Automating Porosity Features Extraction from Second Harmonic Generation Images of Cervical Tissue

Siamak Yousefi, Boram Kim, and Nasser Kehtarnavaz


Imaging of pregnancy stages, second harmonic generation imaging, image processing, automatic feature extraction


Image analysis of collagen SHG (Second Harmonic Generation) signal has potential in preterm birth detection and staging pregnancy. Current interactive methods for extracting collagen features, such as porosity, are cumbersome, subjective, time consuming and prone to error. An automated image processing pipeline is presented in this paper to automate the whole process of porosity features extraction. The proposed automated pipeline includes the following image processing components: nonlinear intensity illumination correction by arithmetic filtering, noise reduction by adaptive Wiener filtering, thresholding by Otsu to obtain a binary image representing pore areas, and finally particle analysis to obtain the porosity features including number of pores, pore size, and pore density. The effectiveness of the developed automated pipeline is examined by comparing the classification outcomes of the interactive manual and automated approaches.

Important Links:

Go Back