Krisha Santhoshi M
Biography
Research Interest
Abstract
Early Detection of Joint Osteonecrosis with Hybridized VGG19
Osteonecrosis of the hip, shoulder, and knee significantly impacts human lifestyle, especially among females and older men, which leads to Joint collapse with Knee replacement and Paralytic attack. In this project, for the early detection of Osteonecrosis, a literature survey is carried out. From the survey, it is identified that Osteonecrosis detection is done with X-ray, MRI scan, DXA, and CT scan. But these images have issues in identifying Osteonecrosis. In this project, Dynamic Contrast Enhanced MRI (DCE-MRI) and Single Photon Emission Computed Tomography (SPECT) are used to identify minute blood supply to the bone tissue. In most of the existing journals, either Humeral, Femoral, Navicular, or for Knee Osteonecrosis, early detection is carried out as Unimodal. Whereas in this project, for combination of all Osteonecrosis for Multi-modal is proposed. In the existing VGG16 model, which lacks fine feature extraction. VGG19 overcomes the above drawback by a deep layered architecture, capturing feature extraction, and improving accuracy for early detection of Osteonecrosis. CNN frameworks with fixed kernel sizes lack for both tiny lesions and large deformities simultaneously. Whereas Vision Transformer (ViT) in this project uses patches of different sizes, so it can detect small necrotic spots as well as large deformities. Hybridising Radiomics with ViT makes Osteonecrosis detection faster and more efficient. Therefore, this project proposal hybridizes VGG19 with ViT radiomics for the early detection of Joint Osteonecrosis.
