scholarly journals Automated analysis of rabbit knee calcified cartilage morphology using micro-computed tomography and deep learning segmentation

2020 ◽  
Author(s):  
Santeri J. O. Rytky ◽  
Lingwei Huang ◽  
Petri Tanska ◽  
Aleksei Tiulpin ◽  
Egor Panfilov ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Deep Learning ◽  
Mixed Model ◽  
Linear Mixed Model ◽  
Anatomical Variation ◽  
Pearson Correlation ◽  
Micro Computed Tomography ◽  
Calcified Cartilage ◽  
Conventional Histology ◽  
Segmentation Models

AbstractPurposeOnly little is known how calcified cartilage (CC) structure changes during exercise, aging and disease. CC thickness (CC.Th) can be analyzed using conventional histological sections. Micro-computed tomography (μCT) allows for three-dimensional (3D) imaging of mineralized tissues, however, the segmentation between bone and CC is challenging. Here, we present state-of-the-art deep learning segmentation for μCT images to enable assessment of CC morphology.MethodsSixteen knees from twelve New Zealand White rabbits were dissected into osteochondral samples from six anatomical regions: lateral and medial femoral condyles, lateral and medial tibial plateaus, femoral groove and patella (n = 96). Samples were imaged with μCT and processed for conventional histology. Manually segmented CC from the histology and reconstructed μCT images was used as the gold standard to train segmentation models with different encoder-decoder architectures. The models with the greatest out-of-fold evaluation Dice score were used for automated CC.Th analysis. Subsequently, the automated CC.Th analysis was compared across a total of 24 regions, co-registered between the imaging modalities, using Pearson correlation and Bland-Altman analyses. Finally, the anatomical variation in CC.Th was assessed via a Linear Mixed Model analysis.ResultsThe best segmentation models yielded average Dice scores of 0.891 and 0.807 for histology and μCT segmentation, respectively. The correlation between the co-registered regions across the modalities was strong (r = 0.897). The Bland-Altman analysis yielded a bias of 21.9 μm and a standard deviation of 21.5 μm between the methods. Finally, both methods could separate the CC morphology between the patella, femoral, and tibial regions (p < 0.001).ConclusionThe presented method allows for ex vivo 3D assessment of CC.Th in an automated and non-destructive manner. We demonstrated its utility by quantifying CC.Th in different anatomical regions. CC.Th was the thickest in the patella and the thinnest in the tibial plateau.Graphical abstractWe present a μCT-based method with deep learning segmentation for analyzing calcified cartilage thickness (CC.Th). The method is compared throughout the study against conventional histology. The comparison against co-registered regions yielded a strong Pearson correlation (r = 0.90). Both methods were able to separate the CC.Th properties between tibia, femur, and patella.

scholarly journals Automated analysis of rabbit knee calcified cartilage morphology using micro‐computed tomography and deep learning

2021 ◽  
Author(s):  
Santeri J. O. Rytky ◽  
Lingwei Huang ◽  
Petri Tanska ◽  
Aleksei Tiulpin ◽  
Egor Panfilov ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Deep Learning ◽  
Automated Analysis ◽  
Micro Computed Tomography ◽  
Rabbit Knee ◽  
Calcified Cartilage

scholarly journals THU0059 INCREASING SURFACE LUBRICATION WITH POLY(2-METHYL-2-OXAZOLINE) HALTS DEGENERATIVE CHANGES IN THE CARTILAGE IN A COLLAGENASE INDUCED OSTEOARTHRITIS (CIOA) RAT MODEL

2020 ◽  
Vol 79 (Suppl 1) ◽  
pp. 241.1-242
Author(s):  
S. Kauppinen ◽  
D. Fercher ◽  
G. Barreto ◽  
G. Morgese ◽  
E. Benetti ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Articular Cartilage ◽  
Mixed Model ◽  
Linear Mixed Model ◽  
Structural Defects ◽  
Control Group ◽  
Surface Erosion ◽  
Micro Computed Tomography ◽  
Medial Side ◽  
Contrast Enhanced

