Histological Image Processing Features Induce a Quantitative Characterization of Chronic Tumor Hypoxia.
Bottom Line:
We use image-processing algorithms to develop a set of candidate image features that can formulate just such a quantitative description of xenographed colorectal chronic tumor hypoxia.Two features in particular give low-variance measures of chronic hypoxia near a vessel: intensity sampling that extends radially away from approximated blood vessel centroids, and multithresholding to segment tumor tissue into normal, hypoxic, and necrotic regions.From these features we derive a spatiotemporal logical expression whose truth value depends on its predicate clauses that are grounded in this histological evidence.
View Article:
PubMed Central - PubMed
Affiliation: Department of Pharmacology and Systems Therapeutics, Icahn School of Medicine at Mount Sinai, New York, NY, United States of America.
ABSTRACT
Hypoxia in tumors signifies resistance to therapy. Despite a wealth of tumor histology data, including anti-pimonidazole staining, no current methods use these data to induce a quantitative characterization of chronic tumor hypoxia in time and space. We use image-processing algorithms to develop a set of candidate image features that can formulate just such a quantitative description of xenographed colorectal chronic tumor hypoxia. Two features in particular give low-variance measures of chronic hypoxia near a vessel: intensity sampling that extends radially away from approximated blood vessel centroids, and multithresholding to segment tumor tissue into normal, hypoxic, and necrotic regions. From these features we derive a spatiotemporal logical expression whose truth value depends on its predicate clauses that are grounded in this histological evidence. As an alternative to the spatiotemporal logical formulation, we also propose a way to formulate a linear regression function that uses all of the image features to learn what chronic hypoxia looks like, and then gives a quantitative similarity score once it is trained on a set of histology images. No MeSH data available. Related in: MedlinePlus |
Related In:
Results -
Collection
License getmorefigures.php?uid=PMC4836667&req=5
Mentions: Next we turn to our image feature for a two-segment gradient. Suppose we have a primitive state variable function ∇+: (x1, x2, p)→(δx1, δx2, δρp) that given a coordinate (x1, x2) and a particle type p returns two items: the coordinate (δx1, δx2) adjacent to (x1, x2) that has the greatest concentration of p, and δρp, the measure of that greatest concentration of p. So ∇+ is a function that performs gradient ascent. For completeness, suppose too that we have the analogous function ∇− to perform gradient descent. Thus by starting at some vessel centroid in viable region V, and extending along a contour iteratively specified by ∇+, we will eventually encounter the boundary of the hypoxic region H, followed by the boundary of the necrotic region N, eventually ending at , all the while ascending the gradient of p (in this case, anti-pimonidazole) concentration (see trajectory (B) in the left panel of Fig 8). This gives our next propositional term:(T(x1,x2)=V)U(x1,x2)←∇+(x1,x2,p),(max1(B)-min1(B),max2(B)-min2(B))(T(x1,x2)=H)U(x1,x2)←∇+(x1,x2,p),(max1(B)-min1(B),max2(B)-min2(B))(T(x1,x2)=N)∧(F(x1,x2)←∇+(x1,x2,p),172+83(∇+(x1,x2,p)=-0.21±0.19))U(x1,x2)←∇+(x1,x2,p),(172+83+267+126)(F(x1,x2)←∇+(x1,x2,p),267+126(∇+(x1,x2,p)=-0.06±0.03)),(3)which in English means “Along the total arc length of trajectory (B), we are in viable tissue until we are in hypoxic tissue until we are in necrotic tissue. And for the arc length of trajectory (B), we verify the gradient trajectory characterized by our experimental results in the following manner: for segment one, we follow a mean gradient slope of -0.21 (bounded from above and below by standard deviation 0.19) for an arc length bounded from above by the mean length of segment one, 172, plus its standard deviation, 83, until we reach segment two; then for segment two, we follow a mean gradient slope of -0.06 (bounded from above and below by standard deviation 0.03) for an arc length bounded from above by the mean length of segment two, 267, plus its standard deviation, 126.” The gradient segment lengths, length bounds, slopes, and slope bounds are given in Table 6, and all measurements are in pixels. This is merely one provisional gradient term, and not intended to represent a comprehensive gradient characterization in spatiotemporal logical terms. Note too that we could write an analogous propositional term by reordering the clauses to reflect the opposite order of encounter, and by using negated slopes and the gradient descent function ∇− (see trajectory (A) in the left panel of Fig 8). |
View Article: PubMed Central - PubMed
Affiliation: Department of Pharmacology and Systems Therapeutics, Icahn School of Medicine at Mount Sinai, New York, NY, United States of America.
No MeSH data available.