Automatic Segmentation and Measurement of Infantile Hemangioma

Serban Oprisescu,Mihai Ciuc,Alina Sultana

doi:10.3390/sym13010138

Abstract

Infantile hemangiomas (IHs) are a type of vascular tumors that affect around 10% of newborns. The measurement of the lesion size and the assessment of the evolution is done manually by the physician. This paper presents an algorithm for the automatic computation of the IH lesion surface. The image scale is computed by using the Hough transform and the total variation. As pre-processing, a geometric correction step is included, which ensures that the lesions are viewed as perpendicular to the camera. The image segmentation is based on K-means clustering applied on a five-plane image; the five planes being selected from seven planes with the use of the Karhunen-Loeve transform. Two of the seven planes are 2D total variation filters, based on symmetrical kernels, designed to highlight the IH specific texture. The segmentation performance was assessed on 30 images, and a mean border error of 9.31% was obtained.

Highlights

Infantile hemangiomas (IHs) represent a type of vascular tumor that has an incidence of 3% to 10% in newborns [1,2]
The automatic image scale computation described above was tested for all the images in the database, and worked without errors
We inspected the geometrically-transformed digit to its ideal frontal template, and there was a precise match, which proves that the geometric correction worked for all the images

Summary

Introduction

Infantile hemangiomas (IHs) represent a type of vascular tumor that has an incidence of 3% to 10% in newborns [1,2] They appear and start to develop rapidly around the first two to five months of life and have a natural slow involution in the following years [3]. Their shape and size are very different from subject to subject; some hemangiomas may become function-threatening, and that is why a precise method of measurement and prediction of evolution is needed. Medical images contain some uncertainty concerning the border of some lesions or organs, uncertainty produced by the image acquisition or reconstruction process, and this uncertainty makes the segmentation task more difficult [5]

Methods

Results

Conclusion