Lesion Segmentation, Novel Lesion Depth/ Cancer Penetration Estimation technique devised. Shape, 3D reconstruction, Colour and Texture feature extraction. Machine learning for Lesion image analysis
Mammogram Segmentation, Identification of Mass/ Microcalcification structures. Feature extraction and Mammogram analysis using machine learning techniques
Fundus RETINA supervised feature based segmentation for identification vessels, cornea, optic disk. Feature extraction and Retinal image analysis to identify diabetes.
Bone radiograph images acquired using BMA™, D3A Medical Systems for lumbar spine, femoral neck and hip regions considered. Texture feature extraction and analysis to ascertain osteoporosis risk category
Foot thermal images acquired from DITI-SCT640, CX640, CG640, digital infrared thermal cameras are used to detect diabetic foot ulcers. Corse and Fine segmentation, Temperature variation measure based feature definition and analysis using machine learning is developed .
Fusion algorithm between Positron Emission Tomography (PET) and Single-photon Emission Computer Tomography (S-PECT), Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) images is realized. Medical image data is fused using Type II Fuzzy Logic. Fusion of multiple radiography techniques enables easy diagnosis by radiologists.
DNA microarray images from tissues are denoised and segmented. DNA data from images are extracted using a gridding mechanism and further analysed.
Histopathology tissue images are analysed for varied types of Breast cancer , cancer cell growth etc. Adaptive conditional random field segmentation, morphological operations, feature extraction and analysis is conducted.
Image retrieval or image search is a critical feature required currently. Novel shape, texture, colour and size features of each image is studied. Domain knowledge is derived from similar images and using machine learning mechanism image retrieval is achieved.
Varied filters/algorithms are required to pre process an image prior to any analysis or feature extraction. A variety of preprocessing are developed catering to varied applications. A partial list of preprocessing filters developed are: ❑ Super resolution ❑ Deblur / Deconvolution ❑ Sparse Reconstruction ❑ Denoising ❑ Demosaicing
Validation of digital scanned documents / social media images is critical. Adopting SIFT features, Clustering mechanisms key points are identified. Matching mechanism is used to detect forgery.
Plants / Crops are effected by various kind of diseases worldwide. A simple application that segments plant / leaves and analyses them for various diseases using machine learning and image processing techniques. Implemented for grape plants in vineyards.
Sparse Reconstruction : Large sizes of images and multi band / frequency data require huge computational resources. A novel sparsification and reconstruction algorithm that is iterative and intuitive is developed. It enables lower data volumes and is almost lossless.
Region Identification / Classification – Identification of vegetative regions , crop types, infrastructure in hyperspectral images is undertaken. Using advanced Convolutional Neural Networks this classification is achieved with desired accuracy. Frequency signatures are used for identification
Copyright © 2020 KnowTeq Info - All Rights Reserved.