Patents and Publications

  • METHODS AND SYSTEMS OF PRIVACY PRESERVING OF OBJECTS IN MEDICAL IMAGES
    • Advenio TecnoSys, India, Patent Filing 2015

 

  • DEVICES FOR RETINAL IMAGING AND ANALYSIS METHODS AND SYSTEMS
    • Advenio TecnoSys, India, Patent Filing 2014

 

  • SMART PHONE BASED ULTRASOUND MACHINE
    • Advenio TecnoSys, India, Patent Filing 2013

 

  • METHOD FOR VISUALIZATION AND ANALYSIS OF MEDICAL IMAGES
    • Advenio TecnoSys, India, Patent Filing 2012

 

  • RETINAL IMAGE ANALYSIS AND METHOD
    • Advenio TecnoSys, India, Patent Filing 2011

 

 

OUR R&D TEAM MEMBERS WERE INVOLVED AS INVENTORS FOR THE FOLLOWING PATENTS


 

  • CURVATURE PROPERTIES AS RIB-CROSS FEATURES FOR REDUCTION OF FALSE POSITIVES IN CHEST X-RAY NODULE DETECTION
    • GRANTED @USPTO, US GRANT NO. 8229195, JULY 24, 2012

 

  • RIB FEATURES EXTRACTION FOR NODULE DETECTION AND FALSE POSITIVE REDUCTION IN CHEST X-RAYS
    • GRANTED @ USPTO, US GRANT NO 8150135, APRIL 03, 2012

 

  • CONTRAST DETECTION THROUGH PROFILE BASED APPROACH
    • GRANTED, @ USPTO, US GRANT NO 8090178, JANUARY 03, 2012

 

  • SYSTEM AND METHOD FOR AUTOMATIC DETECTION OF INTERNAL STRUCTURES IN MEDICAL IMAGES
    • United States Patent 8090178 (Issued January 03, 2012)

 

  • METHOD FOR THE AUTOMATIC ANALYSIS OF IMAGE DATA OF A STRUCTURE
    • United States Patent 20100303358 (Issued December 2, 2010)

 

  • METHOD AND SYSTEM FOR VERIFYING DETECTION OF A LUNG NODULE
    • United States Patent 20100040269 (Issued February 18, 2010)

 

  • CHARACTERIZATION OF LUNG NODULES
    • United States Patent 20090052763 (Issued February 26, 2009)

 

  • IDENTIFYING RIBS IN LUNG X-RAYS
    • United States Patent 20080317322 (Issued December 25, 2008)

 

  • IDENTIFYING BLOOD VESSELS IN LUNG X-RAY RADIOGRAPHS
    • United States Patent 20080298666 (Issued December 4, 2008)

 

  • SYSTEM AND METHOD FOR AUTOMATIC DETECTION OF INTERNAL STRUCTURES IN MEDICAL IMAGES
    • United States Patent 20070248254 (Issued October 25, 2007)

 