Background:Degenerative lesions of articular cartilage (AC) surface are related to disruption of the well-organized collagen network and allow proteoglycans to escape from the tissue. Ultimately, this leads to the development of osteoarthritis (OA). Targeted therapy for early AC lesions could provide an effective way to halt the OA development process.Objectives:This study aims to evaluate the effectiveness of an engineered surface lubricant; poly(2-methyl-2-oxazoline) (PMOXA)1to prevent the destruction of the AC surface. Our recently developed contrast-enhanced µCT (CEµCT) method was used to quantify AC surface erosion2.Methods:OA was induced in 12-18 week-old male Wistar rats (N=17) with an injection of 250 U Collagenase within 25 µL solution into the left hind limb. Both hind legs were treated with a second injection three days after the collagenase injection (CI). Three groups were formed by using either PMOXA (N=5), hyaluronic acid (HA; N=6), or saline (N=6) during the second injection. The animals were sacrificed after 45 days, and harvested knees were fixed in phosphate-buffered formalin for a week. Knees were stored in 70% ethanol, and tibia and femur were carefully dissected free of other tissue, stained with 1% phosphotungstic acid3, and scanned with a desktop µCT with 2.8µm pixel size. The medial and lateral AC surfaces were manually segmented from 3D projections using an in-house developed program (Matlab sofware). These surfaces were analyzed by iteratively fitting a reference surface (RS) to a median-filtered smoothed surface representing a perfectly smooth surface, capturing the realistic shape AC. An offset of 5 pixels (14 µm) was added between the RS and the original surface (OS). Two quantitative parameters were calculated from the data: Average of Maximum Void Depth (MVD) (depth of lesion) and Degeneration-% (area exceeding 20 µm MVD / whole area) *100). Estimates of mean differences from all groups against the CI+Saline -group were determined using a linear mixed model.Results:Boxplots from tested groups are shown in Fig. 1A and MVD results are visualized in Fig. 1B. Collagenase caused structural defects only on the medial and lateral tibial AC surfaces, which was seen as increased MVD and Degeneration-%. CI changes were not seen in PMOXA or HA treated groups. Furthermore, MVD and Degeneration% were lower in CI knees that were treated with PMOXA.Figure 1.A) Boxplots of Maximum Void Depth (MVD) and Degeneration-%. Lateral and medial side are analyzed separately for both tibias and femurs. Stars indicate if a group was statistically different from control group (CI+Saline).CI= red, no CI= blue. B) Representative visualizations for maximum void depth overlayed on top of the 3D AC surface.Conclusion:Our CEµCT analysis method was able to detect subtle changes of the AC surface in the medial and lateral tibial cartilage, caused by the CI. In contrast, the CI did not cause detectable changes in the AC of the femur, which indicates that in the CIOA model, the tibia is more susceptible to structural degradation. Our results show that early intervention with HA or PMOXA can halt the degenerative AC changes caused by CI. However, HA did not suppress the effects of CI in the medial tibia, which indicates that PMOXA could be more effective to prevent the development of OA.References:[1]Morgese G, et al. Hairy and slippery polyoxazoline-based copolymers on model and cartilage surfaces. Biomacromolecules 2018 19 (2), 680-690[2]Ylitalo T, et al. Quantifying Complex Micro-Topography of Degenerated Articular Cartilage Surface by Contrast-Enhanced Micro-Computed Tomography and Parametric Analyses. J Orthop Res. 2019 Apr;37(4):855-866.[3]Nieminen HJ, et al. Determining collagen distribution in articular cartilage using contrast-enhanced micro-computed tomography. Osteoarthritis Cartilage. 2015;23(9):1613–1621Disclosure of Interests:None declared

scholarly journals 3D morphometric analysis of calcified cartilage properties using micro-computed tomography

