Vipin Gupta

Background:
Determining the DNA fragmentation index (DFI) by the sperm chromatin dispersion (SCD) test involves manual counting of stained sperms with halo and no halo.

Aims:
The aim of this study is to build a robust artificial intelligence-based solution to predict the DFI.

Settings and Design:
This is a retrospective experimental study conducted in a secondary in vitro fertilisation setup.

Materials and Methods:
We obtained 24,415 images from 30 patients… Show more

Background:
Determining the DNA fragmentation index (DFI) by the sperm chromatin dispersion (SCD) test involves manual counting of stained sperms with halo and no halo.

Aims:
The aim of this study is to build a robust artificial intelligence-based solution to predict the DFI.

Settings and Design:
This is a retrospective experimental study conducted in a secondary in vitro fertilisation setup.

Materials and Methods:
We obtained 24,415 images from 30 patients after the SCD test using a phase-contrast microscope. We classified the dataset into two, binary (halo/no halo) and multiclass (big/medium/small halo/degraded (DEG)/dust). Our approach consists of a training and prediction phase. The 30 patients' images were divided into training (24) and prediction (6) sets. A pre-processing method M was developed to automatically segment the images to detect sperm-like regions and was annotated by three embryologists.

Statistical Analysis Used:
To interpret the findings, the precision-recall curve and F1 score were utilised.

Results:
Binary and multiclass datasets containing 8887 and 15,528 cropped sperm image regions showed an accuracy of 80.15% versus 75.25%. A precision-recall curve was determined and the binary and multiclass datasets obtained an F1 score of 0.81 versus 0.72. A confusion matrix was applied for predicted and actuals for the multiclass approach where small halo and medium halo confusion were found to be highest.

Conclusion:
Our proposed machine learning model can standardise and aid in arriving at accurate results without using expensive software. It provides accurate information about healthy and DEG sperms in a given sample, thereby attaining better clinical outcomes. The binary approach performed better with our model than the multiclass approach. However, the multiclass approach can highlight the distribution of fragmented and non-fragmented sperms. Show less

Medical images taken from camera based devices (e.g. laparoscope, colposcope, retinoscope, etc) are greatly affected by numerous bright reflection spots (called glare or specular reflections). This may affect the visibility of the abnormal features (if present in the glare locations). We have developed a novel solution to overcome this problem by incorporating a multi-LED lighting solution. This will intelligently and rapidly switch on and off the LED’s in a pattern that dynamically and… Show more

Medical images taken from camera based devices (e.g. laparoscope, colposcope, retinoscope, etc) are greatly affected by numerous bright reflection spots (called glare or specular reflections). This may affect the visibility of the abnormal features (if present in the glare locations). We have developed a novel solution to overcome this problem by incorporating a multi-LED lighting solution. This will intelligently and rapidly switch on and off the LED’s in a pattern that dynamically and geometrically shifts/shuffles these glare spots back and forth in the image such that every glare-affected area of a single image frame can be reconstructed from a few adjacent time-frame images. We have built the prototype that successfully demonstrates how the glare problem in the medical video/image can be satisfactorily solved, significantly enhancing the accuracy of this vital procedure in the diagnosis of diseases. We achieve 65 -95% reduction in specularity on phantom model using the proposed approach. Show less

In this paper we propose a novel approach based on multi-stage random forests to address problems faced by traditional vessel segmentation algorithms on account of image artifacts such as stitches organ shadows etc.. Our approach consists of collecting a very large number of training data consisting of positive and negative examples of valid seed points. The method makes use of a 14x14 window around a putative seed point. For this window three types of feature vectors are computed viz… Show more

In this paper we propose a novel approach based on multi-stage random forests to address problems faced by traditional vessel segmentation algorithms on account of image artifacts such as stitches organ shadows etc.. Our approach consists of collecting a very large number of training data consisting of positive and negative examples of valid seed points. The method makes use of a 14x14 window around a putative seed point. For this window three types of feature vectors are computed viz. vesselness, eigenvalue and a novel effective margin feature. A random forest RF is trained for each of the feature vectors. At run time the three RFs are applied in succession to a putative seed point generated by a naiive vessel detection algorithm based on vesselness. Our approach will prune this set of putative seed points to correctly identify true seed points thereby avoiding false positives. We demonstrate the effectiveness of our algorithm on a large dataset of angio images. Show less

