Browsing by Author "Sumam David, S."

Now showing 1 - 20 of 68

A Deep Learning Approach toÂ Enhance Semantic Segmentation ofÂ Bacteria andÂ Pus Cells fromÂ Microscopic Urine Smear Images Using Synthetic Data
(Springer Science and Business Media Deutschland GmbH, 2024) Kanabur, V.R.; Vijayasenan, D.; Sumam David, S.; Govindan, S.
Urine smear analysis aids in preliminary diagnosis of Urinary Tract Infection. But it is time-consuming and requires a lot of medical expertise. Automating the process using machine learning can save time and effort. However obtaining a large medical dataset is difficult due to data privacy concerns and medical expertise requirements. In this study, we propose a method to synthesize a large dataset of gram-stained microscopic images containing pus cells and bacteria. We train a machine learning model to achieve semantic segmentation of bacteria and pus cells using this dataset. Later we use it to perform transfer learning on a relatively small dataset of gram stained urine microscopic images. Our approach improved the F1-score from 50% to 63% for bacteria segmentation and from 77% to 83% for pus cell segmentation. This method has the potential to improve the turn-around time and the quality of preliminary diagnosis of Urinary Tract Infection. Â© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
A Deep Learning Model for the Automatic Detection of Malignancy in Effusion Cytology
(Institute of Electrical and Electronics Engineers Inc., 2020) Aboobacker, S.; Vijayasenan, D.; Sumam David, S.; Suresh, P.K.; Sreeram, S.
The excessive accumulation of fluid between layers of pleura covering lungs is known as pleural effusion. Pleural effusion may be due to various infections, inflammations or malignancy. The cytologists visually examine the microscopic slide to detect the malignant cells. The process is time-consuming, and interpretation of reactive cells and cells with ambiguous levels of atypia may differ between pathologists. Considerable research is happening towards the automation of fluid cytology reporting. We propose an integrated approach based on deep learning, where the network learns directly to detect the malignant cells in effusion cytology images. Architecture U-Net is used to learn the malignant and benign cells from the images and to detect the images that contain malignant cells. The model gives a precision of 0.96, recall of 0.96, and specificity of 0.97. The AUC of the ROC curve is 0.97. The model can be used as a screening tool and has a malignant cell detection rate of 0.96 with a low false alarm rate of 0.03. Â© 2020 IEEE.
A hybrid CNN-FC approach for automatic grading of brain tumors from non-invasive MRIs
(Institute of Electrical and Electronics Engineers Inc., 2024) Bhaskaracharya, B.; Nair, R.P.; Prakashini, K.; Girish Menon, R.; Litvak, P.; Mandava, P.; Vijayasenan, D.; Sumam David, S.
The grading of brain tumors is essential in treatment planning to effectively control the tumor growth and reduce the associated symptoms. Appropriate treatment planning might help in improving the quality of life and patient life span. Gliomas are indeed the most common type of brain tumor, originating from glial cells. Low-grade gliomas (grades 1 or 2) are typically slow-growing, less invasive, and may be suitable for surgical resection or targeted therapies. On the other hand, higher-grade tumors such as grades 3 or 4 are more aggressive, it might infiltrate the surrounding brain tissue making complete resection challenging. In clinical diagnosis, traditionally tumor grading requires the procedure of resecting a part of the tumor for microscopic examination. To address this, a method to grade the tumor non-invasively using MRIs is proposed. Our work utilized the BraTS2018 dataset to segment the substructure of brain tumors that includes necrosis and non-enhancing, edema, and enhancing regions. These regions are then used to train the proposed grading model. Furthermore, we evaluated the performance of our model on a tertiary hospital dataset consisting of 69 samples. The accuracy scores obtained on the BraTS2018 test sample and tertiary hospital dataset are 0.87 and, 0.85 respectively. This consistent score on both public and tertiary hospital datasets indicates a reliable and stable performance of the model. Â© 2024 IEEE.
A more generalizable DNN based Automatic Segmentation of Brain Tumors from Multimodal low-resolution 2D MRI
(Institute of Electrical and Electronics Engineers Inc., 2021) Bhaskaracharya, B.; Nair, R.P.; Prakashini, K.; Girish Menon, R.; Litvak, P.; Mandava, P.; Vijayasenan, D.; Sumam David, S.