2019 ◽  
Vol 27 (1) ◽  
pp. 172-180 ◽  
Author(s):  
S. Kauppinen ◽  
S.S. Karhula ◽  
J. Thevenot ◽  
T. Ylitalo ◽  
L. Rieppo ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Morphometric Analysis ◽  
Micro Computed Tomography ◽  
Calcified Cartilage

scholarly journals MP13: Association between the quantity of subcutaneous fat and the inter-device agreement of two tissue oximeters

CJEM ◽  
2019 ◽  
Vol 21 (S1) ◽  
pp. S46-S47
Author(s):  
A. Cournoyer ◽  
S. Cossette ◽  
J. Paquet ◽  
R. Daoust ◽  
M. Marquis ◽  
...  
Keyword(s):  
Mixed Model ◽  
Linear Mixed Model ◽  
Pearson Correlation ◽  
Subcutaneous Fat ◽  
Skinfold Thickness ◽  
Repeated Measurements ◽  
The Body ◽  
Absolute Difference ◽  
Specific Patient ◽  
The Impact

Introduction: Near-infrared spectroscopy (NIRS) can be used to monitor the oxygen saturation of hemoglobin in any given superficial tissue. However, the measurements provided by different oximeters can vary a lot. Little is known about the specific patient characteristics that could affect the inter-device agreement of tissular oximeters. This study aimed to evaluate the association between the quantity of subcutaneous fat (assessed by skinfold thickness) and the inter-device agreement of two tissue oximeters, the INVOS 5100c and the Equanox 7600. Methods: In this prospective cohort study, tissue saturations and skinfold thickness were measured at four different sites on both sides of the body in healthy adult (≥18 years old) volunteers. The association between the quantity of subcutaneous fat (assessed by skinfold thickness) and the inter-device agreement (absolute difference between the oximetry values provided by the two oximeters) was first assessed with a Pearson's correlation and a scatter plot. Subsequently, a linear mixed model was used to evaluate the impact of the subcutaneous fat and other covariables (age, sex) on the inter-device agreement while adjusting for the repeated measurements across different sites for the same volunteers. Results: From January to March 2015, 53 healthy volunteers were included in this study with ages ranging between 20 and 81 years old, on which a total of 848 measures were taken. Higher skinfold measures were associated with an increase in the difference between measures provided by both oximeters (Slope = -0.59, Pearson correlation coefficient = -0.51, p &lt; 0.001). This observed association persisted in a linear mixed model (-0.48 [95% confidence interval {CI}-0.61 to -0.36], p &lt; 0.001). The sex of the volunteers also influenced the inter-oximeter agreement (Women:-5.77 [95%CI -8.43 to -3.11], p &lt; 0.001), as well as the forearm sites (Left forearm: −7.16 [95%CI -9.85 to –4.47], p &lt; 0.001; right forearm: −7.01 [95%CI -9.61 to −4.40], p &lt; 0.001). Conclusion: The quantity of subcutaneous fat, as well as the sex of the volunteers and the measurement sites, impacted the inter-device agreement of two commonly used oximeters. Given these findings, monitoring using tissue oximetry should be interpreted with great care when there is a significant quantity of subcutaneous fat.

scholarly journals Prediction of age and brachial-ankle pulse-wave velocity using ultra-wide-field pseudo-color images by deep learning

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Daisuke Nagasato ◽  
Hitoshi Tabuchi ◽  
Hiroki Masumoto ◽  
Takanori Kusuyama ◽  
Yu Kawai ◽  
...  
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Pulse Wave Velocity ◽  
Wave Velocity ◽  
Mixed Model ◽  
Linear Mixed Model ◽  
Pulse Wave ◽  
Wide Field ◽  
The Neural Network ◽  
Pseudo Color

