Limits...
Quantitative image analysis reveals distinct structural transitions during aging in Caenorhabditis elegans tissues.

Johnston J, Iser WB, Chow DK, Goldberg IG, Wolkow CA - PLoS ONE (2008)

Bottom Line: Such approaches are inadequate for the complex changes associated with aging.The processes that underlie these architectural changes may contribute to increased disease risk during aging, and may be targets for factors that alter the aging rate.This work further demonstrates that pattern analysis of an image series offers a novel and generally accessible approach for quantifying morphological changes and identifying structural biomarkers.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Genetics, NIA Intramural Research Program, National Institutes of Health, Baltimore, Maryland, United States of America.

ABSTRACT
Aging is associated with functional and structural declines in many body systems, even in the absence of underlying disease. In particular, skeletal muscles experience severe declines during aging, a phenomenon termed sarcopenia. Despite the high incidence and severity of sarcopenia, little is known about contributing factors and development. Many studies focus on functional aspects of aging-related tissue decline, while structural details remain understudied. Traditional approaches for quantifying structural changes have assessed individual markers at discrete intervals. Such approaches are inadequate for the complex changes associated with aging. An alternative is to consider changes in overall morphology rather than in specific markers. We have used this approach to quantitatively track tissue architecture during adulthood and aging in the C. elegans pharynx, the neuromuscular feeding organ. Using pattern recognition to analyze aged-grouped pharynx images, we identified discrete step-wise transitions between distinct morphologies. The morphology state transitions were maintained in mutants with pharynx neurotransmission defects, although the pace of the transitions was altered. Longitudinal measurements of pharynx function identified a predictive relationship between mid-life pharynx morphology and function at later ages. These studies demonstrate for the first time that adult tissues undergo distinct structural transitions reflecting postdevelopmental events. The processes that underlie these architectural changes may contribute to increased disease risk during aging, and may be targets for factors that alter the aging rate. This work further demonstrates that pattern analysis of an image series offers a novel and generally accessible approach for quantifying morphological changes and identifying structural biomarkers.

Show MeSH

Related in: MedlinePlus

Age-scores for normal animals revealed distinct stages of adult morphology.(A) Age-scores were generated by converting each image's class distance into a class-probability, which were combined with class ages to produce an age-score. For training, 85 images from fem-1(hc17) adults aged day 0 to 12 were used for each class. (B) Age-scores for every test image from normal animals. The horizontal axis was ordered primarily by known age, and secondarily by age-scores. The yellow bars highlight inflections in the age-score distribution for images in each age group. These inflection regions constitute the stable morphology states depicted in (D). The number of test images for days 0–12 was 21, 143, 74, 91, 177, 40, and 62, respectively. (C) Probability distributions of age-scores, which can be interpreted as continuously sampled histograms with peaks indicating the most probable age-score. (D) Distinct stages of adulthood as determined by distribution of age-scores. Yellow ovals for each age state correspond to the inflections in the age-score distribution in part (B). See Table 1 and text for statistics.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2483734&req=5

pone-0002821-g003: Age-scores for normal animals revealed distinct stages of adult morphology.(A) Age-scores were generated by converting each image's class distance into a class-probability, which were combined with class ages to produce an age-score. For training, 85 images from fem-1(hc17) adults aged day 0 to 12 were used for each class. (B) Age-scores for every test image from normal animals. The horizontal axis was ordered primarily by known age, and secondarily by age-scores. The yellow bars highlight inflections in the age-score distribution for images in each age group. These inflection regions constitute the stable morphology states depicted in (D). The number of test images for days 0–12 was 21, 143, 74, 91, 177, 40, and 62, respectively. (C) Probability distributions of age-scores, which can be interpreted as continuously sampled histograms with peaks indicating the most probable age-score. (D) Distinct stages of adulthood as determined by distribution of age-scores. Yellow ovals for each age state correspond to the inflections in the age-score distribution in part (B). See Table 1 and text for statistics.