In the field of Neuro-oncology, there is a need for improved diagnosis and prognosis of brain tumors. Brain tumor segmentation is important for treatment planning and assessing the treatment outcomes. Manual segmentation of brain tumors is tedious, time-consuming, and subjective. In this work, an efficient encoder-decoder based architectures were implemented for automatic segmentation of brain tumors from low resolution 2D images. Ensemble of the multiple architectures (EMMA) improves the performance of the brain tumor segmentation. Furthermore, the computational requirements of the proposed models are lower than that of BraTS-challenge methods. The average Fl-scores on the BraTS-challenge validation dataset for Tumor Core, Whole Tumor, and Enhancing Tumor are 0.82, 0.87, and 0.78, respectively. The average Fl-scores on the KMC-Manipal dataset for TC, WT, and ET are 0.74, 0.82, and 0.68 respectively. Â© 2021 IEEE.
A novel adaptive support window based stereo matching algorithm for 3D reconstruction from 2D images
(2011) Iveel, J.; Sumam David, S.
Three dimensional scene reconstruction, sometimes referred as view synthesis, is a problem in the area of Stereo Vision, which is the most widely used method for gathering depth information from 2D scenes. Stereo vision finds many applications in automated systems such robotics, tracking object in 3D space and constructing a 3D spatial model of a scene. There are many human-machine control applications, such vision based remote control system which exploits stereo vision to control the machine in a touch-free environment. In this paper, we present a new stereo vision matching algorithm using an adaptive support window. In many area based algorithms, the selection and computation of the size and shape of the window is the most crucial factor for obtaining high quality disparity map. We developed an adaptive support window based stereo matching using similar color regions selected by the seed growing algorithm. The proposed approach is tested on Middlebury stereo images and results were promising. Â© 2011 IEEE.
A novel vector-locked loop scheme for synchronized extraction of harmonics
(2010) Bhat, R.A.; Gonda, J.M.; Sumam David, S.
The separation of the fundamental from a distorted waveform is an important component in the implementation of a custom power device. A novel scheme for a vector-locked loop (VLL) for synchronous extraction of harmonics/fundamental in a distorted periodic waveform is described here. An intuitive, though approximate, explanation is provided for the operation of the algorithm, focusing on the locking process and the filtering capability. Performance aspects are analysed and an approximate linear time-invariant model is presented to facilitate in the design. Moreover, a means for choosing design parameters to achieve the optimum performance is provided. Main features of the proposed VLL are its simplicity, excellent insensitivity to harmonics, noise rejection, availability of both fundamental and harmonics without additional processing, and the speed of operation. The claims are verified through extensive simulation studies in MATLABÂ®/SimulinkÂ®. Â© 2010 IEEE.
A system level solution for power quality problems
(2005) Padiyar U, U.H.; Mahant-Shetti, S.S.M.; Sumam David, S.
The power quality deterioration is still an unsolved problem due to the increased power electronic load getting into the system. Every, consumer friendly equipment is power electronic controlled in one-way or other. The current drawn by all of these equipments are generally non-sinusoidal in nature, and are distorting the voltage wave as well. Many solutions proposed so far, though found useful in some cases have never become a system solution for the power quality deterioration problem. This paper proposes a unique innovative solution to these problems by a correction at the generation level by changing the shape of source wave keeping in view the nature of new loads. This paper presents PSPICEÂ® simulated results to justify the need for source wave shape change, which will act as a system level solution to the limit peak currents drawn by the new generation power electronic loads. Â© 2006 IEEE.
A system level solution to improve VRM efficiency
(2009) Padiyar U, H.; Sumam David, S.; Mahant-Shetti, S.S.M.
An investigation of the synchronous buck regulator working as a Voltage regulator module (VRM) for Processor power supply applications, under light load condition is presented. VRM, a very special case of power supply belongs to switched mode power supply family. Important design specifications include low output voltage with extremely small tolerance band, high load current of large slew rates, small foot print, low cost etc. Higher operating efficiencies are desired to enable extended battery mode operation in mobile platforms in addition to reducing heat dissipation. This paper identifies the source of energy inefficiency and suggests a variable frequency approach as a cost effective system level solution, to enhance the efficiency of the converter supplying widely varying load with special emphasis to light load operation.
