A framework for quantification of regional cardiac fibrosis from serial sections using 3D whole slide imaging

Pathological cardiac fibrosis is important in predisposing the heart to arrhythmia and mechanical failure. The regional distribution of fibrosis is often described qualitatively and quantitatively in histological studies of animal hearts after staining collagen with specific colored stains. Currently this description is often piecemeal, as it lacks rigorous spatial registration, matching and methodological standardization between animals and between study groups. We propose a strategy for the quantification of regional fibrosis using the American Heart Association (AHA) cardiac segmentation model. We quantify fibrosis after whole heart 3D histological reconstruction in one normal rat heart and in one rat heart in right heart failure induced by monocrotaline. We then assess the minimum spaced histological sampling which allows for accurate assessment of regional fibrosis. We show that using every section of a set of 5 μm serial sections quantifies regional right ventricular fibrosis, with highly significant (p <; 0.001) differences between heart failure and control hearts. We show that the absolute error of collagen quantification is low when sections are taken spaced by up to 100 μm (error 5.7±5.8%). Likewise, absolute error associated with sectioning starting position is low for sections spaced up to 100 μm (error 13.3±17.2%). Above 100 μm section spacing quantification error is large (tending to 50%) and error associated with sectioning starting position is large (tending to 60%).