya Majumder, Akash Halder, Dmitrii Kaplun, Alexander Voznesensky, Ram Sarkar

  • Open Access

    SAWL-Net: A Statistical Attentions and Wavelet Aided Lightweight Network for Classification of Cancers in Histopathological Images

    by Surya Majumder, Aishik Paul, Friedhelm Schwenker, Ram Sarkar

    Addressing the formidable challenges posed by the diagnosis and management of various types of cancer, including breast, colon, lung, and colorectal cancer, demands innovative solutions to streamline histopathological analysis processes. In this study, we propose a novel lightweight convolutional neural network (CNN) called SAWL-Net (Statistical Attentions and Wavelet Aided Lightweight Network) architecture based on MobileNetV2 equipped with hybrid statistical similarity and wave format-aided attention mechanisms, specifically tailored to the demands of cancer histopathology. By leveraging the capabilities, our model incorporates a lightweight design while ensuring high-performance outcomes. We introduce a unique combination of Pearson Correlation Coefficient, Spearman Rank Correlation, and Cosine Similarity metrics, alongside a specialized wave conversion technique to enhance the detection of similarities across different channels of histopathological data, while providing a holistic approach to the model. In this study, we have considered breast, colorectal, and lung & colon cancer datasets for experimentation. Notably, our model surpasses prevailing state-of-the-art methodologies, showcasing its efficacy in optimizing diagnostic accuracy and expediting treatment strategies for varied cancer types. Our codes are publicly available at the GitHub repository.


  • Open Access

    A Fused Power Spectrum based Feature Selection to Identify Schizophrenia from EEG Signals using Deep Learning Models - An Experimental Study

    by Surya Majumder, Saikat Bandopadhyay, Sujay Saha, Anupam Ghosh

    The brain, a complex and important organ in the human body, is crucial for all our body processes. For the diagnosis and ongoing monitoring of a wide spectrum of brain disorders, accurate and early detection of the proper disorder from neurophysiological monitoring methods is essential. The importance of identification of disorders like Schizophrenia in clinical practice is examined in this research, along with the difficulties in attaining accurate results, particularly when working with small structures and precise details. A novel pre-processing methodology in this stream has been implemented for further feature and knowledge extraction and subsequent image generation. With their ability to automatically extract pertinent features from input images, CNN has made a significant advancement in the domain of image classification.This study presents and investigates in details the effect of our pre-processing on various well-known CNN based architectures. Various models like DenseNet, ResNet, MobileNet, NasNet, EfficientNet and ConvNext families along with Xception, InceptionV3 and InceptionResNetV2 models have been taken into consideration. These models have become optimal approaches to various classification tasks, each providing certain benefits and addressing particular difficulties. We have conducted this research on EEG data from a standard dataset, namely, IBIB PAN - Department of Methods of Brain Imaging and Functional Research of Nervous System dataset. This study presents a thorough review of the performance of different CNN based models and their variants on our preprocessed and generated images. On comparison with state-of-the-art results we have observed that using this approach, almost all our models have exceeded the same. Medical professionals and researchers can use the outcomes of these techniques for better diagnosis and treatment planning in the field of brain disorders. Our codes will be made available at: \href{https://github.com/SuryaMajumder/Brain-EEG-Signal-Analysis---Experimental-Study}{[Link]}

  • Open Access

    Unet based Xception Model Approach for Facial Emotion Recognition

    by Surya Majumder, Adrija Bhattacharya

    Facial emotion recognition is an important topic of study in artificial intelligence and computer vision. In this paper, we propose an automated system for facial emotion identification that uses a novel model architecture based on Xception backbone, with a U-Net based attention mechanism along with a dynamically changing parameterized tan function. MobileNet is widely used deep learning model known for its Depthwise-seperable convoloution. Our study seeks to present a novel model architecture that leverages the capabilities of Xception long with a U-Net based attention mechanism for global attention and a dynamically modified re-parameterized tangent function for local attention. The unsupervised segmentation done by the U-Net block and the attention generated by the said tangent function inside the Xception model is used as attention weights, thereby bringing new insights and advantages for improved facial emotion identification. This research looks at the importance of face expression recognition, which is important in human-computer interaction, affective computing, and a variety of applications in healthcare, entertainment, and security. Because of the intrinsic complexity and diversity of human face emotions, we needed to come up with creative solutions, which led us to this novel model approach. The proposed model is the fundamental contribution of this work, boosting its expressiveness for greater adaptability. Our technique effectively leverages different information by merging layer outputs and using novel attention mechanisms, resulting in higher facial emotion identification accuracy. The combination of complimentary data from diverse sources improves the system's accuracy and robustness. Experimental evaluations utilizing the Cohn-Kanade (CK+) dataset underscore the method's advantages, including improved accuracy and performance.

  • Open Access

    MENet: A Mitscherlich function based ensemble of CNN models to classify lung cancer using CT scans

    by Surya Majumder, Nandita Gautam, Abhishek Basu, Arup Sau, Zong Woo Geem, Ram Sarkar

    Lung cancer is one of the leading causes of cancer-related deaths worldwide. To reduce the mortality rate, early detection and proper treatment should be ensured. Computer-aided diagnosis methods analyze different modalities of medical images to increase diagnostic precision. In this paper, we propose an ensemble model, called the Mitscherlich function-based Ensemble Network (MENet), which combines the prediction probabilities obtained from three deep learning models, namely Xception, InceptionResNetV2, and MobileNetV2, to improve the accuracy of a lung cancer prediction model. The ensemble approach is based on the Mitscherlich function, which produces a fuzzy rank to combine the outputs of the said base classifiers. The proposed method is trained and tested on the two publicly available lung cancer datasets, namely Iraq-Oncology Teaching Hospital/National Center for Cancer Diseases (IQ-OTH/NCCD) and LIDC-IDRI, both of these are computed tomography (CT) scan datasets. The obtained results in terms of some standard metrics show that the proposed method performs better than state-of-the-art methods. The codes for the proposed work are available at https://github.com/SuryaMajumder/MENet.