In this paper, segmentation of coronary arteries of heart is done. The source data is filtered using two dimensional DICOM (Digital imaging & Communications in Medicine) images taken from the 64 Slice Computed tomography Scan Data of the heart, using a filter to highlight only the coronary arteries regions. The work has been implemented using an Intel (R) Core (TM) i3 at 2.20GHz using MATLAB version 7.8.0.347. Segmentation of coronary arteries of heart has been done on 2-Dimensional CT… Show more

In this paper, segmentation of coronary arteries of heart is done. The source data is filtered using two dimensional DICOM (Digital imaging & Communications in Medicine) images taken from the 64 Slice Computed tomography Scan Data of the heart, using a filter to highlight only the coronary arteries regions. The work has been implemented using an Intel (R) Core (TM) i3 at 2.20GHz using MATLAB version 7.8.0.347. Segmentation of coronary arteries of heart has been done on 2-Dimensional CT (Computed Tomography) isolated data. The system has been developed using different image processing functions in MATLAB software. Show less

In this paper, we propose a low cost, robust vision based system for monitoring patient movements during stereotactic radiotherapy: the Monocular Visual Patient Movement Monitoring (MVPM) System. The system consists of a light weight mouth bite with dual fiducial system used for position tracking. One set of fiducials consist of easy-to-detect checkerboard whose 3D position can be tracked using a pre-calibrated off-the-shelf camera and a set of radio-opaque rods suitable for X-ray imaging. We… Show more

In this paper, we propose a low cost, robust vision based system for monitoring patient movements during stereotactic radiotherapy: the Monocular Visual Patient Movement Monitoring (MVPM) System. The system consists of a light weight mouth bite with dual fiducial system used for position tracking. One set of fiducials consist of easy-to-detect checkerboard whose 3D position can be tracked using a pre-calibrated off-the-shelf camera and a set of radio-opaque rods suitable for X-ray imaging. We describe the associated workflow steps for using the system during planning and treatment phases.
Also we report experimental results for tracking accuracy using a
in-house phantom. Show less

The segmentation of images is an every day needed process, especially for clinicians and doctors. Because of the fast increase of possibilities to get image data, more and more image data is available for analysis. Clinicians and doctors often have to focus on important parts of the image, which leads to the need of segmentation. Segmentation is defined as the allocation of an image into a set of distinct and meaningful regions. These regions contain pixels. In each region the pixels have the… Show more

The segmentation of images is an every day needed process, especially for clinicians and doctors. Because of the fast increase of possibilities to get image data, more and more image data is available for analysis. Clinicians and doctors often have to focus on important parts of the image, which leads to the need of segmentation. Segmentation is defined as the allocation of an image into a set of distinct and meaningful regions. These regions contain pixels. In each region the pixels have the same attributes such as intensity, texture, etc. The regions are used to compute important properties of the contained object. Such properties are for example height, width, but also blood volume, etc. Speed, accuracy and robustness are three main goals of every segmentation algorithm. Today different approaches for fast image segmentation exist already. One solution is the CGC (Combinatorial Graph Cuts) framework. It is a fast algorithm which optimizes boundary and regional properties. This algorithm achieves higher speed but is limited to a specific set of applications. The CGC is especially used for graph constructions and max-flow/min-cut computation. The max-flow computation is necessary, because just by knowing the max flow through the graph segmentation is possible. For the max-flow/min-cut two methods based on augmenting path and push-relabel are available. The CGC is an augmenting path based method. Two search trees are built, one from the source and one from the sink. Both trees grow simultaneously to find an augmenting path. It is followed by a stage, in which the allowable flow is pushed and the residual graph is computed. Show less

In early stages of breast cancer surgery can be employed to remove the tumor. To ensure that the tumor is eradicated, radiation is applied subsequently. To avoid the removal of the breast, a radiation treatment can also be given primarily. Fudicials are implanted to keep track of the tumor position. If the tumor is not superficial, it is essential to monitor its movement, since a wrongly positioned beam can affect the chest walls or other critical organs. If the position of the tumor can be… Show more

