Development and validation of a Python script for automatic measurement of root morphological parameters using flatbed scanner images in comparison with WinRHIZO

Abstract

Flatbed scanner-based root morphology analysis is important in crop root research, but the widely used commercial software WinRHIZO is expensive, hindering its adoption. We developed a Python script to automatically measure five root morphological parameters—total root length, root tip count, root surface area, mean root diameter, and root volume—from flatbed scanner images, and validated it against WinRHIZO (version 2009b). The script automatically determines staining status from mean image brightness and switches processing parameters accordingly. Calibration coefficients were determined from eight images each of stained rice and unstained lettuce, then agreement with WinRHIZO was evaluated using 99 validation images (50 unstained, 49 stained). Total root length showed a Pearson correlation coefficient of 0.9998; all 99 images fell within ±10% relative error, with no significant difference from WinRHIZO (paired t-test, P>0.05). Correlation coefficients for root surface area, mean root diameter, and root volume were 0.9944, 0.9587, and 0.9816, respectively, with no significant differences (P>0.05). Root tip count showed high correlation (r=0.9930) but script values averaged 4.4% higher than WinRHIZO (P>0.01). In other words, the script matched WinRHIZO for all parameters except root tip count. The script is a free, cross-platform tool that automatically processed 99 images in approximately 25 minutes.