|
Most medical imaging is inherently three-dimensional (3D) but for validation of pathological
findings, histopathology is commonly used and typically histopathology images are acquired as twodimensional
slices with quantitative analysis performed in a single dimension. Histopathology is
invasive, labour-intensive, and the analysis cannot be performed in real time, yet it remains the gold
standard for the pathological diagnosis and validation of clinical or radiological diagnoses of disease.
A major goal worldwide is to improve medical imaging resolution, sensitivity and specificity to
better guide therapy and biopsy and to one day delay or replace biopsy. A key limitation however is
the lack of tools to directly compare 3D macroscopic imaging acquired in patients with
histopathology findings, typically provided in a single dimension (1D) or in two dimensions (2D).
To directly address this, we developed methods for 2D histology slice visualization/registration to
generate 3D volumes and quantified tissue components in the 3D volume for direct comparison to
volumetric micro-CT and clinical CT. We used the elastase-instilled mouse emphysema lung model
to evaluate our methods with murine lungs sectioned (5 μm thickness/10 μm gap) and digitized with
2μm in-plane resolution. 3D volumes were generated for wildtype and elastase mouse lung sections
after semi-automated registration of all tissue slices. The 1D mean linear intercept (Lm) for wildtype
(WT) (47.1 μm ± 9.8 μm) and elastase mouse lung (64.5 μm ± 14.0 μm) was significantly different
(p<.001). We also generated 3D measurements based on tissue and airspace morphometry from the
3D volumes and all of these were significantly different (p<.0001) when comparing elastase and WT
mouse lung. The ratio of the airspace-to-lung volume for the entire lung volume was also
significantly and strongly correlated with Lm.
|