Abstract This study examined whether age and brachial-ankle pulse-wave velocity (baPWV) can be predicted with ultra-wide-field pseudo-color (UWPC) images using deep learning (DL). We examined 170 UWPC images of both eyes of 85 participants (40 men and 45 women, mean age: 57.5 ± 20.9 years). Three types of images were included (total, central, and peripheral) and analyzed by k-fold cross-validation (k = 5) using Visual Geometry Group-16. After bias was eliminated using the generalized linear mixed model, the standard regression coefficients (SRCs) between actual age and baPWV and predicted age and baPWV from the UWPC images by the neural network were calculated, and the prediction accuracies of the DL model for age and baPWV were examined. The SRC between actual age and predicted age by the neural network was 0.833 for all images, 0.818 for central images, and 0.649 for peripheral images (all P < 0.001) and between the actual baPWV and the predicted baPWV was 0.390 for total images, 0.419 for central images, and 0.312 for peripheral images (all P < 0.001). These results show the potential prediction capability of DL for age and vascular aging and could be useful for disease prevention and early treatment.

scholarly journals PSI-39 Retrospective analysis of semen quality attributes of bulls in the Cal Poly Bull Test

2020 ◽  
Vol 98 (Supplement_4) ◽  
pp. 477-477
Author(s):  
Payton A Thomas ◽  
Catherine E Field ◽  
Mohammed Abo-Ismail ◽  
Zach D McFarlane
Keyword(s):  
Sperm Motility ◽  
Mixed Model ◽  
Performance Test ◽  
Linear Mixed Model ◽  
Semen Quality ◽  
Sperm Count ◽  
Body Weight Gain ◽  
Pearson Correlation ◽  
Average Daily Gain ◽  
Quality Attributes

Abstract The Cal Poly Bull Test has been offering valuable information for genetic improvement via a performance test of yearling beef bulls across the western United States since 1951. Growth and reproductive performance phenotypes were measured and collected over time. Therefore, the objective of this study was to determine the relationship between growth traits and semen quality traits. Performance records and semen samples of Hereford and Angus bulls were collected over a 12-year span from 2001–2013. A total of 1,982 records from Angus (n = 1692) and Hereford (n = 290) bulls were analyzed. All bulls were fall-born and weaned prior to the start of test in May. Bulls were fed for 99.83 ± 0.37 days in accordance with the guidelines of the Beef Improvement Federation where growth performance was assessed monthly during that period. After the conclusion of the test, usually in August, semen was collected from bulls and assessed for sperm motility, morphology and total sperm count. Bulls were retrospectively classified as low (0–1.4 kg), moderate (1.4–1.8 kg), or high (1.8–2.7 kg) average daily gain (ADG) bulls. The correlation among traits was evaluated using Pearson correlation, whereas a linear mixed model was used to evaluate the effect of growth on semen quality attributes. Bull age was moderately, negatively correlated (P &lt; 0.01, r2 = -0.3) with scrotal circumference. Sperm motility and morphology were moderately correlated (P &lt; 0.01, r2 = 0.35). Bull ADG was not strongly correlated (P &lt; 0.01, r2 = 0.15) with semen motility or morphology. The results indicated a significant breed effect (P &lt; 0.01) on semen motility and morphology. Angus bulls had a higher percent of progressively motile spermatozoa (P &lt; 0.01; 72.23% ± 0.73) when compared with Hereford bulls (65.88% ± 1.37). Additionally, Angus bulls had a higher percentage of morphologically normal sperm (P &lt; 0.01; 73.65% ± 0.59) when compared with Hereford bulls (66.29% ± 1.1). However, ADG as a class variable had no impact (P ≥ 0.50) on semen motility or morphology. Thus, these results indicate that higher levels of ADG during the testing period did not negatively impact semen quality attributes. More data must be collected to assess the biology behind the breed effects and validate the effect of body weight gain on semen quality attributes.

Sign in / Sign up

Export Citation Format

Share Document