  • A NOVEL METHOD OF WAVELET-BASED WATERMARKING AND ITS APPLICATION IN MEDICAL DOMAIN
    • Siemens Business, Issued 2007

Publications


  • Low-cost, portable & high performance end-to-end eye care solution with computer aided detection for point-of-care diagnostics, The International Agency for the Prevention of Blindness (IAPB), Durban, South Africa, October 27-30, 2016
    • >Abstract: Current innovation is an affordable and accessible eye-care solution for point-of-care diagnostics. Innovation consists of a computer aided diagnostic (CAD) software enabling auto detection of retinal abnormalities like diabetic retinopathy, DME etc., which are leading causes of blindness and acts as primary-reader specifically  in resource constrained setting.  Our point-of-care solution solves a myriad of health-care needs and not limited to ophthalmologists. Diabetologists, general physicians, trained technicians can access it enabling early intervention of many retinal diseases, and early indications of diabetes.The software for auto detection of retinal abnormalities is computer aided diagnosis (CAD) and is a technology designed to eliminate observational oversights and discordance amongst physicians by providing an expert decision support algorithm, based on advanced image analysis, artificial intelligence and machine learning - the auto detection software trains itself with deep learning techniques for subsequently prompting decisions to humans.
  • RiView: An advanced computer aided diagnosis platform for tuberculosis detection from digital chest x-rays, 47th Union World Conference on Lung Health Liverpool, UK: 26 October - 29 October 2016
    • >Abstract: Chest x-rays are widely used in many settings, especially among private clinicians for diagnosis of lung symptomatic patients. However, variability in inter- & intra- reader agreement, wide ranges of sensitivity and specificity along with a lack of standard classification systems are well documented. The use of newer technologies like computer aided diagnosis (CAD) platforms for radiological image analysis and interpretation can further improve pulmonary tuberculosis interpretation from CXRs. RiView is an advanced lung abnormality detection CAD platform specifically trained on TB related radiological markers and is designed to offer differential diagnosis capability against other pathologies that mimic typical pulmonary tuberculosis abnormalities.
  • A supervised approach for automated detection of haemorrhages in retinal fundus images, 5th Edition of International Conference on Wireless Networks and Embedded Systems, IEEE, October 14-15, 2016
    • >Abstract: Diabetic Retinopathy, also known as DR, is an eye disease in which the retina is damaged because of the leakage of blood from the retinal blood vessels. The major cause of DR is diabetes followed by hypertension. The irregular flow of the blood from the vessels into the retina leads to retinal haemorrhages. DR causes blindness. The early detection of DR can be aided with the presence of haemorrhages. An automatic system to diagnose retinal haemorrhages is suggested. The study involves the analysis of 4546 blobs from 50 retinal fundus images taken from the dataset. The proposed method achieved the sensitivity of 90.42%, specificity of 92.53%. The algorithm proposed in the research will aid the ophthalmologists for the automatic detection of haemorrhages and might be a helpful tool in medical imaging.
  • An Effective Algorithm for Automatic Measurement of Vessel Calibre in Retinal Fundus Images, 5th Edition of International Conference on Wireless Networks and Embedded Systems, IEEE, October 14-15, 2016
    • >Abstract: Different pathologies like diabetic retinopathy, hypertension, arteriosclerosis, age related macular degeneration, occlusion, neovascularization etc. leads to various changes in retinal vasculature. Hence, analysis of retinal vasculature helps in diagnose and screening of eye fundus. This paper presents an effective algorithm for automatic measurement of vessel calibre of retinal blood vessels from the colored fundus images. Retinal blood vessels are extracted using multi-scale line tracking technique. Vessel calibre is measured for the extracted vessels in pixels. The proposed algorithm is validated on online available REVIEW dataset. Our results are very similar to the results given by observers which show high accuracy of our algorithm.
  • Automated Detection of red lesions in the presence of blood vessels in retinal fundus images using morphological operations, International Conference on Power Electronics, Intelligent Control and Energy Systems (IEEE ICPEICES), July 4-6, 2016
    • >Abstract: This paper focuses on the automatic detection of retinal abnormalities such as haemorrhages, also known as red lesions from the retinal fundus images. The primary cause of the red lesions in human eye is the disease which is commonly familiar with the name Diabetic Retinopathy, abbreviated as DR. DR is mainly found in people suffering from Type 1 and Type 2 Diabetes.
  • A Novel Lung Segmentation and Rib Suppression Algorithm for Nodular Lesion Detection from Digital Chest X-RAYS, International Symposium on Biomedical Imaging (ISBI'16), Prague, Czech Republic, April 16, 2016
    • >Abstract: Tuberculosis (TB) is a curable disease provided the major challenge of early detection and notification is overcome efficiently. The WHO has recommended chest radiographs for primary referral purpose while recently it again endorsed digital chest X-rays’ (CXR’s) role as primary screening modality for smear-negative pulmonary TB for prevalence studies in resource limited settings.As CXR’s are a representation of the lung in the 2D projection plane, the presence of overlapping structures make the analysis more challenging. Proper segmentation of the lung area along with the detection of the clavicles and the ribs is imperative to having a good detection of abnormal manifestations like nodular lesions (rough densities with irregular boundaries). Our two-step preprocessing approach performs lung area segmentation while in the second stage we perform rib suppression using our novel techniques for performance enhancement.
  • A hardware-neutral Computer Aided Detection (CADx) platform for Tuberculosis Interpretation from chest X-rays with Advanced false-positive reduction, 46th Union World Conference on Lung Health, Cape Town, South Africa, December 2-6, 2015
    • >Abstract: We propose a two-layered coarse-to-fine cascade framework under supervised learning. The first step is candidate generation module on a training data set. A true candidate is a localized region within the lung which exhibits TB patterns.  Candidate generation step involves extracting several descriptive features within annotated localized patches corresponding to normal and TB infected and subsequently training the first level classifier (random tree) with these features. At this stage we achieve high sensitivity but also with a high false positive level.  The candidates classified as TB infected are positive examples are then described by regions of interest (ROI) and are used as input for the second level classifier.  In the second level each candidate is resampled via scaling, translations, and rotations with respect to the candidate centroid to increase the variation of the training data and to avoid over fitting. These multiple views of candidates are used to train a deep Convolutional Neural Network (CNN) classifier. The CNN provides probability for each ROI view which are averaged and a single probability measure is obtained for each ROI or candidate.  For an unknown test dCXR image, candidates are similarly generated and the CNN classifier is applied to produce the probability measures for each ROI averaging which gives the single probability of each candidate and classifies it to be normal or abnormal based on a threshold value of the probability measure. This second level classifier behaves as a highly discriminating process to discard challenging false positives while still achieving high sensitivity.
  • Retinal image analysis for quantification of ocular disease, Proceedings of the SPIE, Volume 8406, id. 84060L, May 1, 2012
    • >Abstract: In this paper we propose to develop a computer assisted reading (CAR) tool for ocular disease. This involves identification and quantitative description of the patterns in retinal vasculature. The features taken into account are fractal dimension and vessel branching. Subsequently a measure combining all these features are designed which would help in quantifying the progression of the disease. The aim of the research is to develop algorithms that would help with parameterization of the eye fundus images, thus improving the diagnostics.