Accelerating real-time computer vision applications using HW/SW co-design
(Institute of Electrical and Electronics Engineers Inc., 2017) Anirudh, B.K.; Venkatraman, V.; Kumar, A.R.; Sumam David, S.
Video processing applications have become increasingly difficult to implement on hardware, owing to the complex computer vision algorithms involved. This paper presents a real-time video processing architecture based on hardware/software co-design that improves execution speed and reduces the time to market of applications. We have implemented this framework for handwritten digit recognition on the Zybo Zynq-7000 ARM/FPGA SoC using Vivado High Level Synthesis (HLS) and Xillybus tools. Histogram of Oriented Gradients (HOG) feature extraction algorithm has been optimised for hardware execution and acceleration techniques have been applied on Vivado HLS to achieve a speed up of 38.89 for the HOG algorithm and recognition accuracy of 95.6%. Low precision arithmetic along with our approximations for costly functions, produced this significant gain in throughput by reducing 90% of the hardware resources required with just a marginal accuracy reduction by 1%. An overall performance improvement of 77% is obtained through hardware/software co-design over software execution. The framework identified digits seamlessly in a real-time video stream at 30 frames per second and enabled high frame rate video processing. Â© 2017 IEEE.
Adaptive local cosine transform for seismic image compression
(2006) Aparna., P.; Sumam David, S.
A typical seismic analysis involves collection of data by an array of seismometers, transmission over a narrow-band channel, and storage of data for analysis. Transmission and archiving of large volumes of data involves great cost. Hence there is a need to devise suitable methods for compressing the seismic data without compromising on the quality of the reconstructed signal. This paper presents our work on the seismic data compression based on adaptive local cosine transform and its associated multi-resolution and best-basis methodology and compares the results with wavelet based implementation. Â© 2006 IEEE.
Adaptive Reconfigurable Architecture for Image Denoising
(Institute of Electrical and Electronics Engineers Inc., 2015) Hegde, K.V.; Kulkarni, V.; Harshavardhan, R.; Sumam David, S.
In this paper, we propose an adaptive reconfigurable architecture for image denoising. First part of this paper outlines an efficient noise detection hardware for Gaussian & impulse noise detection and suitable filters for denoising. With a robust noise detection method including a novel Gaussian noise detection method, we also explore the dynamic detection of noise in an image giving adaptability to the architecture for a better quality of denoising. Proposed architecture includes a decision making unit to find out the presence of noise as well as type of the noise, based on which a suitable filter is employed during run-time. An onboard microprocessor controls the reconfiguration and dataflow. Proposed architecture is tested on Xilinx Virtex-6 FPGA with localized noise and mixed noise conditions and it gives superior performance compared to the standard filters used. High quality denoising is achieved with simple filters on a reconfigurable region utilizing smaller area and lesser hardware resources. Â© 2015 IEEE.
Address generation for DSP Kernels
(2011) Ramesh Kini, M.; Sumam David, S.
Performance of Signal Processing Algorithms implemented in hardware depend on efficiency of datapath, memory speed, and address computation. Pattern of data access in signal processing applications is complex and it is desirable to execute the innermost loop of a kernel every clock. This demands generation of typically three addresses per clock: two addresses for data sample/coefficient and one for storage of processed data. Presence of a set of dedicated, efficient Address Generator Units (AGU) helps in better utilization of the datapath elements by using them only for kernel operations; and will certainly enhance the performance. This paper focuses on design and implementation of Comprehensive Address Generator Unit (CAGU) for complex addressing modes required by DSP Kernels used in Multimedia Signal Processing. An 8 bit CAGU has been implemented using UMC 0.18 micron, 6 metal layers process, that occupies 21802 sq microns, consuming 2.95 mW and works with a clock period of 6 ns. Â© 2011 IEEE.
