Prostate cancer diagnosis based on Gabor filter texture segmentation of ultrasound image

Abstract

It is generally agreed that the conventional Trans-Rectal Ultrasound (TRUS) examination is an important, costeffective and useful technique for imaging the prostate. TRUS is used in the interpretation of the PSA assay, for monitoring response to nonsurgical and surgical therapy, andfor providing image guidance during some minimally invasive procedures. In This paper, multi-channel filtering is proposed as an excellent method for prostate texture investigation. By processing the TRUS images using multiple resolution techniques, the image is decomposed into appropriate texture features that can be used to classz/$ the textures accordingly. Using Human Visual system (HVS), Medical Doctors use three features for texture analysis, mainly repetition, directionality and complexiq. A bank ofGaborfi1ters that is well distributed to cover the entire frequency plane is designed to mimic the HVS and therefore it is an excellent tool that can be usedfor prostate texture segmentation.

The validation of a prostate segmentation algorithm for 2D ultrasound images

“…A pilot study was performed on a set of prostate ultrasound images (15 images) [1]. We compared the performance of our automatic segmentation algorithm with an expert's radiologist.…”

Computer-Aided Diagnosis of Prostate Cancer with Emphasis on Ultrasound-Based Approaches: A Review

“…Previous literature proposed a PCa diagnosis method based on Gabor filter texture segmentation (Mohamed et al 2003). Vorhoeven and Thijssen (1993) studied the potential application of fractal dimensions for lesion detection in echographic images and, later, fractal feature vectors based on M-band wavelet transform were used for ultrasonic liver tissue characterization (Lee et al 2003).…”

A Medical Texture Local Binary Pattern For TRUS Prostate Segmentation

“…discriminate, and segment the prostate region based on its texture. There have been some attempts by employing the Gabor filter bank for prostate texture segmentation [10], [5], however prostate texture analysis has not yet been investigated seriously. We introduce a novel method by specializing Local Binary Pattern (LBP) for texture analysis of medical images.…”