Abstract
Research Type:
Level 2 - Prospective comparative study, Meta-analysis of Level 2 studies or Level 1 studies with inconsistent results
Introduction/Purpose:
There is an urgent, unmet clinical need for non-invasive imaging markers to monitor the health of tendons and surrounding peritendinous tissues objectively. Magnetic resonance imaging (MRI) and radiomics (i.e., quantitative image analysis) offer a potential solution yet have not been fully explored in preclinical animal models, where the introduction of therapeutic interventions could be studied over time. Ex vivo MRI enables higher resolution imaging with less time and fewer animal handling considerations associated with in vivo MRI. However, the effects of sample preparation for ex vivo MRI on radiomics remain unknown. In this study, we explored the use of in vivo and ex vivo MRI in a rat model of Achilles tendon injury to investigate the Achilles tendon, paratenon, and Kager’s fat pad radiomic phenotypes.
Methods:
The right ankles of male Sprague Dawley rats (n=10) underwent IACUC-approved Achilles tendon transection with primary Kessler repair, followed by one week of cast immobilization; the contralateral ankle was the unoperated control. At 6 weeks post-injury, ankles were imaged in vivo using proton density (PD) and T2 Dixon MRI sequences (94µm sagittal plane resolution), assessing structure and water/fat content. Rats were then euthanized, and ankles were dissected, fixed in 10% formalin for 14 days, and soaked in gadolinium for 48 hours. Ankles were suspended in Fluoinert for ex vivo MRI using PD and T2 Dixon sequences (45µm resolution). Manual segmentation was performed on the sagittal plane of PD sequence to extract tissue area and content (i.e., intensity). One-way repeated-measures ANOVA using the Benjamini, Krieger, and Yekutieli method compared each radiomic feature between healthy and injured tissue. Bland Altman visualized the under or overestimated ex vivo features compared to in vivo.
Results:
Qualitatively (Figure 1A), in vivo and ex vivo MRI appear similar for the Achilles tendon, the anterior and posterior paratenon, and Kager’s Fat Pad. Radiomics (visual summary shown in Figure 1B) revealed that in both in vivo and ex vivo MRI, injury increases the area for the Achilles tendon and paratenon; however, there is increased water content in vivo only. Simultaneously, the injured Kager’s fat pad decreased in both area and fat content in vivo and ex vivo. The Bland-Altman plots indicate that tissue area is under and overestimated in ex vivo compared to in vivo MRI and depends on specific tissues (Figure 1C). Conversely, water and fat content is overestimated in the ex vivo relative to the in vivo MRI.
Conclusion:
We demonstrate that in vivo and ex vivo MRI radiomics identified similar phenotypes of the Achilles tendon, paratenon, and Kager’s fat pad. Injured Achilles tendon and paratenon increase in size and water content, indicating healing and inflammation, while decreases in Kager’s fat pad area and fat content suggest a transition to collagen-rich tissue. Although differences arise between in vivo and ex vivo, particularly for water content, group-wise and tissue-specific analyses reveal similar patterns. Establishing in vivo and ex vivo MRI radiomic phenotypes is crucial to advancing unbiased quantification of Achilles tendon healing for future orthopedic research and clinical applications.
In Vivo and Ex Vivo MRI Radiomic Phenotypes of Achilles Tendon and Peritendious Tissues
Figure 1. (A) Qualitative ex vivo vs in vivo MRI sequences of proton density (PD), T2 Dixon for combo (both water and fat), T2 Dixon Water, and T2 Dixon Fat. Note: blue = Achilles Tendon; Pink = Kager’s fat pad; white = anterior paratenon; yellow = posterior paratenon. (B) Summary of injured group-wise tissue radiomic features for in-vivo and ex-vivo MRI relative to healthy control (arrows indicate significant different in that direction). (C) Bland Altman plots for demonstrate that ex vivo MRI underestimates tissue area while water and fat content is overestimated compared to in vivo.