Adversarial Learning Based Semi-supervised Semantic Segmentation ofÂ Low Resolution Gram Stained Microscopic Images
(Springer Science and Business Media Deutschland GmbH, 2024) Singh, H.; Kanabur, V.R.; Sumam David, S.; Vijayasenan, D.; Govindan, S.
Urinary tract infections (UTIs) are infections that affect the urinary system. It is usually caused by bacteria and pus cells. Analyzing urine samples, including examining pus cells, is a standard method for diagnosing and monitoring UTIs. However, manually detecting bacteria or pus cells in microscopic urine images is a time-consuming and labour-intensive task for microbiologists. Therefore, the segmentation of microscopic pus cell images will ease the process of detecting UTI. Especially low resolution microscopic images are hard to annotate; therefore, in this study, we propose an adversarial learning based semi-supervised segmentation method for segmentation of pus cell images at low resolution i.e. 40Ã— using labeled high resolution images i.e. 100Ã—. The proposed methodology aims to ease the process of UTI detection by automating the segmentation of pus cell images. The results of the proposed methodology demonstrate an increase in the Dice coefficient score percentage by 1%, 1.6% and 2.4% on 40Ã— images when compared to fully supervised segmentation model trained on only 100Ã— data using three different architectures- Unet, ResUnet++, and PSPnet, respectively. Â© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
An Approach for Diagnostically Lossless Coding of Volumetric Medical Data Based on Wavelet and Just-Noticeable-Distortion Model
(Taylor and Francis Ltd., 2023) Chandrika, B.K.; Aparna., P.; Sumam David, S.
This paper explores a technique for visually/diagnostically lossless coding in the wavelet domain to effectively compress the three-dimensional medical image data. The quantisation module based on Just Noticeable Distortion (JND) for wavelets guarantees the visual quality in the reconstructed data. This method has been further extended to present the Volume of Interest (VOI) based technique that enables to preserve the quality of the diagnostically significant VOI region. The proposed method tested on several datasets outperforms the state-of-the-art methods. Apart from the conventional quality metric, Human Visual System (HVS) based quality metrics are also used to evaluate the visual quality of the reconstructed image. A subjective and objective evaluation carried out for VOI based coder shows that the quality-compression needs of the medical community are well addressed. © 2023 IETE.
An Integrated Deep Learning Approach towards Automatic Evaluation of Ki-67 Labeling Index
(Institute of Electrical and Electronics Engineers Inc., 2019) Lakshmi, S.; Vijayasenan, D.; Sumam David, S.; Sreeram, S.; Suresh, P.K.
Ki-67 labeling index is a widely used biomarker for the diagnosis and monitoring of cancer. Many automated techniques have been proposed for evaluating Ki-67 index. In this paper, we introduce an integrated deep learning based approach. We use MobileUnet model for segmentation and classification and connected component based algorithm for the estimation of Ki-67 index in bladder cancer cases. The average F1 score is 0.92 and dice score is 0.96. The mean absolute error in the evaluated Ki-67 index is 2.1. We also explore possible pre-processing steps to generalize the segmentation model to at least one another type of cancer. Histogram matching and re-sizing improve the performance in breast cancer data by 12% in F1 score and 8% in dice score. Â© 2019 IEEE.
Area and power optimised ASIC implementation of adaptive beamformer for hearing AIDS
(Institute of Electrical and Electronics Engineers Inc., 2017) Samtani, K.; Thomas, J.; Deepu, S.P.; Sumam David, S.
Beamforming is a technique used in hearing AIDS to improve the intelligibility of target sound by reducing the interference from other directions. An efficient ASIC implementation of a two omnidirectional microphone array based adaptive beamforming algorithm is presented in this paper with various optimisations proposed at different stages of the hardware design. The beamform patterns and improvements in SNR values obtained from experiments conducted in a conference room environment were analysed to verify the working of the design. The architecture was implemented with 0.18 Î¼m standard cell libraries. Cell area and power reports were analysed for different optimisations. The final area and power obtained are 0.054 mm2 and 60.54 Î¼W respectively. Â© 2017 IEEE.
