Histomorphometry

1. Histomorphometry Or How to get numbers out of slides Stephen Greenwald Pathology Group, Institute of Cell & Molecular Science Barts & The London School of Medicine & Dentistry

2. Outline • What is morphometry? • Why histomorphometry? • Measurement methods • “Manual” • Computerised • Standard processes in computerised histomorph. • Image capture • Enhancement • Thresholding • Measuring • Micro CT of stented arteries

3. What is morphometry? A body of methods for obtaining numerical information about the shape and size of a structure in terms of quantities such as: • volume • surface area • relative amounts of each component • orientation, interconnections • distribution of substructures • etc.

4. Why histomorphometry? • When applied to biological tissue examined microscopically it useful in correlating structure and function e.g. • Alveolar or gut villus surface area • Arterial composition and elasticity • Quantification of • Hyperplasia, dysplasia, hypertrophy • Immunohistochemical or flourescent markers • Area or intensity

5. Major challenge • To extract information about large 3-D structures from microscopic measurements on thin 2-D sections • To do this histomorphometry uses the: Delesse Principle

6. Delesse Principle “In a rock composed of a number of minerals, the area occupied by any given mineral is proportional to the volume of the mineral in the rock” • Repeated determinations of the area fraction will yield an estimate of the volume fraction. • The more determinations; the better the estimate. Delesse A. (1847) Procede mechanique pour determines la composition des roches. Comptes Rendus de l’Academie des Science (Paris) 25, 544)

7. How to estimate area fraction • paper cutting and weighing • planimetry

8. Planimeter

9. How to estimate area fraction • paper cutting and weighing • planimetry • dot counting

10. Dot counting Nuclear area/cell area = number of dots in nuclei/number of dots in cell Absolute area of a structure = number of dots in structure x area of dot square

11. How to estimate area fraction • paper cutting and weighing • planimetry • dot counting • square counting

12. Square counting 34 squares 7 squares

13. How to estimate area fraction • paper cutting and weighing • planimetry • dot counting • square counting • pixel counting in a digital image • semi- or fully automatic system

14. Stimulants Computerised histomorphometry TV camera ADC LUT Microscope ADC Image processor Video memory LUT User. mouse, light-pen Main processor Storage

15. Pixel counting

16. Pixel counting

17. Recognising objects by colour

18. More difficult measurements • length • seminiferous tubules • surface area • alveoli, gut villi etc. • counting discrete objects • cells, nuclei, alveoli, elastic lamellae etc. • Size distribution • cells, nuclei, tumours etc.

19. How to estimate length(Buffon’s needle problem) • If you drop a nail/needle on the floor, what is the probability it will come to rest over a crack between the floor boards? Louis le Clerc, Compte de Buffon, (1707 -1788) French naturalist & polymath

20.  lproj Buffon needle problem  The probability (p) of the needle (or nail) landing on a join depends on the length of the needle (l), the width of the boards (d) and the angle it makes with the direction of the boards (). The angle determines the projected (i.e. effective) length of the nail, (lproj) d l

21. Inverse problem;i.e. throw the grid at the nails • Imagine a contour of length L composed of small elements, l • Now throw the grid (spacing, d) at the nails (i.e. the small elements)

22. Villus and crypt length measurements

23. How to measure surface area • Measure absolute volume (V) of entire organ • Archimedes, weight (knowing density) • Estimate tissue volume fraction from area fraction • calculate tissue volume • Count intercepts (Nint) using grid of total length (L)

24. Normal nucleus: Area = 10m2, perimeter =14m Abnormal nucleus: Area = 10m2, perimeter =26m Pattern recognition • normal v abnormal morphology • displasia, metaplasia • counting poorly stained structures • nuclei, nuclear organelles, leucocytes

25. Standard processes 1.5MB • Image capture • Enhancement • contrast/colour • background correction • Thresholding (identifying structures of interest) • colour • intensity • shape • Measurement • area, perimeter, counting 10kB 1kB

26. Image enhancement:shade correction Uneven background illumination

27. Image enhancement:shade correction Original image with uneven illumination

28. Image enhancement:shade correction Shade corrected image

30. Enhanced Contrast enhancement: Original

31. Enhanced Thresholded Thresholding by colour

32. Measurement

33. The Effect of Stent Oversize Stiffness & Structure on restenosis • In vivo radiographic measurement of stent dimensions in pig carotid and iliac arteries • Development of a micro CT method for stented vessel morphometry on excised arteries

34. Study aims • To quantify degree of restenosis • Effect of stent oversize and stiffness • To compare two stent types • SMART stent (a standard design) • Major problem is restenosis • Compliant ended stent (a novel design). Developed by collaborators, J.E. Moore & Colleagues at Texas A & M

35. Hypothesis By matching the compliance of the stent to that of the “native” artery, flow disturbances and bending stress at the stent/artery junction is reduced and hence restenosis is minimised

36. Stents used in the Study SMART stent Compliant Ended Stent

37. Compliance Matching Stent • Rigid in the centre to provide recoil resistance • Parabolic and cantilevered struts • gradual change in compliance • reduces stress concentration and bending • Less disturbed flow

38. Methods • 65 stents implanted in the iliac and carotid arteries of 17 Large White pigs • Lumen diameter determined before and after implantation by angiography • Follow-up angiography on days 3,7 and 28 • At day 28 the arteries were pressure perfused and removed for histology and CT scanning

39. Lumen diameter [mm] Vessel dimensions determined by automatic edge detecting algorithm

40. Micro CT of excised vessels • Vessels pressure fixed in situ (10% formol saline) • Excised and immersed in oil based contrast medium • Custom built Micro CT scanner (Dental Biophysics QMUL) • Voxel size (30 x 30 x 30µm) • Images processed on custom software developed under KS400 image analysis system

41. A trip through a stented artery

42. One of about 1200 slices cut perpendicular to the long axis of the vessel

43. Image processing Original slice Thresholded Media/Adventita only Circle fitted Stent struts

44. 1.00 18 0.95 16 0.90 14 12 0.85 10 8 0.80 6 0 10 20 30 40 4 0 10 20 30 40 Lumen Stent Slice measurements (CE Stent) Lumen circularity Lumen and stent area [mm2] Distance [mm] Distance [mm]

45. 3D reconstruction

46. And rendering

48. Conclusions • Histomorphometry is useful for counting and measuring clearly defined structures • Limited by a lack of “intelligent” software • Extremely difficult to better the human eye-brain combination for pattern recognition/diagnosis • For histopathologists, may be valuable for quantifying prognosis • Measuring ratio or distribution of different tumour markers • No immediate cause for alarm amongst histopathologists…but watch this space.