Mentions: For a more direct comparison between the test images, we calculated an age-score for each image from its distances to the class centroids (Fig. 3A, Table 1). For this analysis, a larger number of test images were used for each age group (day 0, 21 images; day 2, 143 images; day 4, 74 images; day 6, 91 images; day 8, 177 images; day 10, 40 images and day 12, 62 images). The Pearson's correlation coefficient between calculated age-score and chronological age was 0.54 (p = 5.34×10−47), indicating a statistically significant correlation between chronological age and the assigned age-score. Although most age groups were statistically distinguishable, age-scores overlapped between different age groups (Fig. 3B), but most age groups were statistically distinguishable (Table 1). Overall, these statistical tests indicated that there was a robust, but not perfect, correlation between the predicted age scores and actual age. This imperfect correlation likely reflects morphological heterogeneity of the population during aging. Consistent with increasing heterogeneity during aging, statistical significance was stronger in the younger age groups than among the older age groups. To simplify the representations of predicted age-score, probability density functions were generated for each age group (Fig. 3C). The peaks of these functions correspond to inflections in the rank-ordered age score predictions for each age group, indicating that morphologies occupied preferred states rather than being equally distributed within an age class. The first distinguishable states (Fig. 3D) occurred during early adulthood, at days 0 and 2 (age score, 4.5–5 and 5.5–6). These appeared to constitute subgroups of a single “young adult” state (Ia and Ib). The second major state occurred on days 4, 6 and 8 (age score, 6.5–7). We termed this state II and it appears to encompass mid-life. The third state (state III) represented morphologies associated with late life and included days 10 and 12. Although the day 10 population was not statistically distinguishable from days 6 and 8 (p = 0.47), the peak appeared slightly shifted from the overlapping peaks of days 6 and 8 (Fig. 3C). By day 12, the population was statistically distinguishable from state II, whether or not the calculation included day 10 data (p = 0.003 and p = 0.0025, respectively). This was termed state III and appeared to represent morphologies associated with late mid-life. We note that many images fell outside of the scores defining these morphology states, reflecting the heterogeneity in pharynx morphology at all ages.


Quantitative image analysis reveals distinct structural transitions during aging in Caenorhabditis elegans tissues.

Johnston J, Iser WB, Chow DK, Goldberg IG, Wolkow CA - PLoS ONE (2008)

Age-scores for normal animals revealed distinct stages of adult morphology.(A) Age-scores were generated by converting each image's class distance into a class-probability, which were combined with class ages to produce an age-score. For training, 85 images from fem-1(hc17) adults aged day 0 to 12 were used for each class. (B) Age-scores for every test image from normal animals. The horizontal axis was ordered primarily by known age, and secondarily by age-scores. The yellow bars highlight inflections in the age-score distribution for images in each age group. These inflection regions constitute the stable morphology states depicted in (D). The number of test images for days 0–12 was 21, 143, 74, 91, 177, 40, and 62, respectively. (C) Probability distributions of age-scores, which can be interpreted as continuously sampled histograms with peaks indicating the most probable age-score. (D) Distinct stages of adulthood as determined by distribution of age-scores. Yellow ovals for each age state correspond to the inflections in the age-score distribution in part (B). See Table 1 and text for statistics.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2483734&req=5