In early stages of breast cancer surgery can be employed to remove the tumor. To ensure that the tumor is eradicated, radiation is applied subsequently. To avoid the removal of the breast, a radiation treatment can also be given primarily. Fudicials are implanted to keep track of the tumor position. If the tumor is not superficial, it is essential to monitor its movement, since a wrongly positioned beam can affect the chest walls or other critical organs. If the position of the tumor can be tracked down accurately, the planned treatment region can be revised. Furthermore, it is necessary to stop the treatment immediately if the tumor moves beyond a certain unacceptable distance, e.g. due of coughing or discomfort of the patient. The challenge in tracking breast tumors during IMRT (Intensity-Modulated Radiation Therapy) is that the movement of the skin, which can be monitored by visual cameras, may not completely reflect the movement of the tumor. A significant movement of breast tumors is given by respiration. Show less

Transliteration system for mobile phone is an area that is always in demand given the difficulties and constraints we face in its implementation. In this paper we deal with automatic transliteration system for Kannada which has a non uniform geometry and inter-character spacing unlike non-oriental language text like English. So it is even more a challenging problem. Working model consists of part of the process taking place on a mobile with remaining on a server. Good results are achieved

During radiotherapy of cancer cells the tumor has to be tracked as it changes its position because of the patient's breathing cycle and minor movements. Various techniques to track tumor motion have been developed and applied before. Earliest approaches dealing with this problem basically boil down to asking the patient to be steady either through abdominal compression, shallow breathing or holding the breath. These approaches require patient compliance, provide not much reliability and require… Show more

During radiotherapy of cancer cells the tumor has to be tracked as it changes its position because of the patient's breathing cycle and minor movements. Various techniques to track tumor motion have been developed and applied before. Earliest approaches dealing with this problem basically boil down to asking the patient to be steady either through abdominal compression, shallow breathing or holding the breath. These approaches require patient compliance, provide not much reliability and require active participation by the therapist to monitor any deviations. More recently, in gated radiation therapy where the radiotherapy beam is turned on at the appropriate breathing phase has been applied to solve this technical problem. Typically this strategy involves taking the CT (Computer Tomography) image of the patient at the end of the inhalation or exhalation cycle. For example, before or after an object is placed on the stomach of the patient and its height is tracked. When the height matches the position at the desired phase, the beam is turned on for a specific duration. However, this technique increases the treatment time significantly and measurement errors decrease the accuracy. Other approaches for gated radiotherapy based on two X-rays, the utilization of opto-electronic devices or the attachment of markers are in development. Show less

The present work provides an efficient way of segmenting Left ventricular (LV) cavity and Myocardium (MC) for the cardiac analysis in a clinical environment. It is a fully automatic method based on multi-step binarization and adaptive threshold correction. It offers significant speed advantage over currently available methods. In addition, we provide an option to include/exclude Papillary Muscles (PM) in our algorithm for specific clinical requirements. The key novel characteristics of present… Show more

The present work provides an efficient way of segmenting Left ventricular (LV) cavity and Myocardium (MC) for the cardiac analysis in a clinical environment. It is a fully automatic method based on multi-step binarization and adaptive threshold correction. It offers significant speed advantage over currently available methods. In addition, we provide an option to include/exclude Papillary Muscles (PM) in our algorithm for specific clinical requirements. The key novel characteristics of present approach are: (1) Otsu binarization based initial detection; (2) Threshold correction and PM detection; (3) Optional output of PM; (4) region and distance based MC segmentation. Our method is proved to be robust and faster as compared to recently published results. Show less

Optical character recognition, usually abbreviated to OCR, is the mechanical or electronic translation of images of handwritten, typewritten, or printed text into machine-editable text. OCR is a field of research in pattern recognition, artificial intelligence, and machine vision. In recent scenarios of machine vision, it is a popular choice for reading labels, image indexing, vehicle number plate recognition, and many other automation related applications. It is a major module in Text… Show more

Optical character recognition, usually abbreviated to OCR, is the mechanical or electronic translation of images of handwritten, typewritten, or printed text into machine-editable text. OCR is a field of research in pattern recognition, artificial intelligence, and machine vision. In recent scenarios of machine vision, it is a popular choice for reading labels, image indexing, vehicle number plate recognition, and many other automation related applications. It is a major module in Text Information Extraction (TIE) system. Thus, in general it can be said that there is a certain need for OCR systems in industry applications where reading and verifying of printed information on products is needed. One example can be found in the pharmaceutical industry, where the manufacturing date and expiry dates are important and minor printing errors can lead to a serious threat to the life or health of patients. Show less

