Final Report Summary - 3D-OA-HISTO (Development of 3D Histopathological Grading of Osteoarthritis)
First, we successfully developed a new micro-CT based technique to analyze 3D collagen content distribution in articular cartilage (AC) in vitro using phosphotungstic acid (PTA) as the contrast agent. We have also successfully demonstrated 3D imaging of AC proteoglycan content distribution with micrometer-level resolution using the micro-CT and the novel contrast agent (CA4+). For 3D imaging of AC chondrons, we developed another novel micro-CT imaging protocol that enables obtaining 3D micro-CT data from the AC tissue with significantly better contrast between the chondrons and extracellular matrix than with earlier contrast agent based approaches. This new imaging protocol was also successfully applied to 3D visualization of human meniscus microstructures in normal and osteoarthritic tissue.
We developed novel computer algorithms to quantitatively analyze the subchondral plate and trabecular bone in 3D from micro-CT data sets. The developed algorithms were applied to human osteochondral sample set with different levels of osteoarthritis (OA). Subchondral bone changes due to OA were significantly associated with the true histological degenerative level of AC.
We developed a new method for quantifying morphometric properties of calcified cartilage from micro-CT data sets. The feasibility of the method was tested by analyzing human osteochondral sample set with different levels of OA. Calcified cartilage topography was significantly associated with the histological degenerative level of AC and open subchondral channel properties. It was concluded that our new analysis method can detect topographical changes in tidemark and calcified cartilage perforations associated with OA.
We acquired high-resolution synchrotron-CT data (Saskatoon, Canada) from human osteochondral samples in order to quantify osteocyte morphology in subchondral bone at different histopathological OA grades. It was hypothesized that osteocyte density and volume increase along with OA severity. We found that OA led to more elongated and thinner osteocytes in lower density bone tissue. It was also observed that a total volume of chondrocyte clusters appears to increase in early OA and decrease towards higher OARSI grades. This suggests that there are more hypertrophic chondrocytes in calcified cartilage in early and moderate OA and they disappear when cartilage becomes eroded.
We developed and validated a new, comprehensive, semi-quantitative 3D histopathological grading scale for human osteochondral samples using PTA-enhanced micro-CT. Our results clearly show that micro-CT based 3D histopathological grading is possible and reliable, and it correlates well with the conventional 2D histopathological grading from consecutive sample sections. This is one of the most significant methodological outcomes and it is directly related to the main overall goal of this project.
Furthermore, we developed several computational methods to automatically segment AC surface from micro-CT data sets, which is a prerequisite for automatized 3D histopathological grading of OA. Within this specific task, we innovated a new way to characterize complex OA changes in cartilage surface topology, which was validated with human osteochondral samples. We aim to eventually commercialize this innovation, for which we obtained additional funding from Business Finland (TUTLI project). The IP protection has also been initiated, i.e. patent applications have been filed to Finland, Australia, Canada, China, Europe, India, Japan and USA.
Regarding automatization of 3D histopathological grading of OA, we first developed machine learning based method to automatically segment the interface between non-calcified and calcified cartilage (tidemark) from micro-CT data sets. In the final phase of the project, we created an algorithm and pipeline to automatically conduct 3D histopathological grading of osteochondral samples. This is another major outcome of the project, and our algorithm and pipeline is also open access and freely available to anyone: https://github.com/MIPT-Oulu/3D-Histo-Grading
We also investigated the association of the subchondral bone morphometrics between the clinical resolution cone beam computed tomography (CBCT) and the desktop micro-CT. For human osteochondral samples, we found that quantitative parameters, obtained by applying grey-level co-occurrence matrix (GLCM) texture analysis for CBCT data, were strongly associated with true OA-related sub-resolution morphometrics analyzed by micro-CT. These findings support the use of quantitative image analysis for clinical CBCT data in order to obtain more specific information from OA-related bone alterations. Thus, it may be possible to translate 3D histopathological grading of articular cartilage and subchondral osteochondral to clinical CBCT scanners in the future.
With regard to understanding early OA changes, we found that in human OA subchondral bone changes are involved in the development of OA already at the very early phases. Furthermore, we also found in Fourier transform infrared microspectroscopy (FTIR-MS) study that both quantitative and qualitative changes occur in the AC composition during OA progression.
We developed novel computer algorithms to quantitatively analyze the subchondral plate and trabecular bone in 3D from micro-CT data sets. The developed algorithms were applied to human osteochondral sample set with different levels of osteoarthritis (OA). Subchondral bone changes due to OA were significantly associated with the true histological degenerative level of AC.
We developed a new method for quantifying morphometric properties of calcified cartilage from micro-CT data sets. The feasibility of the method was tested by analyzing human osteochondral sample set with different levels of OA. Calcified cartilage topography was significantly associated with the histological degenerative level of AC and open subchondral channel properties. It was concluded that our new analysis method can detect topographical changes in tidemark and calcified cartilage perforations associated with OA.
We acquired high-resolution synchrotron-CT data (Saskatoon, Canada) from human osteochondral samples in order to quantify osteocyte morphology in subchondral bone at different histopathological OA grades. It was hypothesized that osteocyte density and volume increase along with OA severity. We found that OA led to more elongated and thinner osteocytes in lower density bone tissue. It was also observed that a total volume of chondrocyte clusters appears to increase in early OA and decrease towards higher OARSI grades. This suggests that there are more hypertrophic chondrocytes in calcified cartilage in early and moderate OA and they disappear when cartilage becomes eroded.
We developed and validated a new, comprehensive, semi-quantitative 3D histopathological grading scale for human osteochondral samples using PTA-enhanced micro-CT. Our results clearly show that micro-CT based 3D histopathological grading is possible and reliable, and it correlates well with the conventional 2D histopathological grading from consecutive sample sections. This is one of the most significant methodological outcomes and it is directly related to the main overall goal of this project.
Furthermore, we developed several computational methods to automatically segment AC surface from micro-CT data sets, which is a prerequisite for automatized 3D histopathological grading of OA. Within this specific task, we innovated a new way to characterize complex OA changes in cartilage surface topology, which was validated with human osteochondral samples. We aim to eventually commercialize this innovation, for which we obtained additional funding from Business Finland (TUTLI project). The IP protection has also been initiated, i.e. patent applications have been filed to Finland, Australia, Canada, China, Europe, India, Japan and USA.
Regarding automatization of 3D histopathological grading of OA, we first developed machine learning based method to automatically segment the interface between non-calcified and calcified cartilage (tidemark) from micro-CT data sets. In the final phase of the project, we created an algorithm and pipeline to automatically conduct 3D histopathological grading of osteochondral samples. This is another major outcome of the project, and our algorithm and pipeline is also open access and freely available to anyone: https://github.com/MIPT-Oulu/3D-Histo-Grading
We also investigated the association of the subchondral bone morphometrics between the clinical resolution cone beam computed tomography (CBCT) and the desktop micro-CT. For human osteochondral samples, we found that quantitative parameters, obtained by applying grey-level co-occurrence matrix (GLCM) texture analysis for CBCT data, were strongly associated with true OA-related sub-resolution morphometrics analyzed by micro-CT. These findings support the use of quantitative image analysis for clinical CBCT data in order to obtain more specific information from OA-related bone alterations. Thus, it may be possible to translate 3D histopathological grading of articular cartilage and subchondral osteochondral to clinical CBCT scanners in the future.
With regard to understanding early OA changes, we found that in human OA subchondral bone changes are involved in the development of OA already at the very early phases. Furthermore, we also found in Fourier transform infrared microspectroscopy (FTIR-MS) study that both quantitative and qualitative changes occur in the AC composition during OA progression.