pone-0002821-g003: Age-scores for normal animals revealed distinct stages of adult morphology.(A) Age-scores were generated by converting each image's class distance into a class-probability, which were combined with class ages to produce an age-score. For training, 85 images from fem-1(hc17) adults aged day 0 to 12 were used for each class. (B) Age-scores for every test image from normal animals. The horizontal axis was ordered primarily by known age, and secondarily by age-scores. The yellow bars highlight inflections in the age-score distribution for images in each age group. These inflection regions constitute the stable morphology states depicted in (D). The number of test images for days 0–12 was 21, 143, 74, 91, 177, 40, and 62, respectively. (C) Probability distributions of age-scores, which can be interpreted as continuously sampled histograms with peaks indicating the most probable age-score. (D) Distinct stages of adulthood as determined by distribution of age-scores. Yellow ovals for each age state correspond to the inflections in the age-score distribution in part (B). See Table 1 and text for statistics.
Mentions: For a more direct comparison between the test images, we calculated an age-score for each image from its distances to the class centroids (Fig. 3A, Table 1). For this analysis, a larger number of test images were used for each age group (day 0, 21 images; day 2, 143 images; day 4, 74 images; day 6, 91 images; day 8, 177 images; day 10, 40 images and day 12, 62 images). The Pearson's correlation coefficient between calculated age-score and chronological age was 0.54 (p = 5.34×10−47), indicating a statistically significant correlation between chronological age and the assigned age-score. Although most age groups were statistically distinguishable, age-scores overlapped between different age groups (Fig. 3B), but most age groups were statistically distinguishable (Table 1). Overall, these statistical tests indicated that there was a robust, but not perfect, correlation between the predicted age scores and actual age. This imperfect correlation likely reflects morphological heterogeneity of the population during aging. Consistent with increasing heterogeneity during aging, statistical significance was stronger in the younger age groups than among the older age groups. To simplify the representations of predicted age-score, probability density functions were generated for each age group (Fig. 3C). The peaks of these functions correspond to inflections in the rank-ordered age score predictions for each age group, indicating that morphologies occupied preferred states rather than being equally distributed within an age class. The first distinguishable states (Fig. 3D) occurred during early adulthood, at days 0 and 2 (age score, 4.5–5 and 5.5–6). These appeared to constitute subgroups of a single “young adult” state (Ia and Ib). The second major state occurred on days 4, 6 and 8 (age score, 6.5–7). We termed this state II and it appears to encompass mid-life. The third state (state III) represented morphologies associated with late life and included days 10 and 12. Although the day 10 population was not statistically distinguishable from days 6 and 8 (p = 0.47), the peak appeared slightly shifted from the overlapping peaks of days 6 and 8 (Fig. 3C). By day 12, the population was statistically distinguishable from state II, whether or not the calculation included day 10 data (p = 0.003 and p = 0.0025, respectively). This was termed state III and appeared to represent morphologies associated with late mid-life. We note that many images fell outside of the scores defining these morphology states, reflecting the heterogeneity in pharynx morphology at all ages.

Bottom Line: Such approaches are inadequate for the complex changes associated with aging.The processes that underlie these architectural changes may contribute to increased disease risk during aging, and may be targets for factors that alter the aging rate.This work further demonstrates that pattern analysis of an image series offers a novel and generally accessible approach for quantifying morphological changes and identifying structural biomarkers.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Genetics, NIA Intramural Research Program, National Institutes of Health, Baltimore, Maryland, United States of America.

ABSTRACT
Aging is associated with functional and structural declines in many body systems, even in the absence of underlying disease. In particular, skeletal muscles experience severe declines during aging, a phenomenon termed sarcopenia. Despite the high incidence and severity of sarcopenia, little is known about contributing factors and development. Many studies focus on functional aspects of aging-related tissue decline, while structural details remain understudied. Traditional approaches for quantifying structural changes have assessed individual markers at discrete intervals. Such approaches are inadequate for the complex changes associated with aging. An alternative is to consider changes in overall morphology rather than in specific markers. We have used this approach to quantitatively track tissue architecture during adulthood and aging in the C. elegans pharynx, the neuromuscular feeding organ. Using pattern recognition to analyze aged-grouped pharynx images, we identified discrete step-wise transitions between distinct morphologies. The morphology state transitions were maintained in mutants with pharynx neurotransmission defects, although the pace of the transitions was altered. Longitudinal measurements of pharynx function identified a predictive relationship between mid-life pharynx morphology and function at later ages. These studies demonstrate for the first time that adult tissues undergo distinct structural transitions reflecting postdevelopmental events. The processes that underlie these architectural changes may contribute to increased disease risk during aging, and may be targets for factors that alter the aging rate. This work further demonstrates that pattern analysis of an image series offers a novel and generally accessible approach for quantifying morphological changes and identifying structural biomarkers.

Show MeSH
Related in: MedlinePlus