Cardiac diseases are one of the major causes of deaths in western worlds. One type of cardiac disease is the left ventricular hypertrophy, which implies thickening of the heart muscle’s main pumping chamber i.e. the left ventricle (LV). The result is high blood pressure, which leads to other diseases. For diagnosing these heart diseases, several global parameters are used. These parameters can be calculated by accurate segmentation of the left ventricle and the myocardium (MC), i.e. from a… Show more

Cardiac diseases are one of the major causes of deaths in western worlds. One type of cardiac disease is the left ventricular hypertrophy, which implies thickening of the heart muscle’s main pumping chamber i.e. the left ventricle (LV). The result is high blood pressure, which leads to other diseases. For diagnosing these heart diseases, several global parameters are used. These parameters can be calculated by accurate segmentation of the left ventricle and the myocardium (MC), i.e. from a temporal sequence of images. Show less

Kannada text extraction from images is difficult than extraction of non-oriental language text like English. The geometry and inter-character spacing is non-uniform in Kannada script. Kannada Text extraction from camera captured images is a more challenging problem. The images can be from wide variety of sources and a generic system is difficult to build. We have developed a complete system which takes care of most image variations and the output of the system is Kannada text found in the… Show more

Kannada text extraction from images is difficult than extraction of non-oriental language text like English. The geometry and inter-character spacing is non-uniform in Kannada script. Kannada Text extraction from camera captured images is a more challenging problem. The images can be from wide variety of sources and a generic system is difficult to build. We have developed a complete system which takes care of most image variations and the output of the system is Kannada text found in the image. Testing is done on camera captured images and ICDAR database. Excellent results are achieved. Show less

Every language script has its own characteristics to distinguish it from others. In this paper, we describe some special attributes of character set of Kannada, English and Hindi and we show how we can utilize them for script identification. We propose a method to identify a language at character level in a mixed language document. This method gives very good results with an average accuracy of 99.2%.

In this paper, we have presented the comparative study of bidirectional optical fiber transmission scheme for various cases using Raman amplification process with and without pump depletion. Signal power amplification, pump depletion effect and equivalent fiber loss are discussed for various cases. The results are compared.

This study developed a new algorithm of patient-specific learning based segmentation (PSLS)
to segment lung tumors using multi-modality information of tumor characteristics from
simultaneous PET and CT images. PSLS begins with the user input of a few seed points inside and
outside the likely tumor region in the CT image followed by hybrid PET/CT image segmentation
in two steps. First step, we computed Initial Probability Map (IPM) by adding iteratively
segmented tumor regions… Show more

This study developed a new algorithm of patient-specific learning based segmentation (PSLS)
to segment lung tumors using multi-modality information of tumor characteristics from
simultaneous PET and CT images. PSLS begins with the user input of a few seed points inside and
outside the likely tumor region in the CT image followed by hybrid PET/CT image segmentation
in two steps. First step, we computed Initial Probability Map (IPM) by adding iteratively
segmented tumor regions from different thresholds. We derived Multi-modal Probabilistic
Mapping (MMPM) which accounts for input seed characteristics by combining several component
images like seed intensity based adaptive thresholding, seed’s location based adaptive distance
image (ADI), IPM etc. Probabilistic thresholding (PT) segmentation was obtained by applying
fixed thresholding on MMPM. In the second step, we improve the PT segmentation by use of
patient-specific learning. For each patient (tumor volume), we trained a unique multi-stage patientspecific classifier to distinguish voxels into tumor and non-tumor classes using seeds as training samples and feature set comprising multi-modal voxel intensities, ADI, MMPM values and other metrics like discrete cosine transform capturing global variations. The PSLS successfully
segmented lung tumors in all 35 patient cases. PSLS yields the highest performance among all
compared methods with DSI of 0.84±0.07 and 0.79±0.10, the lowest HD of <8mm and <6mm on average for large and small tumors respectively. The new PSLS algorithm was successfully applied on a complex patient group and had a close correspondence to the manually delineated
tumor. The new method outperforms other clinically used automated segmentation methods. Show less