Volumetric Assessment of Articular Cartilage in Knee Osteoarthritis using MR Imaging

Abstract

Osteoarthritis (OA) is the serious life altering disease that affects various joints newlinein human body such as knee, hip, spine, hand, etc. but knee OA is the most newlineprevalent across worldwide. The assessment of knee OA can be performed based newlineon the changes and gradual loss of Articular Cartilage (AC). Its early detection newlineis necessary to initiate the hyaluronic acids or Non-Steroidal Anti-Inflammatory newlineDrugs (NSAIDs) treatment to stop or control further AC degradation. For early newlinedetection of knee OA volumetric and morphological assessment such as volume newlineand thickness of AC are the suitable features. Till date Magnetic Resonance newlineImaging (MRI) has shown exceptional outcomes in quantifying these features newlinebut with one limitations. These limitations are in terms of inadequate contrast newlineand brightness, weak tissue boundaries, etc. A novel contrast enhancement newlinetechnique called as gray level S-curve transformation is proposed in this thesis newlinethat enhances contrast and improves adjacent tissue boundaries. To overcome newlineother difficulties in automatic AC segmentation, this thesis also proposes the newlinebone segmentation as the prior step of AC segmentation. Since, bone being the newlinemost stable anatomical structure and largely occupied tissue in sagittal-view newlineknee MR images. This is achieved through initial bone region extraction using newlinethe Distance Regularized Level Set Evolution (DRLSE) by using automatic seed newlineinitialization technique followed by the accurate segmentation using boundary newlineleakage detection and correction using boundary displacement technique and newlinemulti-plane morphological technique. These two are the proposed newlinemethodologies for accurate segmentation of subchondral bones. Surface of newlinesegmented subchondral bones are utilized for the localization of AC through BCI newlineextraction using slice specific mask design. Extracted BCI has limited number of newlinetissues that minimizes the computations required to locate the AC. Cartilage, newlinemeniscus, synovial fluid and small part of bone, these are the tissues present in newlineextracted BCI with different int