Artery Vein Segmentation in Handheld Fundus Camera Retinal Images and leveraging Cross Entropy for improved Semantic performance
(Institute of Electrical and Electronics Engineers Inc., 2024) Yohannan, R.P.; Sumam David, S.; Vijayasenan, D.; Chowdary, R.T.; Girish Menon, R.; Menon, S.G.
The segmentation of retinal vessels into arteries and veins in retinal images is a crucial task for analysing the vascular changes with respect to many diseases that manifest ocular symptoms. But most existing research has concentrated on fundus images acquired using tabletop cameras and not much has been studied on images captured by handheld cameras. Such cameras are particularly useful for examining bedridden patients, especially those who may have conditions such as hypertension or diabetes that can affect the retina, since they are portable and can be easily maneuvered by healthcare providers, allowing them to perform retinal examinations conveniently at the patient's bedside. This paper presents an approach to segment such images and assesses the impact of data augmentation on model performance. It further presents a method to compute pixel level weights during training, that allows for fine-grained adjustment of the loss function. Â© 2024 IEEE.
Audio segmentation using a priori information in the context of Karnatic Music
(Institute of Electrical and Electronics Engineers Inc., 2015) Kalyan, V.A.; Sankaranarayanan, S.; Sumam David, S.
Karnatic Music (KM) is distinct because of the prevalence of gamaka - embellishments to musical notes in the form of frequency traversals. Another important aspect of KM is that the performance style is mostly extempore. Hence, Music Information Retrieval (MIR) tasks in the context of KM are highly challenging. This paper deals with the task of Audio Segmentation and its application to MIR challenges of KM at various levels. This work presents a method that incorporates a priori knowledge about the music system and the audio track at hand for segmenting the audio into its constituent notes. The method uses amplitude and energy based features to train a neural network and an accuracy of 95.2% has been achieved on KM audio samples. The paper also elucidates the application of the method to important MIR tasks such as Music Transcription and Score-Alignment in the context of KM. Â© 2015 IEEE.
Automatic speech recognition using audio visual cues
(2004) Yashwanth, H.; Mahendrakar, H.; Sumam David, S.
Automatic speech recognition (ASR) systems have been able to gain much popularity since many multimedia applications require robust speech recognition algorithms. The use of audio and visual information in the speaker-independent continuous speech recognition process makes the performance of the system better compared to the ones with only the audio information. There has been a marked increase in the recognition rates by the use of visual data to aid the audio data available. This is due to the fact that video information is less susceptible to ambient noise than audio information. In this paper a robust Audio-Video Speech Recognition (AVSR) system that allows us to incorporate the Coupled Hidden Markov Model (CHMM) model for fusion of audio and video modalities is presented. The application records the input data and recognizes the isolated words in the input file over a wide range of Signal to Noise Ratio (SNR.) The experimental results show a remarkable increase of about 10% in the recognition rate in the AVSR compared to the audio only ASR and 20% compared to the video only ASR for an SNR of 5dB. Â©2004 IEEE.
Characterization of Fault Signature Due to Combined Air-Gap Eccentricity and Rotor Faults in Induction Motors
(Praise Worthy Prize S.r.l, 2021) Bindu, S.; Sumam David, S.; Thomas, V.V.
An accurate means of non-invasive condition monitoring of the popular industrial drive, three-phase squirrel-cage induction motor, can help to avoid unscheduled maintenance downtime and loss. Faults like air-gap eccentricity can exist even in a newly assembled drive and hence may co-exist with other internal defects. Despite it being a possible situation, the occurrence of simultaneous faults has seldom been studied. Therefore, there is a need for identifying fault signatures of combined fault conditions in a non-invasive manner. This paper presents a detailed model-based study on a three-phase squirrel-cage induction motor with the simultaneous existence of broken rotor-bar and air-gap mixed eccentricity faults using spectral analysis of stator current, instantaneous power, and estimated air-gap torque signals. The modelling of the machine is done using the Multiple Coupled Circuit method and modified to model the presence of the combined fault conditions. A comparative evaluation with various fault conditions and their severity is carried out by spectral analysis, and unique slip-dependent frequency components are identified in the spectra of diagnostic signals. This fault characterization is the most significant contribution of this paper. © 2021 Praise Worthy Prize S.r.l.-All rights reserved.